Split (1)

train · 521k rows

text stringlengths 254 1.16M
--- title: Identification of Tregs-Related Genes with Molecular Patterns in Patients with Systemic Sclerosis Related to ILD authors: - Jiao Luo - Dongdong Li - Lili Jiang - Chunhua Shi - Lihua Duan journal: Biomolecules year: 2023 pmcid: PMC10046355 doi: 10.3390/biom13030535 license: CC BY 4.0 --- # Identification of Tregs-Related Genes with Molecular Patterns in Patients with Systemic Sclerosis Related to ILD ## Abstract Background: Systemic Sclerosis (SSc) is an autoimmune disease that is characterized by vasculopathy, digital ulcers, Raynaud’s phenomenon, renal failure, pulmonary arterial hypertension, and fibrosis. Regulatory T (Treg) cell subsets have recently been found to play crucial roles in SSc with interstitial lung disease (ILD) pathogenesis. This study investigates the molecular mechanism of Treg-related genes in SSc patients through bioinformatic analyses. Methods: The GSE181228 dataset of SSc was used in this study. CIBERSORT was used for assessing the category and proportions of immune cells in SSc. Random forest and least absolute shrinkage and selection operator (LASSO) regression analysis were used to select the hub Treg-related genes. Results: Through bioinformatic analyses, LIPN and CLEC4D were selected as hub Treg-regulated genes. The diagnostic power of the two genes separately for SSc was 0.824 and 0.826. LIPN was associated with the pathway of aminoacyl−tRNA biosynthesis, Primary immunodeficiency, DNA replication, etc. The expression of CLEC4D was associated with the pathway of Neutrophil extracellular trap formation, PPAR signaling pathway, Staphylococcus aureus infection, Systemic lupus erythematosus, TNF signaling pathway, and Toll−like receptor signaling pathway. Conclusion: Through bioinformatic analyses, we identified two Treg-related hub genes (LIPN, CLEC4D) that are mainly involved in the immune response and metabolism of Tregs in SSc with ILD. Moreover, our findings may provide the potential for studying the molecular mechanism of SSc with ILD. ## 1. Introduction Systemic Sclerosis (SSc) as an autoimmune disease has severe clinical manifestations and a high mortality rate, and treatment for this disease is minimal and ineffective [1]. SSc is characterized by vasculopathy, digital ulcers, Raynaud’s phenomenon, renal failure, pulmonary arterial hypertension, and fibrosis [2,3]. Fibrosis, as a hallmark of SSc, frequently involves the lung, manifesting as interstitial lung disease (ILD) [4,5]. It is the most serious complication associated with SSc and is the top cause of death associated with SSc [6]. Clinical and pathologic manifestations of SSc are due to abnormalities in the innate and adaptive immune systems, which result in the production of autoantibodies and cell-mediated autoimmunity. Then, the accumulation of collagen and other matrix components in the skin and internal organs occurs due to microvascular epitheliopathy and fibroblast dysfunction [7]. In SSc with ILD, fibrosis may result from an interplay between autoimmunity, inflammation, and epithelial and vascular injury [8]. However, the pathogenesis of fibrosis often lacks insight into the interactions between key players. Regulatory T (Treg) cell subsets have recently been found to play crucial roles in SSc pathogenesis [9,10,11]. The importance of Tregs for maintaining immune homeostasis and self-tolerance is increasingly recognized. Most studies reported that SSc patients had reduced frequency and/or impairment of circulating Tregs [12]. Fenoglio et al. found an imbalance between circulating Th17 cells and Treg cells (Tregs) in patients with SSc, with an increased proportion of Th17 cells and a decrease in both CD4+CD25+CD127− and CD8+CD28−Treg cells [13]. Also, patients with high computed tomography scores for ILD had elevated Treg cells [14,15]. Nevertheless, studies of Tregs’ phenotype and function are rare. Little is known about the regulatory mechanisms of Tregs alteration in SSc with ILD. Identifying biomarkers related to Tregs will facilitate the exploration of immune infiltration mechanisms of SSc with ILD. Bioinformatics-based studies of the contribution of genes related to Tregs of SSc with ILD have not been conducted yet. To explore the effect of Treg cells and identify potential biomarkers of SSc with ILD, WGCNA was performed using gene expression data in the peripheral blood of SSc with ILD. The T-cell compositions of samples were calculated using the CIBERSORT algorithm. We then identified Treg-related genes from important modules and genes related to Tregs infiltration levels, and machine learning was used to identify hub Treg-related genes. Identifying Treg-related genes of SSc with ILD may provide potential pathogenesis and therapeutic targets of SSc with ILD. ## 2.1. Expression Data Download and Processing The dataset GSE181228 was downloaded from the GEO (https://www.ncbi.nlm.nih.gov/geo/ accessed on 5 October 2022) database. The GSE181228 dataset was last updated on 1 February 2022 and annotated on platform GPL24676. Samples were obtained from peripheral blood and contained 134 SSc with ILD patients (untreated) and 45 healthy controls. Expression data of GSE181228 were DESeq2 normalized and log2 transformed by the uploader. Genes that corresponded to multiple probes were averaged after annotation. Genes with the expression of 0 in more than 30 samples were excluded. Since the data of this study were obtained from public databases, ethics committee approval was not required. ## 2.2. Differentially Expressed Genes (DEGs) DEGs between the SSc with ILD patients and healthy groups were analyzed using the R package “limma” [16]. Genes with adj. $p \leq 0.05$ and an absolute value of log2 (fold change(FC)) > 1 were identified as DEGs. The R package “heatmap” and “ggplot2” were used to map volcanoes and heatmaps. ## 2.3. Immune Infiltration Analysis for the Datasets and Weighted Gene Co-Expression Network Analysis (WGCNA) In this study, the relative expression of 22 immune cells in each sample was determined using the “CIBERSORT” (R package) [17]. Cells with an expression of 0, which were more than $50\%$ in the sample, were excluded. Heat maps and box plots were plotted using the R package “heatmap” and “ggplot2”. Weighted correlation network analysis was performed using the R package “WGCNA” [18]. The samples were clustered, and outlier samples were excluded. To exclude highly correlated genes that did not vary significantly, the MAD method was used to select the 5000 genes with the largest absolute median difference. The correlation matrix was constructed, and a weighted adjacency matrix was generated. Suitable β values were selected to obtain the topological overlap matrix (TOM). Modules of a minimum of 30 genes were constructed using average linkage hierarchical clustering and module dendrograms. To measure the correlation between genes and immune cells, gene significance (GS) was calculated to determine the significance of each module. A threshold of more than 0.25 was used to merge similar modules [19]. The most correlated modules with Tregs were identified, and the intersection between Tregs’ most correlated module genes and DEGs was identified as Tregs-related DEGs (TDEGs). ## 2.4. Identification of Hub Genes Based on differentially expressed feature genes, least absolute shrinkage and selection operator (LASSO) regression analysis [20] was performed using the R package “glmnet” [21], and the variables corresponding to the value of the penalty parameter lambda.1se were selected as marker genes using 10-fold cross-validation. The R package “randomForest” was used to screen important key genes in a random forest (RF) classifier with several binary trees initially ranging from 1 to 100 cycles, respectively. Binary trees were selected based on the lowest value of the error rate, and decision trees were selected based on model stability, thus constructing the random forest model. The random forest uses 10-fold cross-validation repeated 5 times to select the optimal number of marker genes. Intersect genes of two algorithms were selected as hub genes. Hub genes were used to construct logistic regression models using the R package “glmnet”. ROC curve analysis was performed using the R package pROC to calculate the area under the curve (AUC) and assess the diagnostic ability of the model [22]. ## 2.5. Gene Set Enrichment Analysis (GSEA) of Hub Genes Samples were divided into two groups based on the median expression of the signature genes, and GSEA analysis [23] was performed using the “gseKEGG” in the R package “clusterProfiler”. A p-value < 0.05 was indicative of statistical significance. The number of permutations was set to 1000, and the permutation type was set as “gene list”. The most significantly enriched pathway was selected based on the enrichment score. ## 2.6. Statistical Analysis R software (version 4.1.3) was used to perform statistical analysis. Student’s t-test (two-tailed) was used to determine the statistical significance for both groups. A logistic regression algorithm was used to build a prediction model. Roc curve analysis and the area under the curve were calculated. A $p \leq 0.05$ was considered a statistically significant difference. ## 3.1. Identification of Differentially Expressed Genes An expression matrix containing 179 samples was obtained from the GSE150910. Differential expression analysis was performed between 134 SSc with ILD patient groups and 45 healthy groups, and a total of 125 differentially expressed genes (DEGs) were obtained, 83 up-regulated genes and 42 down-regulated genes. One hundred twenty-five DEGs were visualized by volcano map (Figure 1A). The differentially expressed genes with log2 ratio and adjusted p-value are shown in Table S1, and the most significantly differentially up- and down-regulated 25 DEGs each were visualized by heatmap (Figure 1B). ## 3.2. Immune Cell Landscape A total of 12 cell types were obtained, and the expression of 12 immune cell types between the SSc with ILD patient group and the healthy group was visualized by heatmap (Figure 2A). Compared to the healthy group, the expression of peripheral cells (B cells memory, CD8+ T cells, T cells regulatory (Tregs), NK cells resting, NK cells activated) was significantly higher in the SSc with ILD patient group. On the contrary, the expression of infiltration cells (Monocytes, Mast cells resting, Neutrophils) of SSc with ILD patients was significantly decreased compared to the healthy group (Figure 2B). ## 3.3. The WGCNA Co-Expression Network and Identify Differential Tregs-Related DEGs (TDEGs) Using sample clustering, three outlier samples were first excluded from this study. A scale-free network was constructed by selecting β = 6 (no scale R2 = 0.943) as the soft threshold. A total of 5000 genes were grouped into 12 modules (Figure 3A). The black module mostly correlated with the Tregs (Figure 3B). The significance of genes in the black module for Tregs is shown in Figure 3C (0.74, $p \leq 0.001$). TDEGs were obtained by taking the intersection of Tregs-related genes and DEGs (Figure 3D). ## 3.4. Identify Hub Genes Eight genes were identified by LASSO regression analysis (the optimal lambda.1se was 0.056) in 16 TDEGs (Figure 4A,B). Sixteen TDEGs were most stable with a binomial tree of 7 and a decision tree of 400 using the random forest algorithm (Figure 4C). The top 10 significant genes were ranked according to accuracy and Gini coefficient (Figure 4D). A 10-fold cross-validation using random forest was repeated 5 times, and 3 genes were finally selected. The intersection of the two algorithms was taken to obtain a total of three significant hub genes (LIPN, CLEC4D, FAAHP1) (Figure 5A). According to GeneCards (https://www.genecards.org/ accessed on 5 October 2022) database, FAAHP1 is a pseudogene, so this gene was eliminated in this study. A prediction model was developed using a logistic regression algorithm using two genes (LIPN, CLEC4D) with the equation (y = −15.598 + 1.512exp + 1.065exp). The diagnostic power of the two genes separately for SSc with ILD was displayed in Figure 5B (LIPN, CLEC4D; 0.824, 0.826). The diagnostic power of the model for SSc with ILD was 0.877 (Figure 5B). ## 3.5. Signaling Pathways Associated with Hub Genes by GSEA In this study, the functions of hub genes were explored using GSEA. Ten significantly up- and down-regulated enrichment KEGG pathways of hub genes were selected based on enrichment scores. The KEGG pathways of hub genes are shown in Tables S2 and S3. The expression of LIPN was associated with the up-regulated pathway of amoebiasis, Autophagy-animal, Legionellosis, Longevity regulating pathway, Longevity regulating pathway-multiple species, Malaria, Pantothenate and CoA biosynthesis, Renal cell carcinoma, Rheumatoid arthritis, and Ribosome. In contrast, the expression of LIPN was down-regulated in pathways of alanine, aspartate, and glutamate metabolism, aminoacyl-tRNA biosynthesis, arginine and proline metabolism, biosynthesis of amino acids, DNA replication, ECM-receptor interaction, glycine, serine and threonine metabolism, Hypertrophic cardiomyopathy, Primary immunodeficiency, and Tryptophan metabolism (Figure 6C). The expression of CLEC4D was associated with the up-regulated pathway of African trypanosomiasis, Legionellosis, Leishmaniasis, Malaria, Neutrophil extracellular trap formation, PPAR signaling pathway, Staphylococcus aureus infection, Systemic lupus erythematosus, TNF signaling pathway, and Toll-like receptor signaling pathway. The expression of CLEC4D was associated with the down-regulated pathway of alanine, aspartate, and glutamate metabolism, Arrhythmogenic right ventricular cardiomyopathy, Basal cell carcinoma, DNA replication, glycine, serine and threonine metabolism, Hypertrophic cardiomyopathy, Mannose type O-glycan biosynthesis, Proximal tubule bicarbonate reclamation, Proximal tubule bicarbonate reclamation, and Ribosome biogenesis in eukaryotes (Figure 6D). ## 4. Discussion Systemic sclerosis (SSc) is an autoimmune disease with high mortality. Pulmonary fibrosis is the most common and serious clinical manifestation of the disease and is also the main cause of death. The pathophysiology of SSc with ILD is complex and unclear. Due to immune dysfunctions, SSc is characterized by autoimmunity, vasculopathy, and fibrosis. The complex interaction and activation of immune cells are key factors in the formation of fibrosis. Tregs, as a type of CD4 + T cells, have been demonstrated to play an important role in SSc with different mechanisms. In the mouse model of bleomycin-induced pulmonary fibrosis, CD4 + CD25highFoxP3+ cells in the lung were increased after IL-2 complex treatment associated with aggravated lung fibrosis [24]. The role of Tregs in the SSc-ILD is more and more recognized. However, the regulatory mechanisms of T cells in SSc remain poorly understood. In this study, we used bioinformatics methods to explore the role of Tregs in the SSc with ILD and identify the hub gene of Tregs in SSc with ILD and effective diagnostic biomarkers for SSc with ILD. DEGs were obtained from peripheral blood and contained 134 SSc with ILD patients (untreated) and 45 healthy controls. Compared to the healthy group, the expression of peripheral cells (B cells memory, CD8+ T cells, T cells regulatory (Tregs), NK cells resting, NK cells activated) was significantly higher in the SSc with ILD patient group. On the contrary, the expression of infiltration cells (Monocytes, Mast cells resting, Neutrophils) of SSc with ILD patients was significantly decreased compared to the healthy group. Some reports showed that Treg cells without immunosuppressive functions have increased in number, while those with immunosuppressive functions have decreased among Treg cells of SSc patients [10]. SSc patients with active disease exhibit upregulation of FOXP3 gene expression in Treg cells [25]. In this study, Tregs also were significantly higher in the SSc with ILD patients. WGCNA was used to extract Tregs-related genes. Then, TDEGs were obtained by taking the intersection of Tregs-related genes and DEGs. A commonly used algorithm, LASSO analysis which is a machine learning–based algorithm, has been demonstrated to yield clinical efficacy [26,27]. For RF analysis, there is no restriction on variable conditions, which makes it an appropriate ensemble learning algorithm and machine learning method [28]. The RF method can be used to predict continuous variables with no obvious deviations from the prediction [29]. We used both classical algorithms to select hub genes of Tregs in SSc with ILD. Finally, LIPN and CLEC4D were obtained from the intersection of the LASSO regression analysis and random forest based on TDEGs. The LIPN and CLDE4D were demonstrated to be hub genes of Tregs for SSc with ILD. The KEGG pathway of LIPN and CLEC4D was analyzed by GSEA. The expression of LIPN was associated with an up-regulated pathway of amoebiasis, Autophagy-animal, Rheumatoid arthritis, and so on. The expression of LIPN was associated with a down-regulated pathway of aminoacyl-tRNA biosynthesis, Primary immunodeficiency, DNA replication, and so on. The expression of CLEC4D was associated with an up-regulated pathway of Neutrophil extracellular trap formation, PPAR signaling pathway, Staphylococcus aureus infection, Systemic lupus erythematosus, TNF signaling pathway, and Toll−like receptor signaling pathway. The expression of CLEC4D was associated with a down-regulated pathway of alanine, aspartate, and glutamate metabolism, Arrhythmogenic right ventricular cardiomyopathy, Basal cell carcinoma, DNA replication, and so on. Known as LIPN (Rv2970c), it belongs to the Lip family of M. tuberculosis H37Rv and is homologous to the human hormone-sensitive lipase. In addition to its preference for short carbon chain substrates [30], Shirli et al. reported that the LIPN gene encoding epidermal lipase N has been linked to congenital ichthyosis with a late-onset form associated with autosomal-recessive inheritance [31]. In another paper, gestational diabetes mellitus was associated with lipolysis-related genes such as LIPN [32]. However, there are few studies on LIPN in autoimmune diseases, especially SSc with ILD. In this study, LIPN was demonstrated to be correlated with amino acid metabolism. LIPN may be a hub gene to affect the metabolism of Tregs. CLDE4D, as a C-type (Ca2+-dependent) lectin (CLEC) receptor (CLEC), has potential regulatory effects on immune cell trafficking, which is essential in innate pattern recognition [33]. CLEC4D, as a key component of anti-mycobacterial immunity, was expressed by myeloid cells [34]. It has been demonstrated that the deficiency of CLEC4D in the gut promotes the development of colitis by impairing antifungal immune responses [35]. The function of CLEC4D in SSc with ILD, especially in Tregs, was not studied. In this research, CLEC4D was a Tregs related-gene and associated with the immune-related pathway. It means that CLEC4D was associated with the immune function of Tregs. Meanwhile, there was good diagnostic power of LIPN and CLEC4D for SSc with ILD. CLEC4D and LIPN may play a key role in SSc with ILD by affecting the function of Tregs. There are still limitations in this study that we cannot ignore. Generally, the ratio of women to men in SSc ranges from 3:1 to 7:1 [36]. In this study, the female SSc patients were up to $73.1\%$. As we could not get clinical information on the samples from the public database, gender might be a confounding variable in the analyses to impact the results of the differential analysis. In addition, this study was performed only between SSc with ILD and healthy controls. Patients with SSc were not included in the study, so the results could not be verified in SSc and SSc with ILD. The study should have included more forms of SSc patients. Nevertheless, these two genes have not been validated in other gene sets. LIPN and CLEC4D need further examination in in vivo or in vitro experiments; however, this gives us a direction. The mechanism of LIPN and CLEC4D for SSc needs to be further explored. In conclusion, this study first analyzed and assessed molecular patterns of Tregs-related genes of SSc with ILD using bioinformatics methods. Our findings explored two Tregs-related genes of SSc (LIPN and CLEC4D). The Treg-related hub genes were mainly involved in amino acid metabolism and inflammatory pathways of Tregs in SSc with ILD. In addition, the analysis of the sensitivity and specificity of two Tregs-related hub genes unveiled that they may be potential biomarkers for SSc with ILD. ## References 1. Leroy E.C., Black C., Fleischmajer R., Jablonska S., Krieg T., Medsger T.A., Rowell N., Wollheim F.. Scleroderma (systemic sclerosis): Classification, subsets and pathogenesis. J. Rheumatol. (1988) 15 202-205. PMID: 3361530 2. Varga J., Abraham D.. Systemic sclerosis: A prototypic multisystem fibrotic disorder. J. Clin. Investig. (2007) 117 557-567. DOI: 10.1172/JCI31139 3. Moinzadeh P., Riemekasten G., Siegert E., Fierlbeck G., Henes J., Blank N., Melchers I., Mueller-Ladner U., Frerix M., Kreuter A.. Vasoactive Therapy in Systemic Sclerosis: Real-life Therapeutic Practice in More Than 3000 Patients. J. Rheumatol. (2015) 43 66-74. DOI: 10.3899/jrheum.150382 4. Steele R., Hudson M., Lo E., Baron M.. Clinical decision rule to predict the presence of interstitial lung disease in systemic sclerosis. Arthritis Care Res. (2012) 64 519-524. DOI: 10.1002/acr.21583 5. Vonk M.C., Broers B., Heijdra Y.F., Ton E., Snijder R., van Dijk A.P.J., van Laar J.M., Bootsma H., van Hal P.T.W., Hoogen F.H.J.V.D.. Systemic sclerosis and its pulmonary complications in The Netherlands: An epidemiological study. Ann. Rheum. Dis. (2008) 68 961-965. DOI: 10.1136/ard.2008.091710 6. Tyndall A.J., Bannert B., Vonk M., Airò P., Cozzi F., Carreira P.E., Bancel D.F., Allanore Y., Müller-Ladner U., Distler O.. Causes and risk factors for death in systemic sclerosis: A study from the EULAR Scleroderma Trials and Research (EUSTAR) database. Ann. Rheum. Dis. (2010) 69 1809-1815. DOI: 10.1136/ard.2009.114264 7. Epattanaik D., Ebrown M., Postlethwaite B.C., Postlethwaite A.E.. Pathogenesis of Systemic Sclerosis. Front. Immunol. (2015) 6 272. DOI: 10.3389/fimmu.2015.00272 8. Akter T., Silver R.M., Bogatkevich G.S.. Recent Advances in Understanding the Pathogenesis of Scleroderma-Interstitial Lung Disease. Curr. Rheumatol. Rep. (2014) 16 411. DOI: 10.1007/s11926-014-0411-1 9. Frantz C., Auffray C., Avouac J., Allanore Y.. Regulatory T Cells in Systemic Sclerosis. Front. Immunol. (2018) 9 2356. DOI: 10.3389/fimmu.2018.02356 10. Amelsfort J.M.M.-V., Walter G.J., Taams L.S.. CD4. Expert Rev. Clin. Immunol. (2011) 7 499-514. DOI: 10.1586/eci.11.28 11. Slobodin G., Rimar D.. Regulatory T Cells in Systemic Sclerosis: A Comprehensive Review. Clin. Rev. Allergy Immunol. (2016) 52 194-201. DOI: 10.1007/s12016-016-8563-6 12. Banica L., Besliu A., Pistol G., Stavaru C., Ionescu R., Forsea A.-M., Tanaseanu C., Dumitrache S., Otelea D., Tamsulea I.. Quantification and molecular characterization of regulatory T cells in connective tissue diseases. Autoimmunity (2009) 42 41-49. DOI: 10.1080/08916930802282651 13. Mathian A., Parizot C., Dorgham K., Trad S., Arnaud L., Larsen M., Miyara M., Hié M., Piette J.-C., Frances C.. Activated and resting regulatory T cell exhaustion concurs with high levels of interleukin-22 expression in systemic sclerosis lesions. Ann. Rheum. Dis. (2012) 71 1227-1234. DOI: 10.1136/annrheumdis-2011-200709 14. Giovannetti A., Rosato E., Renzi C., Maselli A., Gambardella L., Giammarioli A.M., Palange P., Paoletti P., Pisarri S., Salsano F.. Analyses of T cell phenotype and function reveal an altered T cell homeostasis in systemic sclerosis: Correlations with disease severity and phenotypes. Clin. Immunol. (2010) 137 122-133. DOI: 10.1016/j.clim.2010.06.004 15. Jiang N., Li M., Zeng X.. Correlation of Th17 cells and CD4. Chin. Med. J. (2014) 127 3557-3561. PMID: 25316229 16. Ritchie M.E., Phipson B., Wu D., Hu Y., Law C.W., Shi W., Smyth G.K.. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. (2015) 43 e47. DOI: 10.1093/nar/gkv007 17. Newman A.M., Liu C.L., Green M.R., Gentles A.J., Feng W., Xu Y., Hoang C.D., Diehn M., Alizadeh A.A.. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods (2015) 12 453-457. DOI: 10.1038/nmeth.3337 18. Langfelder P., Horvath S.. WGCNA: An R package for weighted correlation network analysis. BMC Bioinform. (2008) 9. DOI: 10.1186/1471-2105-9-559 19. Pan S., Zhan Y., Chen X., Wu B., Liu B.. Bladder Cancer Exhibiting High Immune Infiltration Shows the Lowest Response Rate to Immune Checkpoint Inhibitors. Front. Oncol. (2019) 9 1101. DOI: 10.3389/fonc.2019.01101 20. Van Gestel T., Suykens J., Lanckriet G., Lambrechts A., De Moor B., Vandewalle J.. Bayesian Framework for Least-Squares Support Vector Machine Classifiers, Gaussian Processes, and Kernel Fisher Discriminant Analysis. Neural Comput. (2002) 14 1115-1147. DOI: 10.1162/089976602753633411 21. Friedman J.H., Hastie T., Tibshirani R.. Regularization Paths for Generalized Linear Models via Coordinate Descent. J. Stat. Softw. (2010) 33 1-22. DOI: 10.18637/jss.v033.i01 22. Robin X., Turck N., Hainard A., Tiberti N., Lisacek F., Sanchez J.-C., Müller M.. pROC: An open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinform. (2011) 12. DOI: 10.1186/1471-2105-12-77 23. Yu G., Wang L.-G., Han Y., He Q.-Y.. clusterProfiler: An R Package for Comparing Biological Themes Among Gene Clusters. OMICS J. Integr. Biol. (2012) 16 284-287. DOI: 10.1089/omi.2011.0118 24. Birjandi S.Z., Palchevskiy V., Xue Y.Y., Nunez S., Kern R., Weigt S.S., Lynch J.P., Chatila T.A., Belperio J.A.. CD4+CD25hiFoxp3+ Cells Exacerbate Bleomycin-Induced Pulmonary Fibrosis. Am. J. Pathol. (2016) 186 2008-2020. DOI: 10.1016/j.ajpath.2016.03.020 25. Ricard L., Jachiet V., Malard F., Ye Y., Stocker N., Rivière S., Senet P., Monfort J.-B., Fain O., Mohty M.. Circulating follicular helper T cells are increased in systemic sclerosis and promote plasmablast differentiation through the IL-21 pathway which can be inhibited by ruxolitinib. Ann. Rheum. Dis. (2019) 78 539-550. DOI: 10.1136/annrheumdis-2018-214382 26. Reichling C., Taieb J., Derangere V., Klopfenstein Q., Le Malicot K., Gornet J.-M., Becheur H., Fein F., Cojocarasu O., Kaminsky M.C.. Artificial intelligence-guided tissue analysis combined with immune infiltrate assessment predicts stage III colon cancer outcomes in PETACC08 study. Gut (2019) 69 681-690. DOI: 10.1136/gutjnl-2019-319292 27. Huang Y.-Q., Liang C.-H., He L., Tian J., Liang C.-S., Chen X., Ma Z.-L., Liu Z.-Y.. Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer. J. Clin. Oncol. (2016) 34 2157-2164. DOI: 10.1200/JCO.2015.65.9128 28. Bureau A., Dupuis J., Hayward B., Falls K., Van Eerdewegh P.. Mapping complex traits using Random Forests. BMC Genet. (2003) 4 S64. DOI: 10.1186/1471-2156-4-S1-S64 29. Ellis K., Kerr J., Godbole S., Lanckriet G., Wing D., Marshall S.. A random forest classifier for the prediction of energy expenditure and type of physical activity from wrist and hip accelerometers. Physiol. Meas. (2014) 35 2191-2203. DOI: 10.1088/0967-3334/35/11/2191 30. Jadeja D., Dogra N., Arya S., Singh G., Singh G., Kaur J.. Characterization of LipN (Rv2970c) of. J. Cell. Biochem. (2015) 117 390-401. DOI: 10.1002/jcb.25285 31. Israeli S., Khamaysi Z., Fuchs-Telem D., Nousbeck J., Bergman R., Sarig O., Sprecher E.. A Mutation in LIPN, Encoding Epidermal Lipase N, Causes a Late-Onset Form of Autosomal-Recessive Congenital Ichthyosis. Am. J. Hum. Genet. (2011) 88 482-487. DOI: 10.1016/j.ajhg.2011.02.011 32. Zhang M., Li Q., Wang K.-L., Dong Y., Mu Y.-T., Cao Y.-M., Liu J., Li Z.-H., Cui H.-L., Liu H.-Y.. Lipolysis and gestational diabetes mellitus onset: A case-cohort genome-wide association study in Chinese. J. Transl. Med. (2023) 21 47. DOI: 10.1186/s12967-023-03902-4 33. Zhang Y., Wei H., Fan L., Fang M., He X., Lu B., Pang Z.. CLEC4s as Potential Therapeutic Targets in Hepatocellular Carcinoma Microenvironment. Front. Cell Dev. Biol. (2021) 9 681372. DOI: 10.3389/fcell.2021.681372 34. Graham L.M., Gupta V., Schafer G., Reid D.M., Kimberg M., Dennehy K.M., Hornsell W.G., Guler R., Campanero-Rhodes M.A., Palma A.D.S.. The C-type Lectin Receptor CLECSF8 (CLEC4D) Is Expressed by Myeloid Cells and Triggers Cellular Activation through Syk Kinase. J. Biol. Chem. (2012) 287 25964-25974. DOI: 10.1074/jbc.M112.384164 35. Wang T., Pan D., Zhou Z., You Y., Jiang C., Zhao X., Lin X.. Dectin-3 Deficiency Promotes Colitis Development due to Impaired Antifungal Innate Immune Responses in the Gut. PLOS Pathog. (2016) 12. DOI: 10.1371/journal.ppat.1005662 36. Calderon L.M., Pope J.E.. Scleroderma epidemiology update. Curr. Opin. Rheumatol. (2021) 33 122-127. DOI: 10.1097/BOR.0000000000000785
--- title: MicroRNA-483-5p Inhibits Hepatocellular Carcinoma Cell Proliferation, Cell Steatosis, and Fibrosis by Targeting PPARα and TIMP2 authors: - Suryakant Niture - Sashi Gadi - Qi Qi - Maxwell Afari Gyamfi - Rency S. Varghese - Leslimar Rios-Colon - Uchechukwu Chimeh - Habtom W. Ressom - Deepak Kumar journal: Cancers year: 2023 pmcid: PMC10046356 doi: 10.3390/cancers15061715 license: CC BY 4.0 --- # MicroRNA-483-5p Inhibits Hepatocellular Carcinoma Cell Proliferation, Cell Steatosis, and Fibrosis by Targeting PPARα and TIMP2 ## Abstract ### Simple Summary Hepatocellular carcinoma (HCC) is the fifth leading and highly aggressive lethal liver cancer. The most common cause of HCC is liver cirrhosis because of multiple underlying etiologies, such as chronic hepatitis, nonalcoholic fatty liver disease (NAFLD), alcoholic fatty liver disease (AFLD), and hepatoxicity. In the current study, we characterize the role of microRNA-483-5p in NAFLD/AFLD and HCC progression and its potential use as a prognostic biomarker. ### Abstract MicroRNAs (miRNAs) are small non-coding RNA molecules that bind with the 3′ untranslated regions (UTRs) of genes to regulate expression. Downregulation of miR-483-5p (miR-483) is associated with the progression of hepatocellular carcinoma (HCC). However, the significant roles of miR-483 in nonalcoholic fatty liver disease (NAFLD), alcoholic fatty liver diseases (AFLD), and HCC remain elusive. In the current study, we investigated the biological significance of miR-483 in NAFLD, AFLD, and HCC in vitro and in vivo. The downregulation of miR-483 expression in HCC patients’ tumor samples was associated with Notch 3 upregulation. Overexpression of miR-483 in a human bipotent progenitor liver cell line HepaRG and HCC cells dysregulated Notch signaling, inhibited cell proliferation/migration, induced apoptosis, and increased sensitivity towards antineoplastic agents sorafenib/regorafenib. Interestingly, the inactivation of miR-483 upregulated cell steatosis and fibrosis signaling by modulation of lipogenic and fibrosis gene expression. Mechanistically, miR-483 targets PPARα and TIMP2 gene expression, which leads to the suppression of cell steatosis and fibrosis. The downregulation of miR-483 was observed in mice liver fed with a high-fat diet (HFD) or a standard Lieber-Decarli liquid diet containing $5\%$ alcohol, leading to increased hepatic steatosis/fibrosis. Our data suggest that miR-483 inhibits cell steatosis and fibrogenic signaling and functions as a tumor suppressor in HCC. Therefore, miR-483 may be a novel therapeutic target for NAFLD/AFLD/HCC management in patients with fatty liver diseases and HCC. ## 1. Introduction Hepatocellular carcinoma (HCC) is the fifth-most common cancer in the world and the third cause of cancer-related mortality [1]. In 2023, the American Cancer Society (ACS) estimated 41,210 new cases of HCC and 29,380 deaths from primary liver cancer and intrahepatic bile duct cancer in the USA (https://www.cancer.org/cancer/liver-cancer/about/what-is-key-statistics.html; accessed on 12 February 2023) [2]. The most common cause of HCC is liver cirrhosis due to various underlying etiologies such as chronic hepatitis B and C viral infection and nonalcoholic fatty liver disease (NAFLD), and alcoholic fatty liver disease (AFLD) [3,4,5,6]. Furthermore, diabetes and obesity epidemics have increased the prevalence of NAFLD and its more severe form, nonalcoholic steatohepatitis (NASH) [7]. Therefore, the risk for HCC, even before the development of cirrhosis, is likely to rise [8,9]. Hepatic steatosis is the earliest event in NAFLD and is characterized by triglyceride accumulation within hepatocytes [10,11]. Ten to twenty percent of steatotic livers develop into NASH, progressing to HCC via fibrosis and cirrhosis [10]. AFLD is another serious public health threat [6,12,13,14,15], and it is estimated that approximately ~$5\%$ of the US adult population is affected by this disease. Similar to NAFLD, AFLD progresses histologically defined stages from hepatic steatosis to NASH or alcoholic steatohepatitis (ASH), fibrosis, cirrhosis, and finally to HCC [5,6]. There are various therapeutic approaches for the treatment of HCC, for example, the use of cell therapy, immune checkpoint inhibitors and new tyrosine kinase inhibitors (sorafenib and regorafenib), epigenetic modifiers, and identification and implementation of predictive biomarkers in the treatment of HCC as reviewed recently [16]. However, life expectancy varies depending on the stage of diagnosis, which broadens therapeutic options and improves prognosis [17]. MicroRNAs (miRNAs) are a class of small non-coding RNA molecules (19–25 nucleotides) that binds to 3′ untranslated regions (UTRs) of genes and inhibit gene expression [18]. Several miRNAs have been shown to regulate cell proliferation/survival/apoptosis, cellular metabolism, and stress-related pathways [19,20,21]. Because of their important role in regulating gene expression, miRNAs have been proposed as diagnostic, prognostic, and risk stratification biomarkers in several human cancers, including HCC [22,23,24,25,26,27,28]. Various earlier studies have reported the dysregulation of miR-483 mature variants -3p and -5p across multiple cancer types. For example, elevated expression of miR-483 has been reported in $100\%$ of Wilms’ tumors [29] and also is up-regulated at the hyperplastic stage of pancreatic tumors [30], and approximately $30\%$ of colon, breast, and liver tumors also showed high or even extremely high levels of miR-483-3p expression [29]. The upregulation of miR-483-5p was also associated with poorer disease-specific survival in patients with adrenocortical carcinomas [31], and the expression of miR-483-5p promotes HCC cell proliferation by targeting the suppressor of cytokine signaling 3 (Socs3) [32]. Interestingly, the deregulation of this miRNA by lncRNA might have a role in various cancers. For example, lncRNA TC39A-AS1 acts as a competing endogenous RNA in breast cancer by sponging miR-483-3p, indirectly increasing MTA2 expression and tumorigenicity of breast cancer. [ 16]. Similarly, in tongue squamous cell carcinoma (TSCC) lncRNA NR_034085, miRNA processing–related lncRNA (MPRL) directly binds to pre-miR-483 within the loop region and blocks pre-miR-483 recognition and cleavage by TRBP–DICER-complex, thereby inhibiting miR-483-5p generation which leads to upregulation miR-483-5p downstream target-FIS1 expression [33]. These studies suggest that miR-483 acts as an oncogenic micro-RNA in several cancers. Given the complexity of the mechanisms behind NAFLD, AFLD, and HCC development, there remains a gap in the knowledge regarding the role of crucial miRNAs and their downstream gene targets in transitioning from a healthy liver to AFLD/ALD-NASH-HCC. In the current study, we examined the association between miR-483-5p (miR-483) expression in NAFLD/AFLD in vivo mice models and HCC patient tissues to better understand the role of this miRNA in liver diseases. Reports suggest a possible role of miR-483 in liver disease. MiR-483-5p targeted the proprotein convertase subtilisin/kexin type 9 (PCSK9) 3′-UTR, leading to decreased PCSK9 protein and mRNA expression, increased hepatic LDL receptor expression, and enhanced LDLcholesterol uptake [34]. Overexpression of miR-483 in mice liver increased hepatic LDL receptor levels by targeting PCSK9, leading to decreased plasma total cholesterol and LDL cholesterol levels, suggesting that microRNA-483 ameliorates hypercholesterolemia [34]. An earlier study demonstrated that miR-483 targets metalloproteinase 2 (TIMP2) and platelet-derived growth factor-β (PDGF-β), thus suppressing CCl4-mediated mouse liver fibrosis in vivo [35]. Our data suggest that miR-483 overexpression inhibited cell proliferation/migration, induced apoptosis, and dysregulated Notch signaling. Overexpression of miR-483 inhibited cell steatosis and downregulated fibrogenic signaling by targeting peroxisome proliferator-activated receptor alpha (PPARa) and tissue inhibitor of metalloproteinases 2 (TIMP2), respectively. We also found downregulation of miR-483-5p in both mice models of NAFLD and AFLD and human HCC tissue samples. ## 2.1. HCC Tumor Samples and miR-483 Expression Patients were recruited at MedStar Georgetown University Hospital (MGUH), Washington, DC. The study is conducted through a protocol approved by the Georgetown-Howard Universities Center for Clinical and Translational Science (GHUCCTS) Institutional Review Board (IRB), Washington, DC, under Protocol #2014–0059 “Multi-omic Approaches for Liver Cancer Biomarker Discovery.” The collection and use of the tissues was approved by the IRB of Georgetown University, Washington, DC, under Protocol #2007–345 “Establishment of the High-Quality Tumor Biobank and Clinical Database”. All patients signed a consent form permitting the use of donated tissue. The consent forms and their content were reviewed and approved by the IRB. Detailed characteristics of the study cohort are described in Supplementary Table S1. In this study, we analyzed 80 samples consisting of 40 tumor tissues (HCC) and 30 adjacent normal tissues (Adj-N), and 10 adjacent cirrhotic tissues (Adj-C) acquired from 40 HCC patients, including 14 African American (AA), 16 European American (EA), and 10 Asian American (AAM). We excluded from analysis 1 sample from an AA patient and one sample from an Asian patient due to outlier screening of the corresponding mRNA-Seq data. Total RNAs were isolated from samples as previously described [36]. Briefly, RNA samples were isolated using the RNeasy Plus Universal Mini Kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. The quality and concentration of RNA were estimated using the NanoDrop ND-1000 spectrophotometer. Further analysis of RNA integrity was performed using the Agilent RNA 6000 Nano Kit on the Agilent 2100 Bioanalyzer. Libraries were prepared using the TruSeq RNA Access Library Prep Kit. Sequencing was performed in an Illumina HiSeq 4000 instrument using a 150 bp pair-end (PE150). The mRNA-Seq data contained an average of 33 M reads per sample. The fastq files were then imported into Partek Flow for quality assessment, alignment, and estimating transcript abundance. Alignment was performed using the spliced transcripts alignment (STAR) algorithm, and reads were quantified using the Expectation Maximization (E/M) method implemented in Partek Flow with Trimmed Mean of M-values (TMM) used for normalization. The miRNA-Seq data were analyzed using the QIAseq miRNA quantification data analysis software (https://geneglobe.qiagen.com/us/analyze, accessed on 12 February 2023). The first step is the primary analysis, where the unique molecular index (UMI) counts are calculated, and primary miRNA mapping is performed. In the secondary analysis step, the UMI counts are analyzed to calculate the changes in miRNA expression. The quantified data were then normalized using the TMM method before any statistical analysis was performed. All raw and pre-processed miRNA-seq, mRNA-seq, and DNA methylation data have been deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE176289 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE176289, accessed on 12 February 2023), as indicated [37]. ## 2.2. Cell Culture Human HCC HepG2 (cat #HB-8065), SK-Hep1(cat #HTB-52), and Hep3B (Cat #HB-8064) cells were obtained from American Type Culture Collection (ATCC). HCC cells were grown in DMEM medium (Invitrogen, Carlsbad, CA, USA) supplemented with $5\%$ Fetal Bovine Serum (FBS, Access Biologicals, Vista, CA, USA) and 50 U/mL penicillin/streptomycin. HepaRG, a human bipotent progenitor cell line capable of differentiating into 2 different cell phenotypes (i.e., biliary-like and hepatocyte-like cells) [38], was obtained from ThermoFisher Scientific (Waltham, MA, USA). The terminally differentiated HepaRG cells (Cat #HPRGC10) and media ingredients were obtained from ThermoFisher Scientific (Waltham, MA, USA). As per the manufacturer’s instructions, specific thawing and plating media (Cat #HPRG770) were used, and cells were expanded in William’s E Medium (Cat #12551032) supplemented with $1\%$ GlutaMax (Cat #35050061). Human cryopreserved hepatocytes (HPCH05+) and hepatocytes thawing and plating medium were obtained from Xenotech (Kansas City, KS, USA). Cells were incubated at 37 °C in a cell culture incubator supplied with $5\%$ CO2 and used in experiments when they reached 70–$80\%$ of the confluence level. ## 2.3. miRNA Transfection HepaRG, HepG2, SK-Hep1, Hep3B cells, and human hepatocytes were grown in 6-well plates (1 × 105 cells/wells) for 24 h before transfection. Cells were transfected with mirVana miRNA-483-5p mimic (miR-483) (AAGACGGGAGGAAAGAAGGGAG; Cat #4464066), mirVana miRNA-483-5p inhibitor (miR-483 Inh.) ( AAGACGGGAGGAAAGAAGGGAG; Cat #4464084) or mirVana miRNA Mimic Negative Control #1 (NC) (Cat #4464058). All mirVana miRNA mimics, negative control, and miRNA inhibitor were obtained from Ambion (Austin, TX). HCC cells and hepatocytes were transfected with 100 nM of negative control (NC) or 25–100 nM of miR-483 mimic or miR-483 inhibitor using the Lipofectamine-2000 reagent (Invitrogen, Waltham, MA, USA). Cells were harvested 30 h after transfection. The expression of miR-483 target genes was analyzed by RT/qPCR, and protein expression was assessed by immunoblotting. ## 2.4. RT/qPCR Total RNAs from HepaRG, HepG2, SK-Hep1, and Hep3B cells were isolated using TRIZOL reagent (Life Technologies, Carlsbad, CA, USA). In other experiments, HepaRG, HepG2, SK-Hep1, Hep3B cells, and human hepatocytes were transfected with NC mimic or miR-483 mimic separately for 30 h. Cells were washed with PBS, and total RNAs were isolated using TRIZOL. Equal amounts of RNA (1 µg) were reverse transcribed using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Waltham, MA, USA). Then, cDNA was incubated with Power SYBR Green PCR master mix (Applied Biosystems) with appropriate forward and reverse primers of indicated genes (Supplementary Table S2). GAPDH was used as an internal control. For miR-483 expression analysis, total miRNAs from HepaRG, HepG2, SK-Hep1, Hep3B cells, and human hepatocytes were isolated using the mirVana microRNA Isolation Kit (Thermo Fisher Scientific, Waltham, MA, USA). The total miRNAs (10 ng) were reverse transcribed using primers specific for miR-483 and RNU44 (Assay ID 002338 and 001094, respectively, Applied Biosystems, Carlsbad, CA, USA) and TaqMan Reverse Transcription reagents (Applied Biosystems). Expression of miR-483 and RNU44 was quantified by RT/qPCR using TaqMan PCR master mixture and Taqman expression assay primers. NU44 expression was used as an internal control. To quantify miR-483 expression in mice liver, we utilized primers specific for miR-483 and snoRNA202 (Assay ID 001232; as an endogenous control). All PCR reactions were performed on a QuantStudio-3 PCR system (Applied Biosystems), and relative quantitation was analyzed according to the manufacturer’s protocols. ## 2.5. Western Blotting Immunoblotting was performed as described previously [39]. Briefly, after transfection with miR-483 or NC mimics/inhibitors, HCC cells were lysed in cell lysis buffer (Cell Signaling Technology, Danvers, MA, USA) containing a protease inhibitor cocktail (Roche, Indianapolis, IN). After centrifugation at 10,000 RPM for 15 min, the cell lysate supernatants were used for protein qualification. Protein concentrations were measured using the Bio-Rad protein assay reagent (Bio-Rad, Hercules, CA, USA). Sixty micrograms of cell lysates were electrophoresed by using NuPAGE 4–$12\%$ Bis-Tris-SDS gels (Invitrogen, Waltham, MA, USA), and proteins were then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, USA). After washing the membranes with 1× Tris-buffered saline with $0.1\%$ Tween 20 Detergent (TBS-T), the membranes were blocked in 1× blocking buffer (Sigma-Aldrich, St. Louis, MO, USA) for 1 h. The membranes were then incubated with primary antibodies overnight at 4 °C as per the manufacturer’s protocols. The following antibodies were obtained from Cell Signaling Technology (Danvers, MA, USA): anti-Notch1 (Cat #4147s), anti-Notch2 (Cat #5132s), anti-Notch3 (Cat #5276s), anti-Hes1 (Cat #11988s), anti-cleaved-PARP (Cat #9541S), anti-LC3B (Cat #4108S), anti-p62 (Cat #5114s), anti-GAPDH (Cat #5174S), anti-β-actin (Cat #4970S), anti-E-cadherin (Cat #3195S), anti-N-cadherin (Cat #13116S), anti-Vimentin (Cat#5741S), anti-Nanog (Cat #4903s), anti-p21(Cat #2947s), anti-CD44 (Cat #3570s), anti-TIMP2 (Cat #5738s), anti-MMP2 (Cat #13132s), anti-TGFβ (Cat #3711s), anti-fatty-acid synthase (FASN) (Cat #3180S), anti-SCD1 (Cat #2794S), and anti-ACC (Cat #3662S). We also obtained anti-L-FABP (Cat #ab7366) from Abcam, anti-TNFAIP8 (Cat #15790-1-AP) antibody, and anti-PPARA (Cat #15540-1-AP) from Proteintech (Rosemont, IL, USA); anti-PPRA-γ (Cat #sc-7273) and anti-SREBP1 (Cat #sc-13551) from Santa Cruz Biotechnology (Dallas, TX, USA). After overnight incubation, the membranes were washed 3 times with TBST and then incubated in the appropriate secondary antibody (1:10,000 dilution) (Jackson ImmunoResearch, West Grove, PA, USA) for 1 h at room temperature. The immunoreactive bands were visualized using Enhanced chemiluminescence (ECL) detection reagents (Signagen Laboratories, Rockville, MD, USA). The immunoblots were visualized using the Azure C-500 Bio-system. ## 2.6. Cell Survival Assay Cells (1 × 104 cells/well) were grown in 96 plates and transfected with NC (100 nM) or increasing concentrations of miR-483 mimic (25–100 nM) for 72 h. In a separate experiment, cells transfected with NC or miR-483 were also treated with sorafenib (5µM), regorafenib (2.5 µM), or their corresponding vehicle controls and incubated for 72 hrs. Cells were then incubated with 5 µL/well of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) reagent (5 mg/mL) for 1 h at 37 °C in a cell culture incubator. Cells were then carefully washed with PBS, and formazan crystals were dissolved in 100 µL DMSO. Cell survival was determined by quantifying absorbance at 570 nm using a Fluostar Omega plate reader (BMG Lab tech, Cary, NC, USA). All experiments were repeated 3 times. ## 2.7. Cell Colony Formation Assay HepaRG, HepG2, and SK-Hep1 cells (1 × 105 cells/well) were grown in 6-well plates in triplicates for 18 h and transfected with NC (100 nM) or miR-483 (100 nM) for 24 h. After transfection, cells were trypsinized and counted, and live cells (5000 cells/well) were re-plated in 6-well plates in triplicates. Cells were allowed to grow for 7–10 days until colonies were visible. Cell colonies were then washed with PBS for 1 min, fixed with cold methanol, and stained with $0.1\%$ crystal violet for 1 h. Cell colonies were washed with distilled water and allowed to dry. Finally, cell colonies were photographed, counted, and plotted. ## 2.8. Cell Migration Assay The effect of miR-483 mimic or miR483 inhibitor (miR-483 Inh.) on the migratory ability of HepaRG and SK-Hep1 cells was determined by wound healing migration assay as described previously [39,40]. Cells (1 × 106 cells/well) were grown in a 6-well plate for 18 h and transfected with 100 nM of NC, miR-483 mimic, or miR-483 inhibitor mimic for 24 h. After transfection, a cell monolayer was scraped using a micropipette tip (A0). At 24 h post-wounding (A24), cells were photographed, and the migration gap length was calculated using ImageJ software v.1.8.0 (https://imagej.nih.gov/ij/, accessed on 12 February 2023). The percent wound closure was calculated using the formula [(A0 − A24)/A0] × 100 and plotted. ## 2.9. Luciferase Assay HepG2 and SK-Hep1 cells (1 × 104 cells/well) were transfected with 0.5 µg of TIMP2-3′UTR-Luciferase construct or PPARA-3′UTR-Luciferase (OriGene, custom designed) in 6-well plates for 18 h. Cells were then transfected with 100 nM of NC and miR-483 mimic or miR-483 mutant mimics (custom-designed from Integrated DNA Technologies) for 24 h. Transfected cells were washed with PBS, lysed, and 20 µg lysates were mixed with luciferase substrate (Promega, Madison, WI, USA). Plates were covered with aluminum foil to protect them from light and incubated at room temperature for 20 min. Fluorescence was measured using a Fluostar Omega plate reader (BMG Lab Tech, Cary, NC, USA), and relative luciferase activity was measured and plotted. ## 2.10. Development of NAFLD and AFLD Mouse Models All animal handling and procedures were carried out as per NIH Guidelines for the Care and Use of Laboratory Animals and approved by the NCCU-Institutional Animal Care and Use Committee (NCCU-IACUC Protocol No. is MG-02-26-2010). For the NAFLD development model, 10-week-old male C57BL/6J mice were fed a regular chow diet (control diet, $12\%$ calories as fat; $$n = 5$$) or a high-fat diet (HFD, $45\%$ calories as fat; $$n = 5$$) for 16 weeks. For the AFLD model, 10-week-old male C57BL/6J mice were ear tagged and randomly assigned to one of 2 groups and either pair-fed a control diet ($$n = 5$$) or a standard Lieber-Decarli liquid diet containing $5\%$ EtOH ($$n = 5$$) (representing $27.5\%$ of the total caloric intake), for 8 weeks as previously described [41,42]. Liquid diets, purchased from DYETS Inc (Bethlehem, PA, USA), were based upon the Lieber-DeCarli EtOH formulation and provided 1 kcal/mL. Our pre-established inclusion/exclusion criteria were that animals would be excluded from the analysis if they were too sick or died before the end of the study. After 16 weeks of high-fat diet feeding or after 8 weeks of a standard Lieber-Decarli liquid diet, mice were anesthetized with isoflurane and sacrificed. Livers were isolated, weighed, and sections were rapidly dissected, snap-frozen in liquid nitrogen, and kept at −80 °C. A part of the fresh liver tissues was fixed in $10\%$ formalin liver slices prepared using a cryostat. The sections were stained with hematoxylin and eosin (H&E) staining for histological examination, as described previously [42]. Total RNAs from liver tissues were isolated and purified, and the expression of miR-483, fibrosis markers, or notch signaling gene expression was analyzed by RT/qPCR. ## 2.11. Statistical Analysis All experiments were performed in triplicates and presented as mean ± SEM. Differences between groups were analyzed using a 2-tailed Student’s t-test. A p-value of < 0.05 was considered statistically significant. Statistical significance was determined by Graph Pad Prism 9 software (GraphPad Software Inc., La Jolla, CA, USA). ## 3.1. Downregulation of miR-483 Expression Activates Notch Signaling in HCC Tissues To investigate the biological significance of miR-483 in HCC, we first analyzed the expression levels of miR-483-5p and miR483-3p in liver hepatocellular carcinoma (LIHC) as reported in The Cancer Genome Atlas (TCGA) data set using the MIR-TV portal (http://mirtv.ibms.sinica.edu.tw/analysis.php; accessed on 9 August 2022). Analysis of the available TCGA data suggests a significant downregulation of both miR-483-5p ($$n = 345$$; $p \leq 0.001$) and miR-483-3p ($$n = 364$$; $p \leq 0.001$) expression in HCC tumors compared with normal liver tissues ($$n = 50$$) (Figure 1A, upper and lower panels). To support this observation, we analyzed the expression of miR-483-5p and miR-483-3p in HCC tumor tissues from African American (AA), European American (EA), and Asian American (AAM) patients as described in the method section (Figure 1B). The expression of miR-483-5p and miR-483-3p were significantly ($p \leq 0.001$) down-regulated in HCC tissues from AA ($$n = 13$$) and EA ($$n = 16$$) patients but not in AAM patients ($$n = 9$$) compared with normal liver tissues (Figure 1B, upper and middle panels). Since numerous reports suggest that the Notch 3 receptor is constitutively active in HCC [43,44], we also analyzed its expression in our sample cohort. Our data suggest that Notch 3 gene expression was significantly ($p \leq 0.05$) higher in the HCC tissues of AA, EA, and AAM patients compared with matched normal liver tissue samples (Figure 1B, lower panel). We then asked whether the miR-483-5p (hereafter called miR-483) expression is co-related with the Notch 3 expression. We analyzed miR-483 and Notch 3 expression in HepaRG (a human bipotent progenitor liver cell line) and HCC cell lines HepG2 and SK-Hep1. RT/qPCR data suggest that miR-483 endogenous expression was not significantly changed in HepG2 as compared with HepaRG cells, but a significantly higher expression was observed in SK-Hep1 cells (Figure 1C, upper panel). However, Notch 3 expression was significantly higher in HCC HepG2 and SK-Hep1cells compared with HepaRG cells (Figure 1C, lower panel). Immunoblotting data suggest that the expression of Notch 3 protein was downregulated in SK-Hep1 cells compared with HepaRG or HepG2 cells. No change in Notch 2 and Notch 1 expression between HepaRG, HepG2, and SK-Hep1 was observed. Interestingly, the Notch downstream target Hes1 was downregulated in SK-Hep1 cells, whereas miR-483 expression was significantly upregulated (Figure 1C (upper panel),D). Our overall data suggest that the expression of miR-483 is downregulated in HCC tumors, and expression of miR-483 may affect Notch(s) signaling in HCC tumors and HCC cell lines. ## 3.2. Overexpression of miR-483 Dysregulates Notch Signaling in HCC Cells Notch signaling plays various roles in HCC by regulating tumorigenesis, angiogenesis, invasion, and metastasis [45]. Increased Notch expression has also been associated with poor prognosis in HCC patients [45,46]. In addition, our data demonstrated that miR-483 expression is downregulated in HCC patients’ tumor tissues, and downregulation of miR-483 is associated with the upregulation of Notch 3. To further explore an association between the regulation of miR-483 and Notch(s) signaling in HCC, we analyzed the impact of miR-483 overexpression on Notch signaling in HepaRG and HCC cells. HepaRG and HepG2, and SK-Hep1 cells were transfected with negative control (NC) or miR-483 mimic, and the overexpression of miR-483 was analyzed by RT/qPCR (Figure 2A). Under similar experimental conditions, we also analyzed the expression of Notch 1, Notch 2, Notch 3 and Notch 4 genes by RT/qPCR (Figure 2B). The overexpression of miR-483 increased the expression of Notch 1 and Notch 3 and decreased the expression of Notch 2 and Notch 4 in HepaRG cells (Figure 2B). Interestingly, overexpression of miR-483 significantly reduced Notch 3 expression in HepG2 and SK-Hep1 cells (Figure 2C, left and right panels). Also, the Notch downstream target HES 1 expression was downregulated in SK-Hep1 (Figure 2C, right panel). No significant change in Notch 4 expression was observed when miR-483 was overexpressed in HepaRG, HepG2, and SK-Hep1 cells (Figure 2A–C). Further, we analyzed the effect of miR-483 overexpression on full-length Notch 3, Notch 2, and Notch 1 proteins, the cleaved Notch (s) transmembrane/intracellular (NTM) fragments, and Notch downstream Hes1 protein expression by immunoblotting (Figure 2D, left and right panels). Transient transfection of miR-483 mimic (50 and 100 nM) decreased expression of full-length Notch 3, Notch 2, and Notch 1 and NTM cleaved fragments in HepaRG and HepG2 cells. Full-length Notch 3 and NTM regions expression also decreased in SK-Hep1 cells after overexpression of miR-483. Hes1 expression decreased in HCC HepG2 and SK-Hep1 cells after overexpression of miR-483 (Figure 2D, left and right panels). On the other hand, the inactivation of miR-483 by miR-483 inhibitor stabilized full-length Notch 3, Notch 2, and NTM regions, and no effect on HES1 expression was observed in our cellular models (Figure 2E). Collectively, our data suggest that miR-483 affects/downregulates Notch signaling in HCC cells. However, the underlying molecular mechanisms need to be further investigated. ## 3.3. miR-483 Inhibits HCC Hallmarks and Increases Sensitivity toward Anti-HCC Drugs Since the overexpression of miR-483 downregulates Notch signaling in HCC cells, we further analyzed the impact of miR-483 on HCC cell survival, colony formation ability, migration, and epithelial-mesenchymal transition (EMT). Our data demonstrate that dose-dependent overexpression of miR-483 inhibits cell survival in HepaRG ($72\%$), HepG2 ($14\%$), SK-Hep1 ($63\%$), and Hep3B ($14\%$) compared with NC (100 nM) transfected cells (Figure 3A). Cell colony formation assay showed that miR-483 inhibits HepaRG ($61\%$), HepG2 ($54\%$), and SK-Hep1 ($56\%$) cell colony formation (Figure 3B, upper and lower panels). EMT is essential in the transition from localized disease to invasion and metastasis in cancers [47] and since EMT plays an important role in cell migration and invasion in HCC [48]. Here we analyzed the effect of miR-483 on cell migration using a cell scratch assay (Figure 3C). We observed a significant reduction in cell migration in HepaRG by ~$70\%$ and SK-Hep1 cells by ~$74\%$ transfected with miR-483 mimics compared with NC transfected cells (Figure 3C). Inhibiting miR-483 did not have any significant effects, compared to NC-transfected cells (Figure 3C, left and right panels). In addition, we studied the changes in the expression of various EMT markers when the miR-483 expression is modulated (Figure 3D). Our immunoblotting data suggest that expression of miR-483 increased E-cadherin and decreased N-cadherin expression in HepaRG and HepG2 cells. In SK-Hep1 cells, miR-483 expression increased E-cadherin, N-cadherin, and decreased Vimentin. Inactivation of miR-483 using an inhibitor increased Vimentin in all three cell lines studied, suggesting that miR-483 suppresses EMT in HepaRG and HCC cells and inactivation of endogenous miR-483 promotes EMT in a cell dependent manner (Figure 3D). Since miR-483 inhibits cell survival and migration, we also analyzed the effect of miR-483 on the expression of Nanog and CD44 cancer stem cell markers and oncogenic TNFα-Induced Protein 8 (TNFAIP8) marker. Overexpression of miR-483 decreased the expression of Nanog, CD44, and TNFAIP8 in HepaRG and CD44 and TNFAIP8 in HepG2 cells (Figure 3E). The expression of Nanog, CD44, and TNFAIP8 in SK-Hep1 also decreased after overexpression of miR-483 (Figure 3E), suggesting that miR-483 inhibits HCC progression. To further explore the role of miR-483 in cell apoptosis, we assessed cleaved-PARP (c-PARP) expression in HepaRG, HepG2, and SK-Hep1 cells after transfection with miR-483 mimic and NC. Overexpression of miR-483 increased cleaved-PARP (c-PARP) expression in HepaRG, HepG2, and SK-Hep1 cells compared with NC transfected cells (Figure 3F). Inactivation of miR-483 decreased cleaved-PARP (c-PARP) expression in all three cell lines, suggesting that miR-483 expression induced cell apoptosis (Figure 3F). Since our data suggest that miR-483 potentially suppresses HCC hallmarks and induces cell apoptosis, we further examined the potential therapeutic role of miR-483 in combination with anti-neoplastic drugs (sorafenib and regorafenib). As expected, expression of miR-483 decreased cell survival in HepaRG and HepG2, SK-Hep1, and Hep3B HCC cells (Figure 3G, left and right panels). Exposure of sorafenib (5 µM) or regorafenib (2.5 µM) alone decreased cell survival in all cell lines. Interestingly, sorafenib (5 µM) or regorafenib (2.5 µM) combined with miR-483 expression further reduced cell survival in all cell lines compared with NC transfected and sorafenib or regorafenib-only treated cells (Figure 3G). These results suggest that miR-483 could have an additive or synergistic effect potentiating drug sensitivity in HCC. Indeed, our data indicate that miR-483 suppressed cancer hallmarks in HCC and could be a potential biomarker for this disease. ## 3.4. miR-483 Inhibits HCC Cell Steatosis by Modulation of Lipogenic Gene Expression NAFLD is a major risk factor for the development of HCC in non-cirrhotic patients, and the expression of lipogenic enzymes/proteins modulates hepatic steatosis and NAFLD development [49]. Since miR-483 inhibits HCC cell proliferation, we next address the impact of miR-483 on steatosis, an early event in NAFLD development. We transfected human hepatocytes and HepaRG, HepG2, SK-Hep1, and Hep3B cells with miR-483 mimic (50 and 100 nM). We then evaluated the effects of miR-483 on the regulation of genes involved in lipogenesis, such as Acetyl-CoA carboxylase (ACC), liver fatty acid-binding protein-1 (L-FABP1), fatty acid synthase (FASN), stearoyl-CoA desaturase-1 (SCD1), transcription factors such as sterol regulatory element-binding protein 1 (SREBP1), and peroxisome proliferator-activated receptor alpha and gamma (PPARG). Overexpression of miR-483 in human hepatocytes slightly decreased FASN and SCD1 expression, but no significant changes in ACC and PPARG were observed. On the other hand, increased FABP1 and SREBP1 expressions were observed compared with NC transfected cells (Supplementary Figure S1A). Similarly, SREBP1, PPARG, and FABP expression were significantly decreased in HepaRG cells, SCD1 in SK-Hep1 cells, and SREBP1 in Hep3B cells. At the same time, the overexpression of miR-483 increased FASN and ACC in HepaRG, HepG2, and SK-Hep1 cells compared with NC transfected cells (Supplementary Figure S1B), suggesting that miR-483 dysregulates lipogenic signaling in hepatocytes and HCC cells. On the contrary, we inactivated endogenous miR-483 by using a miR-483 inhibitor, and lipogenic gene expression was analyzed. Inactivation of endogenous miR-483 expression increased the expression of SCD1, FASN, PPARG, and SREBP1 in HepaRG cells, ACC, SCD1, and FASN in HepG2 cells, and FASN in SK-Hep1 cells compared with NC transfected cells (Figure 4A). These results suggest that miR-483 has a role in the modulation of lipogenic gene expression. We then analyzed the expression of miR-483 in HepaRG and HepG2 after exposure of the cells to fatty acids such as oleic acid (OA), elaidic acid (EA), palmitic acid (PA), lauric acid (LA), stearic acid (SA), myristic acid (MA), linoleic acid (LNA), and cholesterol (CHO) which are known to induce cell steatosis. Treatments with OA, LA, CHO, PA, MA, SA, and LNA increased the expression of miR-483 in HepaRG cells compared to untreated/control (Figure 4B, upper panel). Similarly, OA, LA, CHO, EA, and MA significantly increased miR-483 expression in HepG2 cells (Figure 4B, lower panel), indicating that exposure to fatty acid/cholesterol increased endogenous miR-483 expression in HepaRG and HepG2 cells. Since our data suggest that overexpression or inactivation of miR-483 dysregulates lipogenic signaling, we then examined the role of miR-483 on cell steatosis after transfecting HepaRG, HepG2, SK-Hep1, and Hep3B cells with miR-483 mimic (50 and 100 nM) or miR-483 inhibitor (100 nM) compared to their corresponding NC (100 nM) control. The cells were first transfected with NC mimic, miR-483 mimic, or miR-483 inhibitor for 24 h and exposed to OA for an additional 24 h, and cell steatosis was examined using Oil Red O (ORO) staining (Figure 4C). Overexpression of miR-483 (50–100 nM) significantly decreased lipid droplet accumulation in HepaRG and HCC cells. In contrast, inhibition of miR-483 restored/promoted cell steatosis (Figure 4C). We quantified steatosis after ORO staining as described in the methods section, and our data suggest that miR-483 inhibits cell steatosis in HepaRG, HepG2, SK-Hep1, and Hep3B cells, whereas inactivation of miR-483 promotes cell steatosis significantly (Figure 4D, upper and lower panels). Furthermore, immunoblotting data suggest that overexpression of miR-483 decreased expression of SCD1, FASN, PPARγ, L-FABP, and SREBP1 in HepaRG cells and SCD1, L-FABP, and SREBP1 in HepG2 cells (Figure 4E). Increased expression of FASN, ACC, and PPARG was also observed in HepG2 cells transfected with miR-483 (Figure 4E), suggesting that miR-483 modulates the expression of steatosis/lipogenesis markers that leads to inhibition of cell steatosis. Autophagy plays an essential role in lipid metabolism/lipid droplet clearance [50]. Since overexpression of miR-483 inhibits cell steatosis, we analyzed the effect of miR-483 on the expression of known autophagy biomarkers such as LC3B I/II and p62 in HCC cells (Figure 4F). Overexpression of miR-483 in HepaRG, and HCC HepG2, SK-Hep1, and Hep3B cells increased LC3B I/II and decreased p62 compared with NC transfected cells (Figure 4F), suggesting that miR-483 could inhibit cell steatosis by inducing autophagy. Our data suggest that miR-483 dysregulated lipogenic gene expression and suppressed cell steatosis by activating cellular autophagy. ## 3.5. miR-483 Inhibits Fibrogenic Signaling in HCC NAFLD/AFLD progresses through histologically defined stages from hepatic steatosis to steatohepatitis (NASH), fibrosis, cirrhosis, to HCC. To investigate the role of miR-483 in fibrosis, we inactivated miR-483 in HepaRG and HCC hepG2 and SK-Hep1 cells (Figure 5A). The inactivation of endogenous miR-483 expression by miR-483 inhibitor increased expression of TGFβ, tissue inhibitor of metalloproteinase 2 (TIMP2) and p21 in HepaRG cells, TGFβ, TIMP2, p21, Cytokeratin 7 in HepG2 cells and p21 in SK-Hep1 cells several folds compared with NC transfected cells (Figure 5A) suggesting that miR-483 can affect the expression of p21, TGFβ, TIMP2, and Cytokeratin 7. As earlier reported, miR-483 targets TGFβ [51], and our data indicate that miR-483 overexpression dysregulated TGFβ expression and decreased TIMP2 and matrix metallopeptidase 2 (MMP2) gene expression. We validated these results by immunoblotting using HepaRG and HepG2 cells. The overexpression of miR-483 decreased TIMP2 and TGFβ expression in HepaRG and HepG2 cells and MMP2 in HepG2 cells compared to NC-transfected cells (Figure 5B). Interestingly, when HepaRG and HepG2 cells were exposed to carbon tetrachloride (CCl4), a known fibrosis inducer agent in liver cells, the increased expression of miR-483 was observed in HepG2 cells and significantly decreased expression of miR-483 was observed in HepaRG cells compared with untreated cells (Figure 5C, upper panel). Immunoblotting data demonstrated that overexpression of miR-483 and CCl4 (10 mM) treatments decreased TIMP2 and TGFβ protein expression in HepG2 cells and MMP2 and TIMP2 in HepaRG cells (Figure 5D, left and right panels). These results suggest that CCl4 exposure could modulate miR-483 expression affecting downstream targets TIMP2 and TGFβ expression. Taken together, our data suggest that miR-483 downregulates fibrogenic signaling in HCC cells. ## 3.6. miR-483 Targets PPARa and TIMP2 and Inhibits Cell Steatosis and Fibrosis To further understand the molecular mechanisms of how miR-483 inhibits cell steatosis and fibrogenic signaling in HCC cells, we used TargetScan (http://www.targetscan.org/vert_72/; accessed on 12 August 2022) to identify possible 3′-untranslated regions (UTR) of genes targeted by miR-483. TargetScan analysis revealed that miR-483 binds to the UTRs of the PPARA gene (PPARA isoform 5; 2638–2644), TIMP2 (2089–2095), and p21 (705–711) (Figure 6A). Interestingly, peroxisome proliferator-activated receptors (PPARs) play a critical role in lipid metabolism and homeostasis in multiple cell types [52,53]. Also, an earlier report demonstrated that miR-483 targets TIMP2 and PDGF-β leads to suppressed CCl4-mediated liver fibrosis in mice [35], and miR-483 also targets TGFβ which is involved in fibrogenic responses in pulmonary arterial hypertension (PAH) [51]. The binding sequences of miR-483 with these genes’ UTRs are presented in Figure 6A. The binding of miR-483 to the 3′ UTR of TIMP2 and PPARa genes was analyzed by luciferase reporter assay. Co-transfection of 3′ UTR of TIMP2 and PPARA gene and wild-type miR-483 decreased luciferase activity significantly (Figure 6B, left panel), but no significant change in luciferase activity was observed in mutant-type miR-483 transfected cells (Figure 6A,B (right panel)). To further address the molecular mechanisms of how miR-483 regulates the expression of these genes, we first analyzed the effect of miR-483 on the regulation of PPARA, TMIP2, and p21 gene expression and protein expression in HepaRG and two HCC HepG2 and SK-Hep1 cell lines. RT/qPCR data suggest that overexpression of miR-483 decreased p21, PPARA[5], and TIMP2 expression significantly in HepaRG cells and PPARA[5] and TIMP2 gene expression in HCC cell lines (Figure 6C). Immunoblotting data suggest that overexpression of miR-483 downregulated the expression of PPARa, p21 TIMP2 in HepaRG and HepG2 cells and PPARa and TIMP2 in SK-Hep1 cells (Figure 6D), suggesting that miR-483 inhibits PPARa and TIMP2. The expression of TGFβ also decreased in HepaRG cells after miR-483 overexpression. Collectively our data suggest that targeting PPARa and TIMP2, miR-483 inhibits cell steatosis and fibrosis in HepaRG and HCC cells. ## 3.7. Downregulation of miR-483 Exacerbates NAFLD and AFLD in Mice Liver Finally, we utilized mouse models of NAFLD and AFLD to understand the biological significance of miR-483 in developing liver disease. C57BL/6J mice were fed either a chow diet (control diet, $12\%$ calories as fat; $$n = 5$$) or a high-fat diet (HFD, $45\%$ calories as fat; $$n = 5$$) for 16 weeks as described previously [42] (Figure 7A). Mice fed with HFD showed hepatic steatosis, and this was not observed in the livers of mice fed with the chow diet [42]. We then analyzed the expression of miR-483 in these liver samples (Figure 7B, left panel). RT/qPCR data demonstrated that the expression miR-483 was significantly decreased in the livers of mice fed with HFD. Moreover, we analyzed the expression of fibrosis biomarkers and Notch(s) receptors in mice liver fed with HFD (Figure 7B, right panels, and Supplementary Figure S2A). Mice fed with HFD showed significantly increased expression of TIMP2, TGFβ ($p \leq 0.05$), and downregulated AST ($p \leq 0.001$) and Notch2 ($p \leq 0.01$) expression in liver tissues. No significant change in p21, Notch1, Notch3, or Notch4 was detected (Figure 7B, right panels, and Supplementary Figure S2A). Similarly, we also analyzed the expression of miR-483, fibrosis markers, and Notch(s) receptors in an AFLD mice model (Figure 7C). Mice fed an EtOH-containing diet showed hepatic steatosis and not in mice liver fed with the control diet [42]. Similar to the NAFLD mice model, expression of miR-483 was also downregulated in mice liver fed with EtOH containing diet ($p \leq 0.01$) compared with control diet-fed animals (Figure 7D, left panel). Fibrogenic TGFβ expression was significantly increased ($p \leq 0.01$) in these samples. However, the expression of TIMP2 and AST1 was downregulated. No significant change in the expression of p21, Notch1, Notch2, and Notch4 was observed in these samples (Figure 7D, right panels, and Supplementary Figure S2B). Increased expression of Notch3 was detected in mice liver fed with EtOH containing diet (Figure 7D, right panels, and Supplementary Figure S2B). Considering these results, our data suggest that miR-483 expression is downregulated in both NAFLD and AFLD mice models that could be involved in the modulation of the fibrogenic signaling in mice liver. ## 4. Discussion Hepatocellular carcinoma (HCC) carries a significant threat of cancer-related mortality [1]; thus, identifying biomarkers in the early stages of NAFLD, AFLD, and subsequent HCC progression is crucial. Due to their stability in various bodily fluids, miRNAs are currently studied as prognostic and diagnostic markers for various diseases, including cancer [54]. A recent study showed the under-expression of miR-370-3p and over-expression of miR-196a-5p in serum exosomes of HCC patients compared to control samples. De-regulation of these miRNAs was also associated with increased tumor size, tumor grade, Tumour Node Metastasis (TNM) stage, and worsened prognosis [55]. Another study showed higher expression of circulating miRNAs such as miR-16 and miR-122 in early-stage HCC patients with high diagnostic efficacy [56]. The diagnostic utility of miR-122-5p, miR-21-5p, and miR-222-3p was also analyzed in the serum samples of patients with hepatitis C viral infection and HCC post-direct-acting antiviral (DAAs) therapy [57]. Interestingly, downregulation of miR-21-5p and miR-122-5p was observed in the HCC post-DAA therapy group compared to control samples. In contrast, higher expression of miR-21-5p and miR-122-5p was detected in the HCV-related HCC group indicating altered expression of these miRNAs during DAAs therapy [57]. These studies underscore the prognostic and diagnostic utility of miRNAs in HCC. In the present study, we investigated the biological significance of miR-483 in NAFLD, AFLD, and HCC. We found that miR-483 plays an important role in NAFLD/AFLD/HCC modulation, and our data shows downregulation of the expression of miR-483 in HCC patients with a diverse racial background compared to healthy controls. Our data is also supported by the TCGA database, demonstrating the downregulation of miR-483-5p and miR-483-3p in HCC. Further, we demonstrated that overexpression of miR-483 inhibited HCC cell survival/migration, increased anti-HCC drug sensitivity, and induced cell apoptosis. Since Notch(s) signaling is constitutively activated in HCC and involved in tumor formation [58], we also analyzed the effect of miR-483 expression regulation in Notch(s) signaling in HCC cells. Our data indicated that overexpression of miR-483 inhibits Notch3 in HepG2 and SK-Hep1 cells, thus downregulating the expression of Notch downstream target HES1. However, the molecular mechanism of how miR-483 dysregulates Notch signaling in HCC remains to be investigated. MicroRNAs play a dual role as a tumor suppressor as well as oncogenes. Numerous studies suggest that under or overexpression of specific miRNA or antagomirs can affect the downstream gene regulatory network/cell signaling pathways, which could lead to reversing the phenotypes in cancer cells [59]. MicroRNAs regulate drug resistance in HCC; for example, the downregulation of miR-122 upregulates ABCB1, ABCF2, and PKM2 and increases resistance against doxorubicin [60], whereas the downregulation of miR-340 activates Nrf2 and enhances resistance against cisplatin [61]. Our data suggest that down regelation miR-483 activates Notch, whereas overexpression of miR-483 inhibits HCC hallmarks and increases sensitivity toward anti-HCC drugs. Analysis of miR-483 expression in the different stages of HCC progression needs to be monitored to control not only HCC progression but also NAFLD and AFLD, and designing strategies for overexpression miR-483 in the liver along with anti-HCC drugs may give a better outcome for NAFLD and AFLD and HCC patients. Liver cirrhosis is one of the main causes of HCC and can be caused by various underlying etiologies like chronic hepatitis B/C viral infection, NAFLD, and AFLD [3,4,5,6]. López-Riera, et al. recently identified nine serum microRNAs, miR-16, miR-21, miR-22, miR-27b, miR-30c, miR34a, miR-122, miR-192, and miR-197 associated with NAFLD severity [62]. Furthermore, miR-22, miR-27b, miR-192, and miR-197 appeared to be NAFLD-specific compared with drug-induced liver injury [62]. As previously reviewed [63], the expression of miR-27a, miR-140-5p, miR-191, miR-222, miR-224, miR-378a-3p, miR-140-5p, miR-483, and miR-520d-5p modulates pathogenesis of hyperlipidemia by targeting proprotein convertase subtilisin/kexin type 9 (PCSK9). Specifically, miR-191, miR-222, and miR-224 miR-483-5p control PCSK9 expression [34,64,65], enhanced hypercholesterolemia [34], and LDL-C uptake in mice liver fed with a high-fat diet [65]. Our data indicate that HCC cells exposed to different fatty acids, particularly oleic acid (OA), lauric acid (LA), and cholesterol (CHO), significantly increased miR-483 expression in HepaRG and HepG2 cells and overexpression of miR-483 downregulated cell steatosis by suppressing lipogenic gene expression in vitro. Our data also suggest that miR-483 targets PPARa and downregulates cell steatosis and overexpression of miR-483 increased LC3B autophagy biomarker expression since cellular autophagy modulates lipid metabolism [50,66]. These results suggest that mir-483 suppresses cell steatosis by targeting PPARa and by induction of autophagy. Hepatic steatosis progresses from fibrosis and cirrhosis to HCC, and earlier studies demonstrated the dysregulation of microRNA expression in these transitions [67,68,69]. For example, higher expression of miR-199 was observed in liver fibrosis [70], but its expression was down-regulated in HCC [71]. Interestingly, the increased expression of miR-483 was detected in advanced cirrhosis patients infected with hepatitis C virus [72], and overexpression of MiR-483 suppresses CCl4 mediated induction of fibrosis in mice liver [35], suggesting that miR-483 modulates liver fibrosis. A recent report suggests that under-expression of miR-483 in serum from patients with idiopathic pulmonary arterial hypertension (IPAH) revealed that miR-483 targets several PAH-related genes, including transforming growth factor-β (TGF-β), TGF-β receptor 2 (TGFBR2), β-catenin, connective tissue growth factor (CTGF), interleukin-1β (IL-1β), and endothelin-1 (ET-1) and overexpression of miR-483 in endothelial cells (ECs) inhibited inflammatory and fibrogenic responses [51]. Similarly, we found that overexpression of miR-483, particularly HepG2 and SK-Hep1 cells suppressed the expression of fibrosis markers such as TIMP2, TGF-β, cytoleratin7, and CX3CR1. Increased expression of miR-483 was observed in HepG2 when cells were exposed to CCl4, leading to the downregulation of TIMP2 and TGFβ. This suggests that miR-483 suppresses fibrogenic signaling in HCC, as reported earlier [35]. Since miR-483 modulates cell steatosis and fibrogenic signaling in HCC cells, we wanted to investigate further how miR-483 affects these processes. We first analyzed the possible targets of miR-483 that are involved in steatosis and fibrosis. Interestingly our analysis revealed that miR-483 targets the PPARA-3’UTR sequence (although poorly conserved). Our reported data suggest that miR-483 binds to the PPARA [5]-3’UTR and affects PPARA expression. The PPARs are known to regulate lipid metabolic enzyme expression and modulate intracellular lipid metabolism, entry of fatty acid into peroxisome and mitochondria, and mitochondrial fatty acid catabolism [73]. In the liver, PPARα regulates lipid metabolism and controls liver homeostasis, and dysregulation and overexpression of PPARα may lead to hepatic steatosis, steatohepatitis, steatofibrosis, and liver cancer [53]. An earlier study suggests that miR-483-3p inhibited adipocyte differentiation by reducing the expression of PPARγ2 and FABP4, and miR-483-3p antagonist treatment (9 days) increased the expression of PPARγ2 and FABP4, indicating that miR-483 may dysregulate PPARs expression [74]. Our data also supports the observation that miR-483 inhibits PPARa expression and downregulates cell steatosis. MiR-483 is known to target tissue inhibitors of metalloproteinases 2 (TIMP2), and a recent study suggests that inhibition of miR-483-5p by intra-articular injection of antago-miR-483-5p could prevent the onset of osteoarthritis (OA) pathogenesis by targeting Matrilin 3 (Matn3) and TIMP2 [75]. Bone marrow mesenchymal stem cells derived exosomal miR-483 increased multiple myeloma’s malignant progression by inhibiting TIMP2 expression [76], and the inhibition of miR483 increased expression of p21 and downregulated the expression of c-Myc and Bcl-2 [76]. Li et al. demonstrated that miR-483 suppresses CCl4-mediated mouse liver fibrosis in vivo by targeting TIMP2 and PDGF-β [35]. Our RT/qPCR, immunoblotting, and reporter assay data suggest that miR-483 targets TIMP2 expression and suppresses fibrogenic signaling in HCC cells. Our data also suggest that miR-483 potentially suppressed steatotic and fibrogenic response by targeting PPARA, TIMP2, TGFB1, and p21 since miR-483 binds with the UTRs of these genes. We further established the association between miR-483 expression and NAFLD and AFLD progression in vivo mice models (Figure 7E). Although an earlier study demonstrated that miR-483-5p targets PCSK9, increases hepatic LDL Receptor expression, and ameliorates hypercholesterolemia in mice liver [34], our data suggest that mice fed with a high-fat diet and EtOH show downregulation of miR-483 expression compared with mice fed a regular diet, similar to our HCC tissue samples. Downregulation of miR-483 in the NAFLD mouse model increased TIMP2 and TGF-β expression, whereas down-regulation of miR-483 in the AFLD mouse model increased TGF-β expression but not TIMP2, indicating that miR-483/TIMP2 axis differentially regulated in fatty liver disease. ## 5. Conclusions Our data suggest that the downregulation of miR-483 in NAFLD/AFLD and HCC modulates the progression of fatty liver diseases and HCC, and overexpression of miR-483 inhibited cell steatosis, fibrosis, and HCC cell proliferation; therefore, miR-483 may be used as a novel therapeutic target to treat patients with fatty liver diseases/HCC. ## References 1. Marrero J.A., Kulik L.M., Sirlin C.B., Zhu A.X., Finn R.S., Abecassis M.M., Roberts L.R., Heimbach J.K.. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology (2018) 68 723-750. DOI: 10.1002/hep.29913 2. Siegel R.L., Miller K.D., Fuchs H.E., Jemal A.. Cancer statistics, 2022. CA Cancer J. Clin. (2022) 72 7-33. DOI: 10.3322/caac.21708 3. Sanyal A.J., Yoon S.K., Lencioni R.. The etiology of hepatocellular carcinoma and consequences for treatment. Oncologist (2010) 15 14-22. DOI: 10.1634/theoncologist.2010-S4-14 4. Balogh J., Victor D., Asham E.H., Burroughs S.G., Boktour M., Saharia A., Li X., Ghobrial R.M., Monsour H.P.. Hepatocellular carcinoma: A review. J. Hepatocell. Carcinoma (2016) 3 41-53. DOI: 10.2147/JHC.S61146 5. Friedman S.L., Neuschwander-Tetri B.A., Rinella M., Sanyal A.J.. Mechanisms of NAFLD development and therapeutic strategies. Nat. Med. (2018) 24 908-922. DOI: 10.1038/s41591-018-0104-9 6. Wegermann K., Suzuki A., Mavis A.M., Abdelmalek M.F., Diehl A.M., Moylan C.A.. Tackling NAFLD: Three Targeted Populations. Hepatology (2021) 73 1199-1206. DOI: 10.1002/hep.31533 7. Paschos P., Paletas K.. Non alcoholic fatty liver disease and metabolic syndrome. Hippokratia (2009) 13 9-19. PMID: 19240815 8. Rinella M.E.. Nonalcoholic fatty liver disease: A systematic review. JAMA (2015) 313 2263-2273. DOI: 10.1001/jama.2015.5370 9. White D.L., Kanwal F., El-Serag H.B.. Association between nonalcoholic fatty liver disease and risk for hepatocellular cancer, based on systematic review. Clin. Gastroenterol. Hepatol. (2012) 10 1342-1359.e1342. DOI: 10.1016/j.cgh.2012.10.001 10. Michelotti G.A., Machado M.V., Diehl A.M.. NAFLD, NASH and liver cancer. Nat. Rev. Gastroenterol. Hepatol. (2013) 10 656-665. DOI: 10.1038/nrgastro.2013.183 11. Cohen J.C., Horton J.D., Hobbs H.H.. Human Fatty Liver Disease: Old Questions and New Insights. Science (2011) 332 1519-1523. DOI: 10.1126/science.1204265 12. Dang K., Hirode G., Singal A., Sundaram V., Wong R.J.. Alcoholic Liver Disease Epidemiology in the United States: A Retrospective Analysis of 3 US Databases. Am. J. Gastroenterol. (2020) 115 96-104. DOI: 10.14309/ajg.0000000000000380 13. Pai R.K., Kleiner D.E., Hart J., Adeyi O.A., Clouston A.D., Behling C.A., Jain D., Kakar S., Brahmania M., Burgart L.. Standardising the interpretation of liver biopsies in non-alcoholic fatty liver disease clinical trials. Aliment Pharmacol. Ther. (2019) 50 1100-1111. DOI: 10.1111/apt.15503 14. Takakura K., Oikawa T., Nakano M., Saeki C., Torisu Y., Kajihara M., Saruta M.. Recent Insights Into the Multiple Pathways Driving Non-alcoholic Steatohepatitis-Derived Hepatocellular Carcinoma. Front. Oncol. (2019) 9 762. DOI: 10.3389/fonc.2019.00762 15. Younossi Z.M., Koenig A.B., Abdelatif D., Fazel Y., Henry L., Wymer M.. Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology (2016) 64 73-84. DOI: 10.1002/hep.28431 16. Neureiter D., Stintzing S., Kiesslich T., Ocker M.. Hepatocellular carcinoma: Therapeutic advances in signaling, epigenetic and immune targets. World J. Gastroenterol. (2019) 25 3136-3150. DOI: 10.3748/wjg.v25.i25.3136 17. Wang W., Wei C.. Advances in the early diagnosis of hepatocellular carcinoma. Genes Dis. (2020) 7 308-319. DOI: 10.1016/j.gendis.2020.01.014 18. Bartel D.P.. MicroRNAs: Genomics, biogenesis, mechanism, and function. Cell (2004) 116 281-297. DOI: 10.1016/S0092-8674(04)00045-5 19. Karp X., Ambros V.. Developmental biology. Encountering microRNAs in cell fate signaling. Science (2005) 310 1288-1289. DOI: 10.1126/science.1121566 20. Cheng A.M., Byrom M.W., Shelton J., Ford L.P.. Antisense inhibition of human miRNAs and indications for an involvement of miRNA in cell growth and apoptosis. Nucleic Acids Res. (2005) 33 1290-1297. DOI: 10.1093/nar/gki200 21. Lu S., Sun Y.H., Shi R., Clark C., Li L., Chiang V.L.. Novel and mechanical stress-responsive MicroRNAs in Populus trichocarpa that are absent from Arabidopsis. Plant Cell (2005) 17 2186-2203. DOI: 10.1105/tpc.105.033456 22. Murray M.J., Huddart R.A., Coleman N.. The present and future of serum diagnostic tests for testicular germ cell tumours. Nat. Rev. Urol. (2016) 13 715-725. DOI: 10.1038/nrurol.2016.170 23. Shen S., Lin Y., Yuan X., Shen L., Chen J., Chen L., Qin L., Shen B.. Biomarker MicroRNAs for Diagnosis, Prognosis and Treatment of Hepatocellular Carcinoma: A Functional Survey and Comparison. Sci. Rep. (2016) 6 38311. DOI: 10.1038/srep38311 24. Sierzega M., Kaczor M., Kolodziejczyk P., Kulig J., Sanak M., Richter P.. Evaluation of serum microRNA biomarkers for gastric cancer based on blood and tissue pools profiling: The importance of miR-21 and miR-331. Br. J. Cancer (2017) 117 266-273. DOI: 10.1038/bjc.2017.190 25. Niture S., Tricoli L., Qi Q., Gadi S., Hayes K., Kumar D.. MicroRNA-99b-5p targets mTOR/AR axis, induces autophagy and inhibits prostate cancer cell proliferation. Tumour. Biol. (2022) 44 107-126. DOI: 10.3233/TUB-211568 26. Rios-Colon L., Chijioke J., Niture S., Afzal Z., Qi Q., Srivastava A., Ramalinga M., Kedir H., Cagle P., Arthur E.. Leptin modulated microRNA-628-5p targets Jagged-1 and inhibits prostate cancer hallmarks. Sci. Rep. (2022) 12 10073. DOI: 10.1038/s41598-022-13279-x 27. An Y., Gao S., Zhao W.C., Qiu B.A., Xia N.X., Zhang P.J., Fan Z.P.. Novel serum microRNAs panel on the diagnostic and prognostic implications of hepatocellular carcinoma. World J. Gastroenterol. (2018) 24 2596-2604. DOI: 10.3748/wjg.v24.i24.2596 28. Peng Y., Croce C.M.. The role of MicroRNAs in human cancer. Signal Transduct. Target. Ther. (2016) 1 15004. DOI: 10.1038/sigtrans.2015.4 29. Veronese A., Lupini L., Consiglio J., Visone R., Ferracin M., Fornari F., Zanesi N., Alder H., D’Elia G., Gramantieri L.. Oncogenic role of miR-483-3p at the IGF2/483 locus. Cancer Res. (2010) 70 3140-3149. DOI: 10.1158/0008-5472.CAN-09-4456 30. Olson P., Lu J., Zhang H., Shai A., Chun M.G., Wang Y., Libutti S.K., Nakakura E.K., Golub T.R., Hanahan D.. MicroRNA dynamics in the stages of tumorigenesis correlate with hallmark capabilities of cancer. Genes Dev. (2009) 23 2152-2165. DOI: 10.1101/gad.1820109 31. Soon P.S., Tacon L.J., Gill A.J., Bambach C.P., Sywak M.S., Campbell P.R., Yeh M.W., Wong S.G., Clifton-Bligh R.J., Robinson B.G.. miR-195 and miR-483-5p Identified as Predictors of Poor Prognosis in Adrenocortical Cancer. Clin. Cancer Res. (2009) 15 7684-7692. DOI: 10.1158/1078-0432.CCR-09-1587 32. Ma N., Li F., Li D., Hui Y., Wang X., Qiao Y., Zhang Y., Xiang Y., Zhou J., Zhou L.. Igf2-derived intronic miR-483 promotes mouse hepatocellular carcinoma cell proliferation. Mol. Cell. Biochem. (2012) 361 337-343. DOI: 10.1007/s11010-011-1121-x 33. Tian T., Lv X., Pan G., Lu Y., Chen W., He W., Lei X., Zhang H., Liu M., Sun S.. Long Noncoding RNA MPRL Promotes Mitochondrial Fission and Cisplatin Chemosensitivity via Disruption of Pre-miRNA Processing. Clin. Cancer Res. (2019) 25 3673-3688. DOI: 10.1158/1078-0432.CCR-18-2739 34. Dong J., He M., Li J., Pessentheiner A., Wang C., Zhang J., Sun Y., Wang W.T., Zhang Y., Liu J.. microRNA-483 ameliorates hypercholesterolemia by inhibiting PCSK9 production. JCI Insight (2020) 5 e143812. DOI: 10.1172/jci.insight.143812 35. Li F., Ma N., Zhao R., Wu G., Zhang Y., Qiao Y., Han D., Xu Y., Xiang Y., Yan B.. Overexpression of miR-483-5p/3p cooperate to inhibit mouse liver fibrosis by suppressing the TGF-beta stimulated HSCs in transgenic mice. J. Cell Mol. Med. (2014) 18 966-974. DOI: 10.1111/jcmm.12293 36. Varghese R.S., Zhou Y., Barefoot M., Chen Y., Di Poto C., Balla A.K., Oliver E., Sherif Z.A., Kumar D., Kroemer A.H.. Identification of miRNA-mRNA associations in hepatocellular carcinoma using hierarchical integrative model. BMC Med. Genom. (2020) 13. DOI: 10.1186/s12920-020-0706-1 37. Varghese R.S., Barefoot M.E., Jain S., Chen Y., Zhang Y., Alley A., Kroemer A.H., Tadesse M.G., Kumar D., Sherif Z.A.. Integrative Analysis of DNA Methylation and microRNA Expression Reveals Mechanisms of Racial Heterogeneity in Hepatocellular Carcinoma. Front. Genet. (2021) 12 708326. DOI: 10.3389/fgene.2021.708326 38. Marion M.J., Hantz O., Durantel D.. The HepaRG cell line: Biological properties and relevance as a tool for cell biology, drug metabolism, and virology studies. Methods Mol. Biol. (2010) 640 261-272. DOI: 10.1007/978-1-60761-688-7_13 39. Niture S., Lin M., Odera J.O., Moore J., Zhe H., Chen X., Suy S., Collins S.P., Kumar D.. TNFAIP8 drives metabolic reprogramming to promote prostate cancer cell proliferation. Int. J. Biochem. Cell Biol. (2021) 130 105885. DOI: 10.1016/j.biocel.2020.105885 40. Cagle P., Niture S., Srivastava A., Ramalinga M., Aqeel R., Rios-Colon L., Chimeh U., Suy S., Collins S.P., Dahiya R.. MicroRNA-214 targets PTK6 to inhibit tumorigenic potential and increase drug sensitivity of prostate cancer cells. Sci. Rep. (2019) 9 9776. DOI: 10.1038/s41598-019-46170-3 41. Choi S., Neequaye P., French S.W., Gonzalez F.J., Gyamfi M.A.. Pregnane X receptor promotes ethanol-induced hepatosteatosis in mice. J. Biol. Chem. (2018) 293 1-17. DOI: 10.1074/jbc.M117.815217 42. Niture S., Gyamfi M.A., Lin M., Chimeh U., Dong X., Zheng W., Moore J., Kumar D.. TNFAIP8 regulates autophagy, cell steatosis, and promotes hepatocellular carcinoma cell proliferation. Cell Death Dis. (2020) 11 178. DOI: 10.1038/s41419-020-2369-4 43. Hu L., Xue F., Shao M., Deng A., Wei G.. Aberrant expression of Notch3 predicts poor survival for hepatocellular carcinomas. Biosci. Trends (2013) 7 152-156. PMID: 23836039 44. Gramantieri L., Giovannini C., Lanzi A., Chieco P., Ravaioli M., Venturi A., Grazi G.L., Bolondi L.. Aberrant Notch3 and Notch4 expression in human hepatocellular carcinoma. Liver Int. (2007) 27 997-1007. DOI: 10.1111/j.1478-3231.2007.01544.x 45. Huang Q., Li J., Zheng J., Wei A.. The Carcinogenic Role of the Notch Signaling Pathway in the Development of Hepatocellular Carcinoma. J. Cancer (2019) 10 1570-1579. DOI: 10.7150/jca.26847 46. Zhang Y., Li D., Feng F., An L., Hui F., Dang D., Zhao Q.. Progressive and Prognosis Value of Notch Receptors and Ligands in Hepatocellular Carcinoma: A Systematic Review and Meta-analysis. Sci. Rep. (2017) 7 14809. DOI: 10.1038/s41598-017-14897-6 47. Ribatti D., Tamma R., Annese T.. Epithelial-Mesenchymal Transition in Cancer: A Historical Overview. Transl. Oncol. (2020) 13 100773. DOI: 10.1016/j.tranon.2020.100773 48. Van Zijl F., Zulehner G., Petz M., Schneller D., Kornauth C., Hau M., Machat G., Grubinger M., Huber H., Mikulits W.. Epithelial-mesenchymal transition in hepatocellular carcinoma. Future Oncol. (2009) 5 1169-1179. DOI: 10.2217/fon.09.91 49. Pinyopornpanish K., Khoudari G., Saleh M.A., Angkurawaranon C., Pinyopornpanish K., Mansoor E., Dasarathy S., McCullough A.. Hepatocellular carcinoma in nonalcoholic fatty liver disease with or without cirrhosis: A population-based study. BMC Gastroenterol. (2021) 21. DOI: 10.1186/s12876-021-01978-0 50. Singh R., Kaushik S., Wang Y., Xiang Y., Novak I., Komatsu M., Tanaka K., Cuervo A.M., Czaja M.J.. Autophagy regulates lipid metabolism. Nature (2009) 458 1131-1135. DOI: 10.1038/nature07976 51. Zhang J., He Y., Yan X., Chen S., He M., Lei Y., Zhang J., Gongol B., Gu M., Miao Y.. MicroRNA-483 amelioration of experimental pulmonary hypertension. EMBO Mol. Med. (2020) 12 e11303. DOI: 10.15252/emmm.201911303 52. Liss K.H., Finck B.N.. PPARs and nonalcoholic fatty liver disease. Biochimie (2017) 136 65-74. DOI: 10.1016/j.biochi.2016.11.009 53. Wang Y., Nakajima T., Gonzalez F.J., Tanaka N.. PPARs as Metabolic Regulators in the Liver: Lessons from Liver-Specific PPAR-Null Mice. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21062061 54. Condrat C.E., Thompson D.C., Barbu M.G., Bugnar O.L., Boboc A., Cretoiu D., Suciu N., Cretoiu S.M., Voinea S.C.. miRNAs as Biomarkers in Disease: Latest Findings Regarding Their Role in Diagnosis and Prognosis. Cells (2020) 9. DOI: 10.3390/cells9020276 55. Wei Y., Zhang Q., An L., Fang G., Hong D., Jiao T., Yang H., Wang Z.. Serum exosomal microRNA-370-3p and microRNA-196a-5p are potential biomarkers for the diagnosis and prognosis of hepatocellular carcinoma. Folia Histochem. Cytobiol. (2022) 60 215-225. DOI: 10.5603/FHC.a2022.0019 56. Fang Y., Yan D., Wang L., Zhang J., He Q.. Circulating microRNAs (miR-16, miR-22, miR-122) expression and early diagnosis of hepatocellular carcinoma. J. Clin. Lab. Anal. (2022) 36 e24541. DOI: 10.1002/jcla.24541 57. Khairy A., Ibrahim M.K., AbdElrahman M., Fouad R., Zayed N., Ayman Y., Abdellatef Z., Yosry A.. The diagnostic utility of microRNA 222-3p, microRNA 21-5p, and microRNA 122-5p for HCV-related hepatocellular carcinoma and its relation to direct-acting antiviral therapy. Arab. J. Gastroenterol. (2022) 23 108-114. DOI: 10.1016/j.ajg.2022.04.001 58. Villanueva A., Alsinet C., Yanger K., Hoshida Y., Zong Y., Toffanin S., Rodriguez-Carunchio L., Sole M., Thung S., Stanger B.Z.. Notch signaling is activated in human hepatocellular carcinoma and induces tumor formation in mice. Gastroenterology (2012) 143 1660-1669.e1667. DOI: 10.1053/j.gastro.2012.09.002 59. Fu Z.W., Wang L., Li S.J., Chen F., Au-Yeung K.K.W., Shi C.. MicroRNA as an Important Target for Anticancer Drug Development. Front. Pharmacol. (2021) 12 736323. DOI: 10.3389/fphar.2021.736323 60. Yahya S.M.M., Fathy S.A., El-Khayat Z.A., El-Toukhy S.E., Hamed A.R., Hegazy M.G.A., Nabih H.K.. Possible Role of microRNA-122 in Modulating Multidrug Resistance of Hepatocellular Carcinoma. Indian J. Clin. Biochem. (2018) 33 21-30. DOI: 10.1007/s12291-017-0651-8 61. Shi L., Chen Z.G., Wu L.L., Zheng J.J., Yang J.R., Chen X.F., Chen Z.Q., Liu C.L., Chi S.Y., Zheng J.Y.. miR-340 reverses cisplatin resistance of hepatocellular carcinoma cell lines by targeting Nrf2-dependent antioxidant pathway. Asian Pac. J. Cancer Prev. (2014) 15 10439-10444. DOI: 10.7314/APJCP.2014.15.23.10439 62. López-Riera M., Conde I., Quintas G., Pedrola L., Zaragoza Á., Perez-Rojas J., Salcedo M., Benlloch S., Castell J.V., Jover R.. Non-invasive prediction of NAFLD severity: A comprehensive, independent validation of previously postulated serum microRNA biomarkers. Sci. Rep. (2018) 8 10606. DOI: 10.1038/s41598-018-28854-4 63. Xiang Y., Mao L., Zuo M.-L., Song G.-L., Tan L.-M., Yang Z.-B.. The Role of MicroRNAs in Hyperlipidemia: From Pathogenesis to Therapeutical Application. Mediat. Inflamm. (2022) 2022 3101900. DOI: 10.1155/2022/3101900 64. Naeli P., Mirzadeh Azad F., Malakootian M., Seidah N.G., Mowla S.J.. Post-transcriptional Regulation of PCSK9 by miR-191, miR-222, and miR-224. Front. Genet. (2017) 8 189. DOI: 10.3389/fgene.2017.00189 65. Xu X., Dong Y., Ma N., Kong W., Yu C., Gong L., Chen J., Ren J.. MiR-337-3p lowers serum LDL-C level through targeting PCSK9 in hyperlipidemic mice. Metabolism (2021) 119 154768. DOI: 10.1016/j.metabol.2021.154768 66. Yorimitsu T., Klionsky D.J.. Autophagy: Molecular machinery for self-eating. Cell Death Differ. (2005) 12 1542-1552. DOI: 10.1038/sj.cdd.4401765 67. Wang Y., Liu Z., Zou W., Hong H., Fang H., Tong W.. Molecular regulation of miRNAs and potential biomarkers in the progression of hepatic steatosis to NASH. Biomark. Med. (2015) 9 1189-1200. DOI: 10.2217/bmm.15.70 68. Leti F., Malenica I., Doshi M., Courtright A., Van Keuren-Jensen K., Legendre C., Still C.D., Gerhard G.S., DiStefano J.K.. High-throughput sequencing reveals altered expression of hepatic microRNAs in nonalcoholic fatty liver disease-related fibrosis. Transl. Res. (2015) 166 304-314. DOI: 10.1016/j.trsl.2015.04.014 69. Fang Z., Dou G., Wang L.. MicroRNAs in the Pathogenesis of Nonalcoholic Fatty Liver Disease. Int. J. Biol. Sci. (2021) 17 1851-1863. DOI: 10.7150/ijbs.59588 70. Murakami Y., Toyoda H., Tanaka M., Kuroda M., Harada Y., Matsuda F., Tajima A., Kosaka N., Ochiya T., Shimotohno K.. The progression of liver fibrosis is related with overexpression of the miR-199 and 200 families. PLoS ONE (2011) 6. DOI: 10.1371/journal.pone.0016081 71. Hou J., Lin L., Zhou W., Wang Z., Ding G., Dong Q., Qin L., Wu X., Zheng Y., Yang Y.. Identification of miRNomes in human liver and hepatocellular carcinoma reveals miR-199a/b-3p as therapeutic target for hepatocellular carcinoma. Cancer Cell (2011) 19 232-243. DOI: 10.1016/j.ccr.2011.01.001 72. Cabral B.C.A., Hoffmann L., Bottaro T., Costa P.F., Ramos A.L.A., Coelho H.S.M., Villela-Nogueira C.A., Urmenyi T.P., Faffe D.S., Silva R.. Circulating microRNAs associated with liver fibrosis in chronic hepatitis C patients. Biochem. Biophys. Rep. (2020) 24 100814. DOI: 10.1016/j.bbrep.2020.100814 73. Gervois P., Torra I.P., Fruchart J.C., Staels B.. Regulation of lipid and lipoprotein metabolism by PPAR activators. Clin. Chem. Lab. Med. (2000) 38 3-11. DOI: 10.1515/CCLM.2000.002 74. Ferland-McCollough D., Fernandez-Twinn D.S., Cannell I.G., David H., Warner M., Vaag A.A., Bork-Jensen J., Brons C., Gant T.W., Willis A.E.. Programming of adipose tissue miR-483-3p and GDF-3 expression by maternal diet in type 2 diabetes. Cell Death Differ. (2012) 19 1003-1012. DOI: 10.1038/cdd.2011.183 75. Wang H., Zhang H., Sun Q., Wang Y., Yang J., Yang J., Zhang T., Luo S., Wang L., Jiang Y.. Intra-articular Delivery of Antago-miR-483-5p Inhibits Osteoarthritis by Modulating Matrilin 3 and Tissue Inhibitor of Metalloproteinase 2. Mol. Ther. (2017) 25 715-727. DOI: 10.1016/j.ymthe.2016.12.020 76. Gu J., Wang M., Wang X., Li J., Liu H., Lin Z., Yang X., Zhang X., Liu H.. Exosomal miR-483-5p in Bone Marrow Mesenchymal Stem Cells Promotes Malignant Progression of Multiple Myeloma by Targeting TIMP2. Front. Cell Dev. Biol. (2022) 10 862524. DOI: 10.3389/fcell.2022.862524
--- title: Energy Metabolism, Metabolite, and Inflammatory Profiles in Human Ex Vivo Adipose Tissue Are Influenced by Obesity Status, Metabolic Dysfunction, and Treatment Regimes in Patients with Oesophageal Adenocarcinoma authors: - Fiona O’Connell - Eimear Mylod - Noel E. Donlon - Aisling B. Heeran - Christine Butler - Anshul Bhardwaj - Sinead Ramjit - Michael Durand - Gerard Lambe - Paul Tansey - Ivan Welartne - Kevin P. Sheahan - Xiaofei Yin - Claire L. Donohoe - Narayanasamy Ravi - Margaret R. Dunne - Lorraine Brennan - John V. Reynolds - Helen M. Roche - Jacintha O’Sullivan journal: Cancers year: 2023 pmcid: PMC10046380 doi: 10.3390/cancers15061681 license: CC BY 4.0 --- # Energy Metabolism, Metabolite, and Inflammatory Profiles in Human Ex Vivo Adipose Tissue Are Influenced by Obesity Status, Metabolic Dysfunction, and Treatment Regimes in Patients with Oesophageal Adenocarcinoma ## Abstract ### Simple Summary Oesophageal adenocarcinoma (OAC) is a poor prognosis cancer with limited response rates to current treatment modalities and is strongly linked to obesity status and metabolic dysfunction. This study for the first time has conducted a detailed assessment of adipose tissue metabolism, as well as assessing adipose secreted pro-inflammatory, metabolite, and lipid profiles to determine whether these profiles correlate with significant clinical parameters in OAC patients including obesity, metabolic dysfunction, previous treatment exposure, and tumour regression grades. Overall, in this study, increases in metabolic profiles linked with oxidative phosphorylation, pro-inflammatory cytokines, and metabolites associated with aiding tumorigenesis have been identified in the most viscerally obese OAC patients and in patients with metabolic dysfunction. These results raise the question of whether targeting these altered signalling mechanisms could aid current treatment strategies. ### Abstract Oesophageal adenocarcinoma (OAC) is a poor prognosis cancer with limited response rates to current treatment modalities and has a strong link to obesity. To better elucidate the role of visceral adiposity in this disease state, a full metabolic profile combined with analysis of secreted pro-inflammatory cytokines, metabolites, and lipid profiles were assessed in human ex vivo adipose tissue explants from obese and non-obese OAC patients. These data were then related to extensive clinical data including obesity status, metabolic dysfunction, previous treatment exposure, and tumour regression grades. Real-time energy metabolism profiles were assessed using the seahorse technology. Adipose explant conditioned media was screened using multiplex ELISA to assess secreted levels of 54 pro-inflammatory mediators. Targeted secreted metabolite and lipid profiles were analysed using Ultra-High-Performance Liquid Chromatography coupled with Mass Spectrometry. Adipose tissue explants and matched clinical data were collected from OAC patients ($$n = 32$$). Compared to visceral fat from non-obese patients ($$n = 16$$), visceral fat explants from obese OAC patients ($$n = 16$$) had significantly elevated oxidative phosphorylation metabolism profiles and an increase in Eotaxin-3, IL-17A, IL-17D, IL-3, MCP-1, and MDC and altered secretions of glutamine associated metabolites. Adipose explants from patients with metabolic dysfunction correlated with increased oxidative phosphorylation metabolism, and increases in IL-5, IL-7, SAA, VEGF-C, triacylglycerides, and metabolites compared with metabolically healthy patients. Adipose explants generated from patients who had previously received neo-adjuvant chemotherapy ($$n = 14$$) showed elevated secretions of pro-inflammatory mediators, IL-12p40, IL-1α, IL-22, and TNF-β and a decreased expression of triacylglycerides. Furthermore, decreased secreted levels of triacylglycerides were also observed in the adipose secretome of patients who received the chemotherapy-only regimen FLOT compared with patients who received no neo-adjuvant treatment or chemo-radiotherapy regimen CROSS. For those patients who showed the poorest response to currently available treatments, their adipose tissue was associated with higher glycolytic metabolism compared to patients who had good treatment responses. This study demonstrates that the adipose secretome in OAC patients is enriched with mediators that could prime the tumour microenvironment to aid tumour progression and attenuate responses to conventional cancer treatments, an effect which appears to be augmented by obesity and metabolic dysfunction and exposure to different treatment regimes. ## 1. Introduction Oesophageal adenocarcinoma (OAC) is an aggressive disease associated with a poor prognosis and a five-year survival rate of approximately $20\%$ [1], with current projections indicating that the incidence of this disease is increasing [2]. Currently, the standard of care for treatment involves neo-adjuvant treatment (treatment prior to surgery) with either chemotherapy alone including the FLOT regimen (consisting of 5FU, Folinic acid, Oxaliplatin, Docetaxel) or combination chemo-radiotherapy such as the CROSS regimen (consisting of Carboplatin, paclitaxel with concurrent 41.4 Gy radiation), for locally advanced tumours [3]. Unfortunately, only approximately $30\%$ of patients show a complete response to these current treatment modalities, leaving a large proportion of patients with no therapeutic gain and a possible delay to surgery [4,5]. Furthermore, large-scale epidemiological studies demonstrate a consistent and compelling association between the risk of cancer development/progression and elevated body mass index (BMI) for many gastrointestinal cancers including OAC, with this cancer having one of the strongest associations with obesity [6,7,8]. This makes it an exemplary model for studying the influence of obesity on cancer and the role of the adipose tissue microenvironment in this setting. Numerous factors are associated with the obese adipose tissue microenvironment such as chronic low-grade inflammation, angiogenesis, fibrosis, and the altered secretions of cells, all implicated in the progression and recurrence of cancer [9,10]. One of the significant effects of the obese adipocyte secretome is the release of a series of pro-inflammatory factors, leading to a local environment that is primed to aid tumour development and progression [11]. The altered milieu of the obese tumour environment has been extensively shown to have detrimental effects on the anti-tumour response, diminishing immune cell function and treatment efficacy [12,13,14]. It has been previously shown that adipose-conditioned media from oesophageal cancer patients increases radiosensitivity [15], and this could be linked to the differential expression of leptin receptors and its associated biology in driving inflammation in the adipose tissue microenvironment [16]. Currently, discrepancies are reported in the literature on whether obesity diminishes [17] or ameliorates [15] treatment resistance and whether obese individuals possess an enhanced survival benefit compared with their non-obese counterparts [18], which highlights the importance of identifying the underlying biological mechanisms which play a role in this setting in the complex adipose tissue microenvironment. Previous work has observed elevated oxidative phosphorylation in visceral adipose tissue compared with subcutaneous adipose tissue [16], and adipocytes derived from metabolically unhealthy obese individuals show elevated mitochondrial response profiles [19]. Whilst the role of tumour explant energy metabolism in OAC treatment response has been reported [20], the energy metabolism profiles of visceral adipose tissue between obese and non-obese OAC patients and the influence of its secretome on the cancer cell and immune cell function is still largely unknown. For the first time, this study aims to better characterise the adipose tissue microenvironment using real-time energy metabolism profiles, profiling the secreted inflammatory environment, and assessing altered metabolites and lipid profiles of human ex vivo adipose explants. These profiles were examined based on obesity status, metabolic dysfunction, previous treatment exposure, and treatment responses in OAC patients. Overall, the profiling data and clinical correlations described in this study suggest the adipose microenvironment as well as potentiating a pro-tumorigenic milieu may also be linked with the efficacy of current standard-of-care cancer treatments. With the knowledge that the obesity epidemic is projected to increase, with $50\%$ of the Western world population being obese by 2030 [21], this research endeavours to address an exigent question: what influence might the adipose microenvironment possess in the cancer–obesity link? ## 2.1. Ethics Statement and Patient Recruitment Ethical approval was granted by the St James’s Hospital/AMNCH ethical review board (Ethics number: REC_2019-07 List 25[27]), and written informed consent was collected from all patients in this study. Thirty-two patients were recruited within the period between 1 December 2019 and 30 January 2022, and the patient demographics are listed in Table 1. All fresh adipose tissues were taken at the start of the surgical tumour resection procedure from OAC or OGJ (oesophagogastric junctional adenocarcinoma) patients being treated with curative intent. ## 2.2. Clinical Data Collation and Assessment Obesity was defined using visceral fat area (VFA) measurements with a cut-off value for VFA of 163.8 cm2 for males and 80.1 cm2 for females as previously categorised [22]. Metabolic dysfunction was defined if a patient had 3 or more of the following criteria: visceral obesity (as assessed with the VFA cut-offs mentioned above), previously diagnosed type 2 diabetes or impaired fasting glucose, triglycerides ≥ 1.7 mmol/L or interventional treatment for high triglycerides; high-density lipoprotein cholesterol or interventional treatment for low HDL, systolic blood pressure ≥ 130 mmHg and/or diastolic blood pressure ≥ 85 mmHg or treatment for hypertension [22,23]. Previous treatment was classified as a patient receiving either neo-adjuvant chemotherapy only (FLOT regimen) or chemo-radiotherapy (CROSS regimen), patients who received no neo-adjuvant treatment prior to surgery were classified as treatment naïve. Histological assessment of resected tumours was conducted by a pathologist, using the Mandard tumour regression grade to assess patient’s response to neo-adjuvant treatment; therefore, no TRG scoring is available for treatment naïve patients who did not receive chemotherapy or chemo-radiotherapy. The clinical data summary is listed in Table 1. ## 2.3. Seahorse Analysis of Metabolic Profiles from Adipose Tissue Explants and Generation of Adipose Conditioned Media (ACM) Fresh omental tissue was collected from theatre and processed within 30 min by dissecting it into pieces weighing approximately 20 mg. Tissue was plated in triplicates in 1 mL of M199 (Gibco, Thermofisher, Waltham, MA, USA) supplemented with $0.1\%$ gentamicin (Lonza, Switzerland), in a 24-well plate (Sarstedt, Nümbrecht, Germany). Adipose explants were cultured for 24 h at 37 °C and $5\%$ carbon dioxide in a humidified incubator (Thermofisher, MA, USA). In the last hour of culture, adipose tissue and ACM were transferred to an islet capture microplate with capture screens (Agilent Technologies, Santa Clara, CA, USA) and incubated in a non-CO2 incubator at 37 °C (Whitley, West Yorkshire, UK) prior to analysis. Seahorse Xfe24 analyser was used to assess metabolic profiles in adipose explants (Agilent Technologies, CA, USA). Following a 12 min equilibrate step, three basal measurements of OCR (Oxygen Consumption Rate) and ECAR (Extracellular Acidification Rate) were taken over 24 min consisting of three repeats of the following sequence “mix (3 min)/wait (2 min)/measurement (3 min)” to establish basal respiration. Adipose Conditioned Media (ACM) was extracted in a sterile environment and tissue was weighed using a benchtop analytical balance (Radwag, Radom, Poland) and snap frozen. All samples were then stored at −80 °C for further processing. ## 2.4. Multiplex ELISA The collected Adipose Conditioned Media (ACM) was processed according to MSD (Meso Scale Discovery, Rockville, Maryland, USA) multiplex protocol. To assess angiogenic, vascular injury, pro-inflammatory, and cytokine and chemokine secretions from ACM, a 54-plex ELISA kit separated across 7 plates was used (Meso Scale Discovery, Rockville, Maryland, USA). The multiplex kit was used to quantify the secretions of CRP, Eotaxin, Eotaxin-3, FGF(basic), Flt-1, GM-CSF, ICAM-1, IFN-γ, IL-10, IL-12/IL-23p40, IL-12p70, IL-13, IL-15, IL-16, IL-17A, IL-17A/F, IL-17B, IL-17C, IL-17D, IL-1RA, IL-1α, IL-1β, IL-2, IL-21, IL-22, IL-23, IL-27, IL-3, IL-31, IL-4, IL-5, IL-6, IL-7, IL-8, IL-8 (HA), IL-9, IP-10, MCP-1, MCP-4, MDC, MIP-1α, MIP-1β, MIP-3α, PlGF, SAA, TARC, Tie-2, TNF-α, TNF-β, TSLP, VCAM-1, VEGF-A, VEGF-C, and VEGF-D from ACM. All assays were run as per the manufacturer’s recommendation, and an overnight supernatant incubation protocol was used for all assays except Angiogenesis Panel 1 and Vascular Injury Panel 2, which were run according to the same-day protocol. ACM was run undiluted for all assays except Vascular Injury Panel 2, where a one-in-four dilution was used, as per previous optimization experiments. Assays were run on a MESO QuickPlex SQ 120, and all analyte concentrations were calculated using Discovery Workbench software (version 4.0). Secretion data for all factors were normalized to adipose post-incubation weight and expressed as pg/mL per gram of adipose tissue. ## 2.5. Metabolomic and Lipidomic Screening ACM was analysed using a targeted metabolomic platform and was prepared according to the MxP® Quant 500 assay manual (Biocrates Life Sciences, Innsbruck, Austria). Samples were dried and derivatised using derivatization solution ($5\%$ phenyl isothiocyanate in ethanol/water/pyridine (volume ratio $\frac{1}{1}$/1)) and incubated for 1 h at room temperature and then dried under nitrogen for 1 h. After the addition of 300 μL of 5 mM ammonium acetate in methanol, the plate was shaken for 30 min and then centrifuged at 500 g for 2 min, and then 150 μL of high-performance liquid chromatography (HPLC)-grade water was added for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Additionally, 10 μL of the eluate was diluted with 490 μL of methanol running solvent for flow injection analysis tandem mass spectrometry (FIA-MS/MS) analysis. Samples were analysed using the Sciex ExionLC series UHPLC system coupled to a Sciex QTRAP 6500+ mass spectrometer. The UHPLC columns (Biocrates Life Sciences, Innsbruck, Austria) were installed, and mobile phases A and B were $100\%$ water and $95\%$ acetonitrile (both added $0.2\%$ formic acid), respectively. In the LC-MS/MS analysis, amino acids ($$n = 20$$) and amino acid-related ($$n = 30$$), bile acids ($$n = 14$$), biogenic amines ($$n = 9$$), carboxylic acids [7], hormones and related ($$n = 4$$), indoles and derivatives ($$n = 4$$), nucleobases and related ($$n = 2$$), fatty acids ($$n = 12$$), trigonelline, trimethylamine N-oxide, p-Cresol sulphate, and choline were quantified. Lipid classes such as lysophosphatidylcholines ($$n = 14$$), phosphatidylcholines ($$n = 76$$), sphingomyelins ($$n = 15$$), ceramides ($$n = 28$$), dihydroceramides ($$n = 8$$), hexosylceramides ($$n = 19$$), dihexosylceramides ($$n = 9$$), trihexosylceramides ($$n = 6$$), cholesteryl esters ($$n = 22$$), diglycerides ($$n = 44$$), and triglycerides ($$n = 242$$) were quantified using the FIA-MS/MS analysis. Furthermore, acylcarnitines ($$n = 40$$) and the sum of hexose were also quantified using the FIA-MS/MS analysis. The multiple reaction monitoring (MRM) method, which was optimized by Biocrates Life Sciences, was applied to identify and quantify all metabolites. ## 2.6. Statistical Analysis All statistics were conducted using GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA). A significance level of $p \leq 0.05$ was used in all analyses and all p-values reported were two-tailed. The Mann–Whitney test was used for the continual non-parametric dependent variable analysis of data with two groups. For more than two groups, the Kruskal–Wallis test with Dunn’s correction was used. Details of specific statistical tests are given in each corresponding figure legend. To determine if protein expression levels identified using 54-plex ELISA correlated with patient clinical factors, Spearman correlations were carried out using R software version 3.6.2 [24]. Correlations were generated using the R package ‘Hmisc’ version 4.4-0 [25]. Graphical representations of correlations were generated with the R package ‘corrplot’ version 0.84 [26]. All correlations with an associated p-value < 0.05 were considered statistically significant. The Holm–Bonferroni post hoc correction was used to control for multiple comparison testing during correlation analysis. ## 3.1. Increased Oxidative Phosphorylation Metabolism and Elevated Secreted Pro-Inflammatory Mediators Were Observed in Adipose Tissue Explants from Viscerally Obese Patients To assess whether obesity determined using visceral fat area alters metabolic and secreted profiles of visceral omental ex vivo explants from obese OAC patients ($$n = 16$$) and non-obese patients ($$n = 16$$), the Agilent Seahorse Xfe24 analyser was used to assess real-time metabolic parameters. To further assess the influence of these clinical parameters on this explant model, the matched ACM secretome was evaluated using MSD 54 plex ELISA and metabolomic and lipidomic profiling. Significant increases were observed in OCR ($$p \leq 0.0009$$) and ECAR ($$p \leq 0.0260$$) (Figure 1A) profiles in visceral adipose explants derived from obese oesophageal cancer patients compared with non-obese patients. Increased secretions of cytokines Eotaxin-3, IL-17A, IL-17D, IL-3, MCP-1, and MDC and a decrease in the secretion of VEGF-D ($$p \leq 0.0356$$, 0.0281, 0.0457, 0.0369, 0.0358, and 0.0408, respectively) (Figure 1B) were observed in adipose explants from obese patients compared with their non-obese counterparts. A decreased expression of metabolites Aspartic acid, Glutamine, and PC aa C42:6, ($$p \leq 0.0240$$, 0.0280, and 0.0063, respectively) (Figure 1C) was observed in the secretome of adipose explants from obese patients compared with their non-obese counterparts. An elevated expression of metabolites GABA, Glutamic acid, TG (16:0_35:3), TG(18:2_38:4), TG(22:5_34:3) (0.0419, 0.0343, 0.0315, 0.0468, 0.0316 respectively) (Figure 1C,D) was additionally observed in the adipose secretome from obese patients compared with non-obese patients. Significant correlations observed between the experimental data and visceral fat area were visualised using corrplot (Figure 1E), and the associated R numbers and p-values are detailed in Table 2. ## 3.2. Adipose Explants Derived from OAC Patients with Metabolic Dysfunction Show Increased Oxidative Phosphorylation Associated Metabolism and Secreted Pro-Inflammatory Mediators Aberrant biological mechanisms such as obesity, diabetes, high triacylglycerides, high cholesterol, and high blood pressure have all been identified as contributors to the development of metabolic syndrome, a pro-inflammatory condition to aid cancer progression [27]. In this study, to assess the influence of metabolic dysfunction, on visceral adipose explants from metabolically healthy ($$n = 17$$) and metabolically dysfunctional ($$n = 15$$) OAC patients, four profiling assays looking at metabolism, secreted pro-inflammatory mediators, and lipid/metabolite profiles were used. In adipose explants derived from patients with clinically annotated metabolic dysfunction, significant increases in OCR ($$p \leq 0.0486$$) and the OCR/ECAR ratio ($$p \leq 0.0402$$) were identified (Figure 2A) in metabolic profiles compared with metabolically healthy patients. Furthermore, significant increases in secreted cytokines including IL-5, IL-7, SAA, and VEGF-C ($$p \leq 0.0473$$, 0.0257, 0.0253, 0.0301) (Figure 2B) were detected in metabolically unhealthy patients compared with metabolically healthy patients. The following metabolites were also identified to be significantly elevated in the adipose secretome of patients with metabolic dysfunction: C0, Glycine, Histidine, Phenylalanine, Tryptophan, Asymmetric dimethylarginine, Homocysteine, Hypoxanthine, Taurine, beta-Ala, CE(22:5), and PC aa C38:4 ($$p \leq 0.0137$$, 0.0299, 0.0275, 0.0452, 0.0372, 0.0122, 0.0016, 0.0299, 0.0172, 0.0219, 0.0327, and 0.0291, respectively) (Figure 2C). Following the lipidomic analysis, triglycerides including TG(16:0_35:3), TG(18:0_34:3), TG(18:1_33:3), TG(18:2_38:4), TG(20:2_36:5), TG(20:4_33:2), TG(20:4_34:0), and TG(20:4_36:5) ($$p \leq 0.0105$$, 0.0036, 0.0009, 0.0025, 0.0152, 0.0327, 0.0321, and 0.0265, respectively) (Figure 2D) were also seen to be elevated in the adipose secretome of OAC patients with metabolic dysfunction compared with patients who were metabolically healthy. Significant correlations were also observed between the experimental data with metabolic dysfunction, Barrett’s oesophagus, smoking history, the ASA grade, and the Clavien–Dindo grade, which were visualised using corrplot (Figure 2E), and the associated R numbers and p-values are detailed in supplemental Table S1. ## 3.3. Adipose Explants from Patients Receiving the FLOT Chemotherapy Regimen Showed Increased Oxidative Phosphorylation and Pro-Inflammatory Mediators and Decreased Triacylglycerides Patients were classified as treatment naïve if they had received no treatment prior to surgical intervention ($$n = 10$$), FLOT if they had received neo-adjuvant chemotherapy only regimen FLOT ($$n = 14$$), or CROSS if they received neo-adjuvant chemo-radiotherapy regimen CROSS (=8). The four experimental assays looking at metabolism, secreted pro-inflammatory mediators, and lipid/metabolite profiles were utilised to identify whether any associations were observed between previous treatment exposure and adipose tissue functionality. Increased OCR ($$p \leq 0.0338$$) (Figure 3A) metabolic profiles were observed in adipose explants derived from patients who had previously been treated with the FLOT chemotherapy regimen compared with patients who did not have neo-adjuvant treatment. Increased secretion of cytokines IL-12p40, IL-1α, IL-22, and TNF-β ($$p \leq 0.0340$$, 0.0426, 0.0350, and 0.0392, respectively) (Figure 3B) were observed in the adipose secretome of patients previously treated with FLOT compared with patients who were treatment-naïve. Decreased secretion of cytokines IL-7 and VEGF-C ($$p \leq 0.0061$$ and 0.0231, respectively) (Figure 3B) were observed in the adipose explants from patients who had received neo-adjuvant chemo-radiotherapy regimen CROSS compared with FLOT. Metabolites including Cer(d16:$\frac{1}{22}$:0), SM (OH) C22:1, TG(16:0_34:3), TG(16:0_40:6), TG(16:1_34:1), TG(16:1_36:4), and TG(18:0_36:3) ($$p \leq 0.0106$$, 0.0165, 0.0375, 0.0092, 0.0138, 0.0470, and 0.0370, respectively) (Figure 3C,D) were significantly decreased in the secretome of the adipose explants derived from patients who had previously been treated with the FLOT regimen compared with treatment-naïve patients. Significant decreases were observed in metabolites p-Cresol-SO4, Hex3Cer(d18:$\frac{1}{24}$:1), and TG(20:2_34:4) ($$p \leq 0.0333$$, 0.0422, and 0.0177, respectively) (Figure 3C,D) whilst metabolite TG(16:1_36:5) ($$p \leq 0.0364$$) was significantly increased in the adipose secretome of patients who had previously received the CROSS regimen compared with patients who were treatment-naïve. Metabolites GUDCA, Hex2Cer(d18:$\frac{1}{14}$:0), and TG(20:2_34:4) ($$p \leq 0.0031$$, 0.0173, and 0.0106, respectively) (Figure 3C,D) were all observed to be decreased and metabolites DCA, TG(14:0_36:2), TG(16:0_33:2), TG(16:0_30:2), TG(18:0_36:3), and TG(18:2_35:2), TG(18:3_32:1) ($$p \leq 0.0069$$, 0.0040, 0.0177, 0.0010, 0.0406, 0.0348, and 0.0131, respectively) (Figure 3C,D) were observed as increased in the adipose secretome of patients who received the chemo-radiotherapy CROSS regimen compared with patients who received the chemotherapy only FLOT regimen. In particular, the metabolite TG(18:0_36:3) was decreased in the adipose secretome of patients receiving the FLOT regimen compared with both the treatment-naïve patients and patients receiving the CROSS regimen ($$p \leq 0.0370$$ and 0.0406, respectively) (Figure 3D) whilst metabolite TG(20:2_34:4) was decreased in the adipose secretome of patients receiving the CROSS regimen compared with both treatment-naïve patients and patients receiving the FLOT regimen ($$p \leq 0.0177$$, 0.0106) (Figure 3D). Significant correlations observed between the experimental data with neo-adjuvant treatment, chemotherapy only, chemoradiotherapy, tumour differentiation, lymph involvement, venous involvement, and perineural involvement were visualised using corrplot (Figure 3E), and the associated R numbers and p-values are detailed in Supplemental Tables S2 and S3. ## 3.4. Increased ECAR and Altered Metabolites Are Observed in Adipose Explants from OAC Patients with Increasing Tumour Regression Grades Tumour regression grading was used to assess patients’ response to treatment received; therefore, experimental data within this section only relates to patients who received neo-adjuvant treatment prior to surgery, and patients who did not receive treatment prior to surgery were excluded from this analysis. TRG 1–2 ($$n = 6$$) indicate patients who had a complete or good response to therapy, patients with TRG 3 ($$n = 8$$) showed an intermediate response, and patients with TRG 4–5 ($$n = 6$$) had a poor response. The association of this staging with adipose explant metabolism and secretome was assessed using the four profiling assays looking at metabolism, secreted pro-inflammatory mediators, and lipid/metabolite profiles. Significantly increased ECAR profiles ($$p \leq 0.0248$$) (Figure 4A) were observed in the secretome of adipose explants derived from patients with TRG scoring of 4–5 compared with patients with TRG scorings of 1–2. A decreased expression of metabolites Cer(d18:$\frac{1}{23}$:0), PC aa C36:1, PC aa C40:3, PC ae C34:2, HexCer(d18:$\frac{1}{18}$:0), TG(17:1_36:3), and TG(18:1_33:3) ($$p \leq 0.0452$$, 0.0445, 0.0183, 0.0318, 0.0269, 0.0262, and 0.0081, respectively) (Figure 4B,C) and an increased expression of DG(18:1_20:1) ($$p \leq 0.0393$$) (Figure 4C) were observed in the adipose secretome from patients with a TRG of 3 compared with patients with TRG scorings of 1–2. A decreased expression of metabolites Cer(d18:$\frac{1}{23}$:0), TG(17:0_36:4), and TG(22:6_34:3) ($$p \leq 0.0409$$, 0.0249, and 0.0451, respectively) (Figure 4B,C) and an increased expression of TG(18:0_38:6) ($$p \leq 0.0434$$) (Figure 4C) were observed in the secretome of adipose explants derived from patients with a TRG of 4–5 compared with patients with TRG scorings of 1–2. A significant decrease was observed in metabolites Cer(d16:$\frac{1}{23}$:0), PC aa C36:1, PC aa C40:3, PC ae C40:2, PC ae C44:5, Hex2Cer(d18:$\frac{1}{22}$:0), TG(17:1_36:3), TG(18:0_38:6), TG(20:3_34:3), and TG(20:4_36:5) ($$p \leq 0.0163$$, 0.0303, 0.0246, 0.0446, 0.0324, 0.0493, 0.0425, 0.0478, 0.0173, and 0.0046, respectively) (Figure 4B,C) and a significant increase in triglyceride TG(20:4_36:3) ($$p \leq 0.0252$$) (Figure 4C) were observed in the adipose secretome of patients with a TRG of 3 compared with patients with a TRG of 4–5. Of note, metabolite Cer(d18:$\frac{1}{23}$:0) ($$p \leq 0.0452$$, 0.0409) (Figure 4B) was decreased in the adipose secretome of patients with a TRG score of 3 and TRG of 4–5 compared with patients who had a TRG of 1–2. A decreased expression was also observed in TG(18:0_38:6) ($$p \leq 0.0434$$, 0.0478) (Figure 4C) in the secretome of adipose explants derived from patients with a TRG of 1–2 and a TRG of 3 compared with patients who possessed a TRG score of 4–5. It was also identified that metabolites including PC aa C36:1 ($$p \leq 0.0445$$, 0.0303), PC aa C40:3, ($$p \leq 0.0183$$, 0.0246), and TG(17:1_36:3) ($$p \leq 0.0262$$, 0.0425) (Figure 4B,C) had increased expression in the adipose secretome of patients with TRGs of 1–2 and 4–5, respectively, compared with patients with a TRG scoring of 3. Significant correlations observed between experimental data with tumour regression grade, clinical tumour stage, clinical nodal stage, pathological tumour stage, pathological nodal stage, and no evidence of disease were visualised using corrplot (Figure 4D), and the associated R numbers and p-values are detailed in Supplemental Tables S4 and S5. ## 4. Discussion This study for the first time has conducted a detailed assessment of adipose tissue metabolism, as well as adipose secreted pro-inflammatory, metabolite, and lipid profiles and determined whether these profiles correlate with significant clinical parameters in OAC patients including obesity, metabolic dysfunction, previous treatment exposure and tumour regression grades. The recent literature has identified an elevated reliance of certain subtypes of cancers on oxidative phosphorylation rather than glycolysis [28]. Previous research from our group has shown an increased reliance on oxidative phosphorylation correlating with enhanced radioresistance in oesophageal cancer cells [29]. In this study, elevated OCR and ECAR metabolic profiles have been observed in adipose explants from obese OAC patients compared with their non-obese counterparts, indicating higher metabolic rates with increasing visceral adiposity. Previous work conducted in mouse models has shown that diminished oxidative phosphorylation function ameliorated obesity in mice on high-fat diets disrupting weight gain and improving glucose tolerance [30]. This study further shows that elevated OCR and OCR/ECAR ratio profiles were observed in patients classified with metabolic dysfunction compared with metabolically healthy patients. Previously, increased mitochondrial respiration was observed in adipocytes of metabolically unhealthy patients compared with metabolically healthy patients [19]. This shift towards utilisation of oxidative phosphorylation pathways within adipose tissue of obese and metabolically unhealthy individuals may provide insight into aberrant mitochondria function promoting pro-tumorigenic signalling. An elevated utilisation of oxidative phosphorylation pathways was observed in adipose tissue from our patients who received FLOT chemotherapy treatment compared with treatment-naïve patients. Previous research has shown that treatment with 5-FU and oxaliplatin upregulated genes associated with oxidative phosphorylation in mouse models [31]. This may be further augmented by elevated oxidative phosphorylation in adipose tissue sequestering chemotherapy within adipose tissue [32], which may hinder chemotherapy efficacy. Additionally, an increased reliance on glycolysis was observed in adipose explants with increasing TRGs (poor response). Studies have shown cancer cells co-cultured with adipose stromal cells upregulate glycolysis and mediated chemoresistance, which may support why elevated ECAR is observed only in adipose explants of patients with the poorest response to chemotherapy [33]. The secretome of adipose tissue comprises many pro-tumorigenic cytokines that are primed to aid cancer cell growth and survival [34]. Pro-inflammatory cytokines and metabolism are interlinked and essential in regulating adipocyte function and lipid metabolism, particularly in metabolic diseases [35]. Eotaxin-3, an adipose-associated cytokine, was elevated with obesity status in this study. Eotaxin-3 correlates with increased BMI [36] and is highly expressed in the circulation of OAC patients with longer survival rates [37]. However, Eotaxin-3 had also been reported to be highly expressed in tumour tissue in aggressive disease [38] and with increased macrophage infiltration [38]. Coupling this knowledge with the observed increased secretion of MDC and MCP-1 in this study, both macrophage-associated cytokines [39,40], the adipose secretome of obese individuals may aid the recruitment of macrophages and other immune cells to the adipose tissue leading to a diminished immune response in the local tumour microenvironment. The secretion of IL-17 A and IL-17 D was also increased in the adipose secretome of obese patients. Studies have linked elevated IL-17 in adipose tissue to increases in infiltrating immune cells [41,42] and poor prognosis [43], and the disruption of this pathway diminishes metastasis and enhances treatment response [44,45,46]. This cytokine may be a potential target in the obese tumour microenvironment. This study has also shown a decreased secretion of VEGF-D in the adipose secretome of obese OAC patients. Previous work has observed that while VEGF-A, VEGF-B, and VEGF-C showed increased circulating levels in obese individuals, the levels of VEGF-D were significantly diminished [47]. VEGF-D in particular has been linked with metastatic disease, and this diminished secretion may be amelioratory [48]. Increased expression of IL-5 in the adipose secretome of patients with metabolic dysfunction was identified. Previous studies in mouse models with suppressed IL-5/CD300f expression have shown decreased diet-induced weight gain and insulin resistance [49]. IL-5 facilitates lung metastasis [50] and targeting the IL-5 axis could ameliorate metabolic dysfunction symptoms. Previous findings have identified elevated circulating SAA in patients with metabolic syndrome, as was observed in the adipose secretome of metabolically unhealthy patients in this cohort [51]. SAA relates to the inflammatory processes, which play pivotal roles in both obesity and metabolic syndrome [52]. Elevated levels of IL-7 were observed in this study in the adipose secretome of metabolically unhealthy patients and in patients who received FLOT chemotherapy compared with CROSS chemo-radiotherapy. Previous work has shown that IL-7 overexpression in mouse models is associated with glucose and insulin resistance [53]. However, elevated IL-7 could prove beneficial in the context of FLOT chemotherapy as IL-7 can aid in T cell growth [54]. Recent studies have indicated that IL-7 could potentially re-sensitize tumours to cisplatin [55]. As previously mentioned, obese individuals have elevated circulating levels of VEGF-C, and the overexpression of VEGF-C in mouse models has been linked to increased weight gain and insulin resistance [47,56]. Consequently, this elevated secretion in the adipose secretome could potentiate metabolic disorder and prime the local tumour microenvironment for cancer cell survival. Elevated VEGF-C has also been linked to chemoresistance [57], and the increased secretion of VEGF-C observed within the adipose secretome of FLOT chemotherapy-treated patients compared to CROSS-treated patients may indicate a more systemic effect as this VEGF-C enriched adipose secretome could potentiate VEGF-C mediated metastasis [58]. Increased secretion of IL-22 and TNF-β was observed in the adipose secretome of patients who received FLOT chemotherapy compared with patients who received no neo-adjuvant therapy. IL-22, whilst possessing both pro- and anti-inflammatory effects, can promote cancer cell growth and enhance chemoresistance [59], allowing infiltration of M2-like macrophages in adipose tissue, and driving a systemic anti-tumour effect [60]. TNF-β promotes proliferation and invasion in cancer cell models [61]; however, the influence of chemotherapies on the functional role of TNF-β in adipose tissue is unknown. This study reports elevated levels of pro-inflammatory cytokines IL-12p40 and IL-1α in the adipose secretome of patients who received FLOT chemotherapy compared with treatment-naïve patients. Expression of IL-1 α in tumour tissue of gastric cancers has been shown to correlate with a more aggressive disease state and metastasis [62]; however, these cytokines can also act in immune cell recruitment [63]. Recent research has supported the utility of the metabolome in many diseases [64]. This study observed altered secretions in key metabolites involved in the glutamate/GABA-glutamine cycle. Increased secretion of GABA and glutamic acid (ionic form glutamate) and decreased secretion of glutamine were observed in the adipose secretome of obese patients compared with their non-obese counterparts. Previous research has also observed decreased glutamine and increased glutamate in the circulation of obese individuals [65,66]. Glutamine is effective in polarisation of anti-inflammatory M2-like macrophages, which may act as a potential therapeutic target to resolve the low-grade inflammation associated with obesity [66]. Aspartic acid was also decreased in the adipose secretome of obese patients compared with non-obese patients, which emulates previous research where decreased N-acetylaspartate was observed in patients with higher BMIs [67]. A combination of aspartic acid and glutamic acid may be capable of inducing tumour cell death [68]. Interestingly, these amino acids were decreased in the obese setting in our study. Elevated secreted levels of triacylglycerides TG(16:0_35:3), TG(18:2_38:4), and TG(22:5_34:3) were also detected in adipose tissue from obese OAC patients compared with non-obese. Accumulation of triacylglycerides has previously been reported in obese adipose tissue [69] and in cancers [70,71]. Elevated levels of asymmetric dimethylarginine, homocysteine, and hypoxanthine have been reported in individuals with metabolic syndrome [72,73,74]. This study also identified increased secretion of these amino acids in the adipose tissue of OAC patients that have metabolic dysfunction compared with metabolically healthy patients. Previous research has identified increased circulating levels of both homocysteine and hypoxanthine in cancer patients; however, the role they play in tumorigeneses or cancer cell survival is not fully understood [75,76]. However, reports of elevated asymmetric dimethylarginine in cancer have suggested it may attenuate apoptosis in response to stress and chemotherapy [77]. This study has also identified elevated levels of glycine and taurine in the adipose secretome of metabolically unhealthy versus metabolically healthy OAC patients. Elevated circulating levels of these amino acids should ameliorate the pro-inflammatory effects of metabolic syndrome and attenuate cancer cell proliferation as well as enhance the efficacy of chemotherapy [78,79]. Furthermore, p-Cresyl sulfate was decreased in the adipose secretome of patients who received the CROSS regimen compared with patients who received no neo-adjuvant therapy. p-Cresyl sulphate aids in cancer cell migration and epithelial–mesenchymal transition [80,81] and its diminished secretion in the adipose secretome of CROSS-treated patients could possess beneficial effects. In this study, DCA was observed to be increased in the adipose secretome of CROSS regimen-treated patients compared with FLOT-treated patients, whilst GUDCA (glycoursodeoxycholic acid) was observed to be decreased. DCA (dichloroacetate) has previously been reported to decrease cell proliferation and migration in mouse models [82]. A series of triglycerides were also identified to be decreased in the adipose secretome of the FLOT regimen receiving patients compared with treatment-naïve patients. Only TG(20:2_34:4) was found to be decreased in the adipose secretome of patients receiving CROSS compared with treatment-naïve and FLOT-receiving patients. In particular, the metabolite TG(18:0_36:3) was decreased in the adipose secretome of patients receiving the FLOT regimen compared with both treatment-naïve patients and patients receiving the CROSS regimen, whilst metabolite TG(20:2_34:4) was decreased in the adipose secretome of patients receiving the CROSS regimen compared with both treatment-naïve patients and patients receiving the FLOT regimen. Previous research has identified an increase in circulating triglycerides during neo-adjuvant chemotherapy which gradually decreases to normal levels [83]. Decreased secretion of triglycerides in the adipose secretome of patients only receiving chemotherapy raises the question of whether this decreased expression correlates with increased circulating levels and whether this can be utilised to enhance current therapies. Diminished levels of ceramides, glycosylceramides, and sphingomyelins were also observed in the adipose secretome of treated patients compared with treatment-naïve patients. Ceramide metabolism has previously been reported to induce cancer cell death through induction of cellular stress [84], and the diminished presence of ceramide and associated molecules following cancer treatment exposure could prove interesting as a therapeutic target. Metabolite Cer(d18:$\frac{1}{23}$:0) was observed to be decreased in the adipose secretome of patients with a TRG 3 and TRG 4–5 compared with patients who had a TRG 1–2. Diminished levels of ceramides could prevent ceramide metabolism, which induces cancer cell death through the induction of cellular stress [84]. The diminished presence of ceramide in the adipose secretome of patients with more aggressive cancers could indicate that anti-cancer ceramide analogues may aid poor responding tumours [85]. Decreased expression was observed in TG(18:0_38:6) in the secretome of adipose explants derived from patients with a TRG of 1–2 and a TRG of 3 compared with patients who had a TRG of 4–5. Higher levels of TG(18:0_38:6) were observed in the adipose secretome of patients with a TRG 4–5 compared with a TRG 1–2, in addition to increased expression of TG(17:1_36:3), TG(18:0_38:6), TG(20:3_34:3), and TG(20:4_36:5) in TRG 4–5 patients compared with TRG 1–2. Higher expression of triglycerides has been linked with more aggressive cancers and decreased disease-free survival and overall survival [86]. Metabolites including PC aa C36:1 and PC aa C40:3 had increased expression in the adipose secretome of patients with TRGs of 1–2 and 4–5 compared with patients with a TRG 3. Phosphatidylcholine metabolism is linked to both cellular proliferation and cell death [87]. The decreased expression found in this study in the adipose secretome of patients with a TRG of 3 compared with patients with more regressive TRG 1–2 and more aggressive cancers with TRG scoring of 4–5 pose an interesting question of whether the adipose secretome and phosphatidylcholine metabolism may attenuate or potentiate cancer cells response to current treatment modalities. Additionally, elevated Clavien–Dindo grade, a classification of post-operative complications was positively correlated with a series of adipose-secreted cytokines associated with pro-inflammatory response and angiogenesis including IL-6, IL-16, FLt-1, PlGF, VEGF-A, and VEGF-D. It has been previously reported that increased expression of circulating IL-6 following major abdominal surgery was associated with an increased risk of complications [88]. Furthermore, previously elevated circulating levels of VEGF have been reported following major abdominal surgery [89] and are associated with poor cancer-specific survival [90]. The elevation of these cytokines in the adipose secretome could potentiate post-op complications in these patients and may act as a future therapeutic target to ameliorate the locally based and circulating effects of these cytokines. Within this study, positive correlations were observed between increasing secretion of IL-3, IL-5, and lactic acid in the adipose secretome and patients with increasing nodal invasion. Previous research has indicated that tumour-associated leukocytes from patients with metastatic lymph nodes secreted higher levels of IL-3 and IL-5 [91]. Additionally, elevated levels of circulating lactate were previously identified in OAC patients whose tumours had metastasised to the lymph nodes compared with patients with no lymph node metastasis [92]. The potential of the adipose secretome to augment these analytes in circulation and whether this plays a pivotal role in aiding tumour metastasis to the lymph node requires further investigation to determine if these associations could be utilised for therapeutic benefit. Furthermore, within this study, decreased levels of PlGF and lactic acid in the adipose secretome were correlated with patients who showed no evidence of disease following treatment and resection. As previously mentioned, elevated lactate levels have been associated with nodal metastasis [92], and it has also been reported to be generated by cancer cells and holds critical roles in cancer cell proliferation, promoting angiogenesis and acting as a key immunosuppressive analyte [93]. This decreased expression of lactate in the adipose secretome in patients whose tumours have regressed is of interest as it raises questions on how adipose tissue plays a role in sequestering or releasing lactate and whether this could potentiate the tumour microenvironment to aid nodal invasion or recede to support cancer regression. Additionally, decreasing levels of PlGF were observed in the adipose secretome of patients whose cancers had regressed. PlGF has been reported to promote tumour desmoplasia in pancreatic mouse models with PlGF blockade., enhancing the efficacy of chemotherapy [94]. It is unknown if PlGF reduction could enhance tumour regression. ## 5. Conclusions Overall, in this study, increases in OCR-linked oxidative phosphorylation, pro-inflammatory cytokines, and metabolites associated with aiding tumorigenesis have been identified in the most viscerally obese OAC patients and in patients with metabolic dysfunction. This now raises the question of whether targeting these altered signalling mechanisms could aid current treatment strategies. With obesity levels steadily increasing, understanding of the complex nature of adipose tissue and whether its effects can be harnessed for therapeutic gain is critically important. ## References 1. Zhang Y.. Epidemiology of Esophageal Cancer. World J. Gastroenterol. (2013.0) 19 5598-5606. DOI: 10.3748/wjg.v19.i34.5598 2. De Vegt F., Gommers J.J.J., Groenewoud H., Siersema P.D., Verbeek A.L.M., Peters Y., Broeders M.J.M.. Trends and Projections in the Incidence of Oesophageal Cancer in the Netherlands: An Age-Period-Cohort Analysis from 1989 to 2041. Int. J. Cancer (2022.0) 150 420-430. DOI: 10.1002/ijc.33836 3. Reynolds J.V., Preston S.R., O’Neill B., Lowery M.A., Baeksgaard L., Crosby T., Cunningham M., Cuffe S., Griffiths G.O., Roy R.. Neo-AEGIS (Neoadjuvant Trial in Adenocarcinoma of the Esophagus and Esophago-Gastric Junction International Study): Preliminary Results of Phase III RCT of CROSS versus Perioperative Chemotherapy (Modified MAGIC or FLOT Protocol). (NCT01726452). J. Clin. Oncol. (2021.0) 39 4004. DOI: 10.1200/JCO.2021.39.15_suppl.4004 4. Noble F., Lloyd M.A., Turkington R., Griffiths E., O’Donovan M., O’Neill J.R., Mercer S., Parsons S.L., Fitzgerald R.C., Underwood T.J.. Multicentre Cohort Study to Define and Validate Pathological Assessment of Response to Neoadjuvant Therapy in Oesophagogastric Adenocarcinoma. Br. J. Surg. (2017.0) 104 1816. DOI: 10.1002/bjs.10627 5. Donohoe C.L., O’Farrell N.J., Grant T., King S., Clarke L., Muldoon C., Reynolds J.V.. Classification of Pathologic Response to Neoadjuvant Therapy in Esophageal and Junctional Cancer: Assessment of Existing Measures and Proposal of a Novel 3-Point Standard. Ann. Surg. (2013.0) 258 784-792. DOI: 10.1097/SLA.0b013e3182a66588 6. Ryan A.M., Rowley S.P., Fitzgerald A.P., Ravi N., Reynolds J.V.. Adenocarcinoma of the Oesophagus and Gastric Cardia: Male Preponderance in Association with Obesity. Eur. J. Cancer (2006.0) 42 1151-1158. DOI: 10.1016/j.ejca.2005.12.024 7. Renehan A.G., Roberts D.L., Dive C.. Obesity and Cancer: Pathophysiological and Biological Mechanisms. Arch. Physiol. Biochem. (2008.0) 114 71-83. DOI: 10.1080/13813450801954303 8. Fang X., Wei J., He X., Lian J., Han D., An P., Zhou T., Liu S., Wang F., Min J.. Quantitative Association between Body Mass Index and the Risk of Cancer: A Global Meta-Analysis of Prospective Cohort Studies. Int. J. Cancer (2018.0) 143 1595-1603. DOI: 10.1002/ijc.31553 9. Park J., Morley T.S., Kim M., Clegg D.J., Scherer P.E.. Obesity and Cancer—Mechanisms Underlying Tumour Progression and Recurrence. Nat. Rev. Endocrinol. (2014.0) 10 455-465. DOI: 10.1038/nrendo.2014.94 10. Quail D.F., Dannenberg A.J.. The Obese Adipose Tissue Microenvironment in Cancer Development and Progression. Nat. Rev. Endocrinol. (2019.0) 15 139-154. DOI: 10.1038/s41574-018-0126-x 11. Iyengar N.M., Gucalp A., Dannenberg A.J., Hudis C.A.. Obesity and Cancer Mechanisms: Tumor Microenvironment and Inflammation. J. Clin. Oncol. (2016.0) 34 4270-4276. DOI: 10.1200/JCO.2016.67.4283 12. Davern M., Bracken-Clarke D., Donlon N.E., Sheppard A.D., Butler C., Donohoe C., Ravi N.. Visceral Adipose Tissue Secretome from Early and Late-Stage Oesophageal Cancer Patients Differentially Affects Effector and Regulatory T Cells. J. Cancer Res. Clin. Oncol. 2022. DOI: 10.1007/s00432-023-04620-6 13. Mylod E., O’connell F., Donlon N.E., Butler C., Reynolds J.V., Lysaght J., Conroy M.J.. The Omentum in Obesity-Associated Cancer: A Hindrance to Effective Natural Killer Cell Migration towards Tumour Which Can Be Overcome by CX3CR1 Antagonism. Cancers (2021.0) 14. DOI: 10.3390/cancers14010064 14. Melo A.M., Mylod E., Fitzgerald V., Donlon N.E., Murphy D.M., Foley E.K., Bhardwaj A., Reynolds J.V., Doherty D.G., Lysaght J.. Tissue Distribution of Γδ T Cell Subsets in Oesophageal Adenocarcinoma. Clin. Immunol. (2021.0) 229 108797. DOI: 10.1016/j.clim.2021.108797 15. Mongan A.M., Lynam-Lennon N., Doyle S.L., Casey R., Carr E., Cannon A., Conroy M.J., Pidgeon G.P., Brennan L., Lysaght J.. Visceral Adipose Tissue Modulates Radiosensitivity in Oesophageal Adenocarcinoma. Int. J. Med. Sci. (2019.0) 16 519-528. DOI: 10.7150/ijms.29296 16. Heeran A.B., McCready J., Dunne M.R., Donlon N.E., Nugent T.S., Bhardwaj A., Mitchelson K.A.J., Buckley A.M., Ravi N., Roche H.M.. Opposing Immune-Metabolic Signature in Visceral Versus Subcutaneous Adipose Tissue in Patients with Adenocarcinoma of the Oesophagus and the Oesophagogastric Junction. Metabolites (2021.0) 11. DOI: 10.3390/metabo11110768 17. Lashinger L.M., Rossi E.L., Hursting S.D.. Obesity and Resistance to Cancer Chemotherapy: Interacting Roles of Inflammation and Metabolic Dysregulation. Clin. Pharmacol. Ther. (2014.0) 96 458-463. DOI: 10.1038/clpt.2014.136 18. Ross K.H., Gogineni K., Subhedar P.D., Lin J.Y., McCullough L.E.. Obesity and Cancer Treatment Efficacy: Existing Challenges and Opportunities. Cancer (2019.0) 125 1588-1592. DOI: 10.1002/cncr.31976 19. Böhm A., Keuper M., Meile T., Zdichavsky M., Fritsche A., Häring H.U., de Angelis M.H., Staiger H., Franko A.. Increased Mitochondrial Respiration of Adipocytes from Metabolically Unhealthy Obese Compared to Healthy Obese Individuals. Sci. Rep. (2020.0) 10 12407. DOI: 10.1038/s41598-020-69016-9 20. Buckley A.M., Dunne M.R., Morrissey M.E., Kennedy S.A., Nolan A., Davern M., Foley E.K., Clarke N., Lysaght J., Ravi N.. Real-Time Metabolic Profiling of Oesophageal Tumours Reveals an Altered Metabolic Phenotype to Different Oxygen Tensions and to Treatment with Pyrazinib. Sci. Rep. (2020.0) 10 12105. DOI: 10.1038/s41598-020-68777-7 21. Finkelstein E.A., Khavjou O.A., Thompson H., Trogdon J.G., Pan L., Sherry B., Dietz W.. Obesity and Severe Obesity Forecasts through 2030. Am. J. Prev. Med. (2012.0) 42 563-570. DOI: 10.1016/j.amepre.2011.10.026 22. Doyle S.L., Bennett A.M., Donohoe C.L., Mongan A.M., Howard J.M., Lithander F.E., Pidgeon G.P., Reynolds J.V., Lysaght J.. Establishing Computed Tomography–Defined Visceral Fat Area Thresholds for Use in Obesity-Related Cancer Research. Nutr. Res. (2013.0) 33 171-179. DOI: 10.1016/j.nutres.2012.12.007 23. Grundy S.M., Cleeman J.I., Daniels S.R., Donato K.A., Eckel R.H., Franklin B.A., Gordon D.J., Krauss R.M., Savage P.J., Smith S.C.. Diagnosis and Management of the Metabolic Syndrome. Circulation (2005.0) 112 2735-2752. DOI: 10.1161/CIRCULATIONAHA.105.169404 24. 24. R Core Team R: A Language and Environment for Statistical Computing(R software version 3.6.2)R Foundation for Statistical ComputingVienna, Austria2019. R: A Language and Environment for Statistical Computing (2019.0) 25. Harrell F.E., Dupont C.. Hmisc: Harrell Miscellaneous, Version 4.8-0 (2020.0) 26. Wei T., Simko V.. R Package “Corrplot”: Visualization of a Correlation Matrix, Version 0.92 (2017.0) 27. Zhang J., Wu H., Wang R.. Metabolic Syndrome and Esophageal Cancer Risk: A Systematic Review and Meta-analysis. Diabetol. Metab. Syndr. (2021.0) 13 8. DOI: 10.1186/s13098-021-00627-6 28. Ashton T.M., Gillies McKenna W., Kunz-Schughart L.A., Higgins G.S.. Oxidative Phosphorylation as an Emerging Target in Cancer Therapy. Clin. Cancer Res. (2018.0) 24 2482-2490. DOI: 10.1158/1078-0432.CCR-17-3070 29. Lynam-Lennon N., Maher S.G., Maguire A., Phelan J., Muldoon C., Reynolds J.V., O’Sullivan J.. Altered Mitochondrial Function and Energy Metabolism is Associated with a Radioresistant Phenotype in Oesophageal Adenocarcinoma. PLoS ONE (2014.0) 9. DOI: 10.1371/journal.pone.0100738 30. Choi M.J., Jung S.B., Lee S.E., Kang S.G., Lee J.H., Ryu M.J., Chung H.K., Chang J.Y., Kim Y.K., Hong H.J.. An Adipocyte-Specific Defect in Oxidative Phosphorylation Increases Systemic Energy Expenditure and Protects against Diet-Induced Obesity in Mouse Models. Diabetologia (2020.0) 63 837-852. DOI: 10.1007/s00125-019-05082-7 31. Vellinga T.T., De Boer V.C.J., Fatrai S., Van Schelven S., Trumpi K., Verheem A., Snoeren N., Emmink B.L., Koster J., Rinkes I.H.M.B.. SIRT1/PGC1a-Dependent Increase in Oxidative Phosphorylation Supports Chemotherapy Resistance of Colon Cancer. Clin. Cancer Res. (2015.0) 21 2870-2879. DOI: 10.1158/1078-0432.CCR-14-2290 32. Sheng X., Parmentier J.-H., Tucci J., Pei H., Cortez-Toledo O., Dieli-Conwright C.M., Oberley M.J., Neely M., Orgel E., Louie S.G.. Adipocytes Sequester and Metabolize the Chemotherapeutic Daunorubicin. Mol. Cancer Res. (2017.0) 15 1704-1713. DOI: 10.1158/1541-7786.MCR-17-0338 33. Rizi B.S., Caneba C., Nowicka A., Nabiyar A.W., Liu X., Chen K., Klopp A., Nagrath D.. Nitric Oxide Mediates Metabolic Coupling of Omentum-Derived Adipose Stroma to Ovarian and Endometrial Cancer Cells. Cancer Res. (2015.0) 75 456-471. DOI: 10.1158/0008-5472.CAN-14-1337 34. O’Connell F., O’Sullivan J.. Help or Hindrance: The Obesity Paradox in Cancer Treatment Response. Cancer Lett. (2021.0) 522 269-280. DOI: 10.1016/j.canlet.2021.09.021 35. Chen Y., Yu C.Y., Deng W.M.. The Role of Pro-Inflammatory Cytokines in Lipid Metabolism of Metabolic Diseases. Int. Rev. Immunol. (2019.0) 38 249-266. DOI: 10.1080/08830185.2019.1645138 36. Gentili A., Zaibi M.S., Alomar S.Y., De Vuono S., Ricci M.A., Alaeddin A., Siepi D., Boni M., Vaudo G., Trayhurn P.. Circulating Levels of the Adipokines Monocyte Chemotactic Protein-4 (MCP-4), Macrophage Inflammatory Protein-1β (MIP-1β), and Eotaxin-3 in Severe Obesity and Following Bariatric Surgery. Horm. Metab. Res. (2016.0) 48 847-853. DOI: 10.1055/s-0042-108731 37. Donlon N.E., Sheppard A., Davern M., O’connell F., Phelan J.J., Power R., Nugent T., Dinneen K., Aird J., Greene J.. Linking Circulating Serum Proteins with Clinical Outcomes in Esophageal Adenocarcinoma—An Emerging Role for Chemokines. Cancers (2020.0) 12. DOI: 10.3390/cancers12113356 38. Lan Q., Lai W., Zeng Y., Liu L., Li S., Jin S., Zhang Y., Luo X., Xu H., Lin X.. CCL26 Participates in the PRL-3-Induced Promotion of Colorectal Cancer Invasion by Stimulating Tumor-Associated Macrophage Infiltration. Mol. Cancer Ther. (2018.0) 17 276-289. DOI: 10.1158/1535-7163.MCT-17-0507 39. Godiska R., Chantry D., Raport C.J., Sozzani S., Allavena P., Leviten D., Mantovani A., Gray P.W.. Human Macrophage–Derived Chemokine (MDC), a Novel Chemoattractant for Monocytes, Monocyte-Derived Dendritic Cells, and Natural Killer Cells. J. Exp. Med. (1997.0) 185 1595-1604. DOI: 10.1084/jem.185.9.1595 40. Yoshimura T., Imamichi T., Weiss J.M., Sato M., Li L., Matsukawa A., Wang J.M.. Induction of Monocyte Chemoattractant Proteins in Macrophages via the Production of Granulocyte/Macrophage Colony-Stimulating Factor by Breast Cancer Cells. Front. Immunol. (2016.0) 7 2. DOI: 10.3389/fimmu.2016.00002 41. Chehimi M., Vidal H., Eljaafari A.. Pathogenic Role of IL-17-Producing Immune Cells in Obesity, and Related Inflammatory Diseases. J. Clin. Med. (2017.0) 6. DOI: 10.3390/jcm6070068 42. Carolan E., Tobin L.M., Mangan B.A., Corrigan M., Gaoatswe G., Byrne G., Geoghegan J., Cody D., O’Connell J., Winter D.C.. Altered Distribution and Increased IL-17 Production by Mucosal-Associated Invariant T Cells in Adult and Childhood Obesity. J. Immunol. (2015.0) 194 5775-5780. DOI: 10.4049/jimmunol.1402945 43. Punt S., Langenhoff J.M., Putter H., Fleuren G.J., Gorter A., Jordanova E.S.. The Correlations between IL-17 vs. Th17 Cells and Cancer Patient Survival: A Systematic Review. Oncoimmunology (2015.0) 4 e984547. DOI: 10.4161/2162402X.2014.984547 44. Coffelt S.B., Kersten K., Doornebal C.W., Weiden J., Vrijland K., Hau C.S., Verstegen N.J.M., Ciampricotti M., Hawinkels L.J.A.C., Jonkers J.. IL-17-Producing Γδ T Cells and Neutrophils Conspire to Promote Breast Cancer Metastasis. Nature (2015.0) 522 345-348. DOI: 10.1038/nature14282 45. Lotti F., Jarrar A.M., Pai R.K., Hitomi M., Lathia J., Mace A., Gantt G.A., Sukhdeo K., DeVecchio J., Vasanji A.. Chemotherapy Activates Cancer-Associated Fibroblasts to Maintain Colorectal Cancer-Initiating Cells by IL-17A. J. Exp. Med. (2013.0) 210 2851-2872. DOI: 10.1084/jem.20131195 46. Lee E.J., Park H.J., Lee I.J., Kim W.W., Ha S.J., Suh Y.G., Seong J.. Inhibition of IL-17A Suppresses Enhanced-Tumor Growth in Low Dose Pre-Irradiated Tumor Beds. PLoS ONE (2014.0) 9. DOI: 10.1371/journal.pone.0106423 47. Gómez-Ambrosi J., Catalán V., Rodríguez A., Ramírez B., Silva C., Gil M.J., Salvador J., Frühbeck G.. Involvement of Serum Vascular Endothelial Growth Factor Family Members in the Development of Obesity in Mice and Humans. J. Nutr. Biochem. (2010.0) 21 774-780. DOI: 10.1016/j.jnutbio.2009.05.004 48. Stacker S.A., Caesar C., Baldwin M.E., Thornton G.E., Williams R.A., Prevo R., Jackson D.G., Nishikawa S.I., Kubo H., Achen M.G.. VEGF-D Promotes the Metastatic Spread of Tumor Cells via the Lymphatics. Nat. Med. (2001.0) 7 186-191. DOI: 10.1038/84635 49. Rozenberg P., Reichman H., Zab-Bar I., Itan M., Pasmanik-Chor M., Bouffi C., Qimron U., Bachelet I., Fulkerson P.C., Rothenberg M.E.. CD300f:IL-5 Cross-Talk Inhibits Adipose Tissue Eosinophil Homing and Subsequent IL-4 Production. Sci. Rep. (2017.0) 7 5922. DOI: 10.1038/s41598-017-06397-4 50. Zaynagetdinov R., Sherrill T.P., Gleaves L.A., McLoed A.G., Saxon J.A., Habermann A.C., Connelly L., Dulek D., Peebles R.S., Fingleton B.. Interleukin-5 Facilitates Lung Metastasis by Modulating the Immune Microenvironment. Cancer Res. (2015.0) 75 1624. DOI: 10.1158/0008-5472.CAN-14-2379 51. Deetman P.E., Bakker S.J.L., Dullaart R.P.F.. High Sensitive C-Reactive Protein and Serum Amyloid A Are Inversely Related to Serum Bilirubin: Effect-Modification by Metabolic Syndrome. Cardiovasc. Diabetol. (2013.0) 12 166. DOI: 10.1186/1475-2840-12-166 52. Malle E., De Beer F.C.. Human Serum Amyloid A (SAA) Protein: A Prominent Acute-Phase Reactant for Clinical Practice. Eur. J. Clin. Investig. (1996.0) 26 427-435. DOI: 10.1046/j.1365-2362.1996.159291.x 53. Lucas S., Taront S., Magnan C., Fauconnier L., Delacre M., Macia L., Delanoye A., Verwaerde C., Spriet C., Saule P.. Interleukin-7 Regulates Adipose Tissue Mass and Insulin Sensitivity in High-Fat Diet-Fed Mice through Lymphocyte-Dependent and Independent Mechanisms. PLoS ONE (2012.0) 7. DOI: 10.1371/journal.pone.0040351 54. Morre M., Beq S.. Interleukin-7 and Immune Reconstitution in Cancer Patients: A New Paradigm for Dramatically Increasing Overall Survival. Target. Oncol. (2012.0) 7 55. DOI: 10.1007/s11523-012-0210-4 55. Ke B., Wei T., Huang Y., Gong Y., Wu G., Liu J., Chen X., Shi L.. Interleukin-7 Resensitizes Non-Small-Cell Lung Cancer to Cisplatin via Inhibition of ABCG2. Mediat. Inflamm. (2019.0) 2019 7241418. DOI: 10.1155/2019/7241418 56. Karaman S., Hollmén M., Yoon S.Y., Alkan H.F., Alitalo K., Wolfrum C., Detmar M.. Transgenic Overexpression of VEGF-C Induces Weight Gain and Insulin Resistance in Mice. Sci. Rep. (2016.0) 6 31566. DOI: 10.1038/srep31566 57. Su J.L., Yen C.J., Chen P.S., Chuang S.E., Hong C.C., Kuo I.H., Chen H.Y., Hung M.C., Kuo M.L.. The Role of the VEGF-C/VEGFR-3 Axis in Cancer Progression. Br. J. Cancer (2006.0) 96 541-545. DOI: 10.1038/sj.bjc.6603487 58. Wang C.A., Harrell J.C., Iwanaga R., Jedlicka P., Ford H.L.. Vascular Endothelial Growth Factor C Promotes Breast Cancer Progression via a Novel Antioxidant Mechanism That Involves Regulation of Superoxide Dismutase 3. Breast Cancer Res. (2014.0) 16 462. DOI: 10.1186/s13058-014-0462-2 59. Kobold S., Völk S., Clauditz T., Küpper N.J., Minner S., Tufman A., Düwell P., Lindner M., Koch I., Heidegger S.. Interleukin-22 Is Frequently Expressed in Small- and Large-Cell Lung Cancer and Promotes Growth in Chemotherapy-Resistant Cancer Cells. J. Thorac. Oncol. (2013.0) 8 1032-1042. DOI: 10.1097/JTO.0b013e31829923c8 60. Kim E.Y., Noh H.M., Choi B., Park J.E., Kim J.E., Jang Y., Lee H.K., Chang E.J.. Interleukin-22 Induces the Infiltration of Visceral Fat Tissue by a Discrete Subset of Duffy Antigen Receptor for Chemokine-Positive M2-Like Macrophages in Response to a High Fat Diet. Cells (2019.0) 8. DOI: 10.3390/cells8121587 61. Buhrmann C., Yazdi M., Popper B., Shayan P., Goel A., Aggarwal B.B., Shakibaei M.. Evidence that TNF-β Induces Proliferation in Colorectal Cancer Cells and Resveratrol can down-modulate it. Exp. Biol. Med. (2019.0) 244 1-12. DOI: 10.1177/1535370218824538 62. Baker K.J., Houston A., Brint E.. IL-1 Family Members in Cancer; Two Sides to Every Story. Front. Immunol. (2019.0) 10 1197. DOI: 10.3389/fimmu.2019.01197 63. Tugues S., Burkhard S.H., Ohs I., Vrohlings M., Nussbaum K., Vom Berg J., Kulig P., Becher B.. New Insights into IL-12-Mediated Tumor Suppression. Cell Death Differ. (2015.0) 22 237. DOI: 10.1038/cdd.2014.134 64. Clish C.B.. Metabolomics: An Emerging but Powerful Tool for Precision Medicine. Cold Spring Harb. Mol. Case Stud. (2015.0) 1 a000588. DOI: 10.1101/mcs.a000588 65. Maltais-Payette I., Boulet M.M., Prehn C., Adamski J., Tchernof A.. Circulating Glutamate Concentration as a Biomarker of Visceral Obesity and Associated Metabolic Alterations. Nutr. Metab. (2018.0) 15 78. DOI: 10.1186/s12986-018-0316-5 66. Ren W., Xia Y., Chen S., Wu G., Bazer F.W., Zhou B., Tan B., Zhu G., Deng J., Yin Y.. Glutamine Metabolism in Macrophages: A Novel Target for Obesity/Type 2 Diabetes. Adv. Nutr. (2019.0) 10 321-330. DOI: 10.1093/advances/nmy084 67. Daniele G., Campi B., Saba A., Codini S., Ciccarone A., Giusti L., Del Prato S., Esterline R.L., Ferrannini E.. Plasma N-Acetylaspartate Is Related to Age, Obesity, and Glucose Metabolism: Effects of Antidiabetic Treatment and Bariatric Surgery. Front. Endocrinol. (2020.0) 11 216. DOI: 10.3389/fendo.2020.00216 68. Yamaguchi Y., Yamamoto K., Sato Y., Inoue S., Morinaga T., Hirano E.. Combination of Aspartic Acid and Glutamic Acid Inhibits Tumor Cell Proliferation. Biomed. Res. (2016.0) 37 153-159. DOI: 10.2220/biomedres.37.153 69. Gregoire F.M., Smas C.M., Sul H.S.. Understanding Adipocyte Differentiation. Physiol. Rev. (1998.0) 78 783-809. DOI: 10.1152/physrev.1998.78.3.783 70. Hama K., Fujiwara Y., Hayama T., Ozawa T., Nozawa K., Matsuda K., Hashiguchi Y., Yokoyama K.. Very Long-Chain Fatty Acids Are Accumulated in Triacylglycerol and Nonesterified Forms in Colorectal Cancer Tissues. Sci. Rep. (2021.0) 11 6163. DOI: 10.1038/s41598-021-85603-w 71. Tomin T., Fritz K., Gindlhuber J., Waldherr L., Pucher B., Thallinger G.G., Nomura D.K., Schittmayer M., Birner-Gruenberger R.. Deletion of Adipose Triglyceride Lipase Links Triacylglycerol Accumulation to a More-Aggressive Phenotype in A549 Lung Carcinoma Cells. J. Proteome Res. (2018.0) 17 1415-1425. DOI: 10.1021/acs.jproteome.7b00782 72. Palomo I., Contreras A., Alarcón L.M., Leiva E., Guzmán L., Mujica V., Icaza G., Díaz N., González D.R., Moore-Carrasco R.. Elevated Concentration of Asymmetric Dimethylarginine (ADMA) in Individuals with Metabolic Syndrome. Nitric Oxide (2011.0) 24 224-228. DOI: 10.1016/j.niox.2011.03.002 73. Hajer G.R., Van Der Graaf Y., Olijhoek J.K., Verhaar M.C., Visseren F.L.J.. Levels of Homocysteine Are Increased in Metabolic Syndrome Patients but Are Not Associated with an Increased Cardiovascular Risk, in Contrast to Patients without the Metabolic Syndrome. Heart (2007.0) 93 216. DOI: 10.1136/hrt.2006.093971 74. Furuhashi M.. New Insights into Purine Metabolism in Metabolic Diseases: Role of Xanthine Oxidoreductase Activity. Am. J. Physiol.—Endocrinol. Metab. (2020.0) 319 E827-E834. DOI: 10.1152/ajpendo.00378.2020 75. Wu L.L., Wu J.T.. Hyperhomocysteinemia Is a Risk Factor for Cancer and a New Potential Tumor Marker. Clin. Chim. Acta (2002.0) 322 21-28. DOI: 10.1016/S0009-8981(02)00174-2 76. Park J., Shin Y., Kim T.H., Kim D.H., Lee A.. Plasma Metabolites as Possible Biomarkers for Diagnosis of Breast Cancer. PLoS ONE (2019.0) 14. DOI: 10.1371/journal.pone.0225129 77. Li H., Zhou Y., Zhao A., Qiu Y., Xie G., Jiang Q., Zheng X., Zhong W., Sun X., Zhou Z.. Asymmetric Dimethylarginine Attenuates Serum Starvation-Induced Apoptosis via Suppression of the Fas (APO-1/CD95)/JNK (SAPK) Pathway. Cell Death Dis. (2013.0) 4 e830. DOI: 10.1038/cddis.2013.345 78. Jain M., Nilsson R., Sharma S., Madhusudhan N., Kitami T., Souza A.L., Kafri R., Kirschner M.W., Clish C.B., Mootha V.K.. Metabolite Profiling Identifies a Key Role for Glycine in Rapid Cancer Cell Proliferation. Science (2012.0) 336 1040-1044. DOI: 10.1126/science.1218595 79. Baliou S., Kyriakopoulos A.M., Spandidos D.A., Zoumpourlis V.. Role of Taurine, its Haloamines and its LncRNA TUG1 in Both Inflammation and Cancer Progression. On the Road to Therapeutics? (Review). Int. J. Oncol. (2020.0) 57 631-664. DOI: 10.3892/ijo.2020.5100 80. Wu T.K., Wei C.W., Pan Y.R., Hsu R.J., Wu C.Y., Yu Y.L.. The Uremic Toxin P-Cresyl Sulfate Induces Proliferation and Migration of Clear Cell Renal Cell Carcinoma via MicroRNA-21/ HIF-1α Axis Signals. Sci. Rep. (2019.0) 9 3207. DOI: 10.1038/s41598-019-39646-9 81. Peng Y.S., Syu J.P., Wang S.D., Pan P.C., Kung H.N.. BSA-Bounded p-Cresyl Sulfate Potentiates the Malignancy of Bladder Carcinoma by Triggering Cell Migration and EMT through the ROS/Src/FAK Signaling Pathway. Cell Biol. Toxicol. (2020.0) 36 287-300. DOI: 10.1007/s10565-019-09509-0 82. Klindworth A., Pruesse E., Schweer T., Peplies J., Quast C., Horn M., Glöckner F.O.. Evaluation of General 16S Ribosomal RNA Gene PCR Primers for Classical and Next-Generation Sequencing-Based Diversity Studies. Nucleic Acids Res. (2013.0) 41 e1. DOI: 10.1093/nar/gks808 83. Tian W., Yao Y., Fan G., Zhou Y., Wu M., Xu D., Deng Y.. Changes in Lipid Profiles during and after (Neo)Adjuvant Chemotherapy in Women with Early-Stage Breast Cancer: A Retrospective Study. PLoS ONE (2019.0) 14. DOI: 10.1371/journal.pone.0221866 84. Morad S.A.F., Cabot M.C.. Ceramide-Orchestrated Signalling in Cancer Cells. Nat. Rev. Cancer (2012.0) 13 51-65. DOI: 10.1038/nrc3398 85. Moro K., Nagahashi M., Gabriel E., Takabe K., Wakai T.. Clinical Application of Ceramide in Cancer Treatment. Breast Cancer (2019.0) 26 407. DOI: 10.1007/s12282-019-00953-8 86. Chen X.Q., Wu P.W., Liu D.H., Yan S.J., Shen X.M., Yang L.Y.. Prognostic Significance of High Triglyceride and Apolipoprotein B Levels in Patients with Stage III and High-Risk Stage II Colorectal Cancer Undergoing Curative Surgery. Oncol. Lett. (2020.0) 20 705. DOI: 10.3892/ol.2020.11617 87. Saito R.D.F., Andrade L.N.. de S.; Bustos, S.O.; Chammas, R. Phosphatidylcholine-Derived Lipid Mediators: The Crosstalk Between Cancer Cells and Immune Cells. Front. Immunol. (2022.0) 13 215. DOI: 10.3389/fimmu.2022.768606 88. Rettig T.C.D., Verwijmeren L., Dijkstra I.M., Boerma D., Van De Garde E.M.W., Noordzij P.G.. Postoperative Interleukin-6 Level and Early Detection of Complications After Elective Major Abdominal Surgery. Ann. Surg. (2016.0) 263 1207-1212. DOI: 10.1097/SLA.0000000000001342 89. Belizon A., Balik E., Feingold D.L., Bessler M., Arnell T.D., Forde K.A., Horst P.K., Jain S., Cekic V., Kirman I.. Major Abdominal Surgery Increases Plasma Levels of Vascular Endothelial Growth Factor: Open More So Than Minimally Invasive Methods. Ann. Surg. (2006.0) 244 792. DOI: 10.1097/01.sla.0000225272.52313.e2 90. Yang C., Bork U., Schölch S., Kulu Y., Kaderali L., Bolstorff U.L., Kahlert C., Weitz J., Rahbari N.N., Reissfelder C.. Postoperative Course and Prognostic Value of Circulating Angiogenic Cytokines after Pancreatic Cancer Resection. Oncotarget (2017.0) 8 72315. DOI: 10.18632/oncotarget.20315 91. El-Ghonaimy E.A., El-Shinawi M., Ibrahim S.A., El-Ghazaly H., Abd-El-Tawab R., Nouh M.A., El-Mamlouk T., Mohamed M.M.. Positive Lymph-Node Breast Cancer Patients—Activation of NF-ΚB in Tumor-Associated Leukocytes Stimulates Cytokine Secretion That Promotes Metastasis via C-C Chemokine Receptor CCR7. FEBS J. (2015.0) 282 271-282. DOI: 10.1111/febs.13124 92. Davern M., Donlon N.E., O’Connell F., Gaughan C., O’Donovan C., Habash M., Sheppard A.D., MacLean M., Dunne M.R., Moore J.. Acidosis Significantly Alters Immune Checkpoint Expression Profiles of T Cells from Oesophageal Adenocarcinoma Patients. Cancer Immunol. Immunother. (2023.0) 72 55-71. DOI: 10.1007/s00262-022-03228-y 93. Wang Y.. Cancer-Generated Lactic Acid: A Regulatory, Immunosuppressive Metabolite?. J. Pathol. (2013.0) 230 350. DOI: 10.1002/PATH.4218 94. Aoki S., Inoue K., Klein S., Halvorsen S., Chen J., Matsui A., Nikmaneshi M.R., Kitahara S., Hato T., Chen X.. Placental Growth Factor Promotes Tumour Desmoplasia and Treatment Resistance in Intrahepatic Cholangiocarcinoma. Gut (2022.0) 71 185-193. DOI: 10.1136/gutjnl-2020-322493
--- title: Minocycline Attenuates Sevoflurane-Induced Postoperative Cognitive Dysfunction in Aged Mice by Suppressing Hippocampal Apoptosis and the Notch Signaling Pathway-Mediated Neuroinflammation authors: - Junjie Liang - Shanshan Han - Chao Ye - Haimeng Zhu - Jiajun Wu - Yunjuan Nie - Gaoshang Chai - Peng Zhao - Dengxin Zhang journal: Brain Sciences year: 2023 pmcid: PMC10046414 doi: 10.3390/brainsci13030512 license: CC BY 4.0 --- # Minocycline Attenuates Sevoflurane-Induced Postoperative Cognitive Dysfunction in Aged Mice by Suppressing Hippocampal Apoptosis and the Notch Signaling Pathway-Mediated Neuroinflammation ## Abstract Postoperative cognitive dysfunction (POCD), an important postoperative neurological complication, is very common and has an elevated incidence in elderly patients. Sevoflurane, an inhaled anesthetic, has been demonstrated to be associated with POCD in both clinical and animal studies. However, how to prevent POCD remains unclear. Minocycline, a commonly used antibiotic can cross the blood-brain barrier and exert an inhibitory effect on inflammation in the central nervous system. The present work aimed to examine the protective effect and mechanism of minocycline on sevoflurane-induced POCD in aged mice. We found that $3\%$ sevoflurane administered 2 h a day for 3 consecutive days led to cognitive impairment in aged animals. Further investigation revealed that sevoflurane impaired synapse plasticity by causing apoptosis and neuroinflammation and thus induced cognitive dysfunction. However, minocycline pretreatment (50 mg/kg, i.p, 1 h prior to sevoflurane exposure) significantly attenuated learning and memory impairments associated with sevoflurane in aged animals by suppressing apoptosis and neuroinflammation. Moreover, a mechanistic analysis showed that minocycline suppressed sevoflurane-triggered neuroinflammation by inhibiting Notch signaling. Similar results were also obtained in vitro. Collectively, these findings suggested minocycline may be an effective drug for the prevention of sevoflurane-induced POCD in elderly patients. ## 1. Introduction As society ages, more than $50\%$ of all elderly individuals are estimated to undergo at least one surgical procedure [1]. Thus, a growing number of studies are focusing on the effects of surgery and/or anesthesia on brain function in aged patients. Postoperative delirium (POD) is an acute neurological complication characterized by decreased environmental awareness and attention deficit that occurs early after surgery and affects nearly $60\%$ of elderly individuals [2]. Postoperative cognitive dysfunction (POCD), featuring memory deficit, reduced capability of processing information, and anxiety, is a further exacerbation of POD [3]. Sevoflurane, a frequently utilized inhalation anesthetic in clinics, has been shown to be strongly associated with POCD. Chai et al. suggested that sevoflurane enhances the expression of acidic leucine-rich nuclear phosphoprotein-32A (ANP32A, the key component of inhibitors of acetyltransferases) by inducing transcription factor CCAAT/enhancer binding protein beta (C/EBPβ), which may inhibit histone acetylation and promote cognitive impairment in aged mice [4]. Yu and colleagues demonstrated that repeated sevoflurane administration increases the concentrations of NUAK family SNF1-like kinase 1 (Nuak1, an AMPK-related kinase) in neonatal mice, which could induce brain Tau phosphorylation and lead to cognitive dysfunction [5]. Wang and collaborators found that sevoflurane inhalation might induce cognitive deficit in young mice via autophagy induction in the hippocampus [6]. Many studies have examined sevoflurane-induced POCD, demonstrating that sevoflurane causes POCD through a variety of mechanisms in mice of different ages. However, the exact underpinning mechanism remains undefined, making it a hot topic for many anesthesiologists. Inflammation is the initial stage in multiple pathological processes. Inflammation in the hippocampus, the major brain area related to learning and memory, negatively affects cognitive ability. Sevoflurane has been shown to cause central nervous system inflammation. Qin et al. found that sevoflurane reduces immune response in young rats [7]. Dong et al. demonstrated that sevoflurane enhances the generation of the proinflammatory cytokine IL-6 as well as cognitive dysfunction by inducing Tau migration from neurons to microglia [8]. Li et al. suggested that sevoflurane induces NLRP3 inflammasome formation and reduces brain-derived neurotrophic factor (BDNF) production to promote neurodegeneration and neuroinflammation [9]. Gui et al. found that long-term inhalation of sevoflurane and surgery could induce neuroinflammation, down-regulate glial cell-derived neurotrophic factor (GDNF), and reduce the level of neurogenesis in the hippocampus, which contributed to cognitive dysfunction in neonatal rats [10]. Notch is a conserved pathway controlling cell fate. Previously reported findings indicated that Notch signaling is tightly associated with tumor development [11]. Recently, it was demonstrated that aberrant Notch pathway induction is strongly associated with inflammatory events, e.g., allergic airway inflammation [12] and renal inflammation [13]. However, whether the Notch pathway participates in hippocampal inflammation induced by sevoflurane remains unknown. Minocycline, a second-generation tetracycline belonging to broad-spectrum antibacterial drugs, has high effective tissue permeability and antibacterial activity. Minocycline has been reported to exert a variety of pharmacological effects that are different from its antimicrobial activity, e.g., anti-inflammatory and antiapoptotic activities, in the central nervous system [14,15]. Tian et al. found that minocycline plays a protective role in hippocampal neurons of aged rats by alleviating sevoflurane-induced apoptotic death, inflammatory response, and beta amyloid (Aβ) production. [ 16]. Recently, Yang et al. found minocycline ameliorates diabetic neuropathic pain (DNP) by downregulating Notch and inactivating Notch signaling in microglia in the spinal cord [17]. However, whether minocycline attenuates sevoflurane-induced neuroinflammation by modulating Notch signaling remains unknown. Based on the above findings, we hypothesized that the Notch signaling pathway plays a critical role in sevoflurane-induced neuroinflammation, and minocycline may alleviate such inflammation by inhibiting excessive activation of Notch signaling, thereby improving cognitive impairment. Consequently, the present work aimed to examine the role of Notch signaling in sevoflurane-associated neuroinflammation and to determine the therapeutic impact of minocycline. ## 2.1. Animals Sixty-three specific pathogen-free (SPF) male C57BL/6J mice aged eighteen months (Changzhou Cavens Experimental Animal Co., Ltd., Changzhou, China) were utilized throughout this study. Mouse housing was carried out in a specific pathogen-free environment with a 12-h photoperiod and full access to food and water. The experimental protocol had approval from the Animal Ethics Committee of Jiangnan University (JN. No. 20201230c1401231[379]). We used as few animals as possible for experiments and minimized the severity of their suffering. After one week of adaptation, mice were randomized into the control (Con), sevoflurane (Sev), and sevoflurane+minocycline (Sev + Min) groups, with 21 animals/group. In each group, 10 mice were used for the Morris water maze and fear conditioning test, 8 were used for immunofluorescence, and 3 were used for LTP. After the behavior tests, anesthesia was carried out with $2\%$ sodium pentobarbital solution at 80 mg/kg and the animals were sacrificed. The hippocampal tissue was isolated at 4 °C and kept at −80 °C for subsequent assays. ## 2.2. Minocycline Treatment and Sevoflurane Exposure The sevoflurane anesthesia procedure was carried out as previously reported [4]. In brief, mice were placed into a transparent airtight box and administered $95\%$ oxygen and $3\%$ sevoflurane (Shanghai Hengrui Pharmaceutical, Shanghai, China) by inhalation at a flow rate of 1–2 L/min on an anesthesia machine (R620-S1, RWD Life Technology, Shenzhen, China) for two hours daily for three consecutive days, while control mice received $95\%$ oxygen under the same environmental conditions. Animal body temperature was kept at 38 °C throughout the entire anesthesia procedure using a heating blanket. After the end of the sevoflurane anesthesia, mice were transferred to a new cage maintained at room temperature and ambient air until free movements occurred. Hydrochloride minocycline was obtained from MCE (MedChemExpress, Monmouth Junction, NJ, USA) and dissolved in saline. One hour before sevoflurane exposure, minocycline was administered by intraperitoneal injection in mice at 50 mg/kg according to a previous study [18], the animal in the Con and Sev groups were intraperitoneally administered equivalent amounts of saline. ## 2.3. Cell Culture and Sevoflurane Treatment BV2 microglia and HT-22 cells were utilized as representative microglial cells and hippocampal neurons, respectively. Mouse hippocampal HT-22 cells, as a cell model, were obtained from mouse hippocampus and have been widely used in many neural system disease studies. BV2 microglia cells were obtained from immortalized mouse microglial cell lines showing inflammation and phagocytic feature upon induction [19]. Both HT-22 and BV2 cells were provided by Procell Life Science & Technology (Wuhan, China). Cell culture was performed at 37 °C in a humid environment containing $5\%$ CO2 in DMEM (Cytiva, Shanghai, China) with antibiotics (penicillin [100 units/mL] and streptomycin [100 g/mL], Beyotime, Shanghai, China) and $10\%$ FBS (BIOEXPLORER, Carolina, USA). The sevoflurane treatment protocol was described previously [20]. After 48 h, cells in culture dishes were randomized into the control, sevoflurane, minocycline (50 μM), and sevoflurane plus minocycline (1, 10, and 50 μM) groups. Culture dishes underwent incubation in an anesthesia induction chamber at 37 °C (RWD Life Science, Shenzhen, China) with a freshly prepared gas mixture ($21\%$ O2, $5\%$ CO2, and $69\%$ N2) or $3\%$ sevoflurane for 6 h. Minocycline was added to the medium 1 h before sevoflurane exposure. ## 2.4. Bromodeoxyuridine (BrdU)-Labeling BrdU was obtained from Sigma-Aldrich Trading (China) and intraperitoneally injected into mice at the beginning and end of sevoflurane inhalation in each group while establishing the sevoflurane anesthesia model. The BrdU powder was dissolved in 37 °C saline, and the injection dose was 100 mg/kg b.i.d for three days. ## 2.5. Morris Water Maze (MWM) Test Thirty mice ($$n = 10$$) were assessed for learning and memory function after the final day of sevoflurane exposure. The MWM experimental method was described previously [21]. Briefly, the MWM system consisted of a circular tank filled with water that appeared white with milk powder, and the water temperature was kept at 19–20 °C. The tank was evenly split into four quadrants, of which one contained the platform placed 1 cm below the water surface. After receiving sevoflurane anesthesia, mice underwent training with four trials daily for five days. Two consecutive trials were separated by a minimum time interval of 15 min. To perform a trial, the animals were placed in four distinct quadrants of the water maze sequentially and allowed 60 s to freely find the hidden platform; otherwise, they were directed toward the platform and forced to remain on it for 30 s. On day 6, mice underwent the spatial probe test, where they were tested without the platform present. Escape latency, the target time ratio, the swimming track, and the platform crossing times were recorded using a WMT-100 Morris water maze video tracking system. ## 2.6. Fear Conditioning Assay The fear conditioning assay procedure was described previously [22]. Briefly, after sevoflurane or air exposure, the animals were positioned in a box containing a stainless-steel grid floor for foot shock and allowed to adapt to the environment for 3 min. In cued fear conditioning (day 1 after sevoflurane or air exposure), the mice were submitted to a sound stimulation at 70 dB for 30 s, followed by an electrical stimulation at 1 mA for 2 s, a cycle that was repeated 3 times. During this period, the percentage of freezing time in mice was determined. In the cued fear memory test (5 days after cued fear conditioning), the animals were positioned in a box for 3 min, followed by exposure to only 70 dB sound stimulation for 30 s three times. Memory in the cued fear assay was determined as the percentage of time that the animals remained frozen over the course of the three trials (Figure 1F). ## 2.7. Brain Immunofluorescence Brains underwent fixation with $4\%$ paraformaldehyde overnight, followed by transfer to $20\%$ and $30\%$sucrose in PBS for 2 days and 3 days, respectively. After embedding in the Tissue-Tek OCT compound (Sakura, Tokyo, Japan), brain sections underwent 25-µm sectioning using a vibratome slicer (Leica, Nussloch, Germany). These sections underwent permeabilization with $0.5\%$ Triton X-100 dissolved in PBS for 30 min at ambient temperature to dissolve the phospholipid membrane and blocking with $5\%$ bovine serum albumin (BSA) for 1 h to block nonspecific sites. Incubation with primary antibodies was performed at 4 °C overnight. Following three PBS washes, subsequent incubation was carried out with secondary antibodies for 1 h at 37 °C. A microscope (Carl Zeiss, Freistaat Thüringen, Germany) was utilized for imaging. ImageJ (version 1.53m) was employed to assess positive cells in the hippocampal DG region of a given section, and then the numbers of positive cells in all sections from the same mouse were added for statistical analysis. Table 1 lists all primary antibodies applied in immunofluorescence. ## 2.8. Cell Immunofluorescence Cells (5 × 104 cells/well) seeded in 24-well dished with cover slips underwent overnight incubation and exposure to minocycline (1, 10, and 50 μM) 1 h before incubation with $3\%$ sevoflurane for 6 h. Next, a 10-min fixation was carried out with $4\%$ paraformaldehyde, and cells were permeabilized with $0.5\%$ Triton X-100 in PBS. Blocking ($5\%$ BSA, 1 h) was followed by successive incubations with primary (overnight, 4 °C) and secondary (2 h, ambient). Finally, DAPI (2.5 μg/mL) counterstaining was carried out before analysis under a fluorescence microscope (Carl Zeiss). Table 1 lists all primary antibodies applied for immunofluorescence. ## 2.9. TUNEL-Based Cell Apoptosis Analysis Cells (1 × 104/well) seeded in 96-well plates with cover slips underwent overnight incubation and treatment with minocycline (1, 10, and 50 μM) for 1 h before induction with $3\%$ sevoflurane for 6 h. According to the TUNEL FITC detection apoptosis kit (Vazyme, Nanjing, China) protocol, cells underwent fixation with $4\%$ formaldehyde (25 min) and permeabilization with $0.2\%$ Triton X-100 PBS (5 min). Next, 1× equilibration buffer was equilibrated for 10–30 min at ambient. This was followed by a 1 h with TdT incubation buffer at 37 °C. DAPI (2.5 μg/ML) counterstaining was carried out, followed by fluorescence microscopy. Cells can be divided into two groups using this technique: living (extremely low background fluorescence) and apoptotic (strong green, fluorescent signals) cells. ## 2.10. Western Blotting Hippocampal tissue specimens or cells underwent lysis with RIPA buffer (Beyotime, Shanghai, China) plus 1 mmol/L PMSF. Proteins separation utilized $12\%$ SDS-PAGE gels (Vazyme, Nanjing, China) for approximately 1 h, followed by transfer onto nitrocellulose membranes. After a 1 h blocking (with $5\%$ BSA in Tris-buffered saline Tween-20 (TBST), the samples were submitted to successive incubation) with primary (overnight, 4 °C) and secondary (1 h, ambient) antibodies. Finally, these bands were detected on a Tanon-2500B chemiluminescence imager (Shanghai Tanon Technology, Shanghai, China), and ImageJ was utilized for densiometric quantitation. The target protein’s expression was standardized to total β-actin levels. Table 2 lists all primary antibodies utilized for Western blotting. ## 2.11. Reverse Transcription and Real-Time Quantitative PCR Total RNA extraction was carried out from BV2 microglia cells using with TRIzol reagent. Reverse transcription employed Hifair® III 1st Strand cDNA Synthesis SuperMix for qPCR (Yeasen, Shanghai, China) with primer sets for iNOS, IL-1β, and GAPDH. 2−△△Ct method was utilized for the analysis of data after GAPDH normalization. Table 3 lists the primers utilized for PCR. ## 2.12. Electrophysiological Analysis Mice were anesthetized by intraperitoneally injecting $2\%$ sodium pentobarbital at 80 mg/kg and then sacrificed. The brain was obtained after decapitation and placed in artificial cerebrospinal fluid (ACSF) saturated with mixed gases ($95\%$ O2 and $5\%$ CO2) at 0~4 °C. After cooling the brain tissue to 0~4 °C, we dissected the hippocampus, which was sliced along the sagittal plane with a 400 µm thick vibrating slicer. Then, hippocampal sections were incubated with ACSF saturated with the above gas mixture at 32.0 ± 0.5 °C for 2–3 h. During the experiment, a single brain slice was immersed in ACSF saturated with the gas mixture (4 mL/min, 30 °C) and placed on an 8 × 8 array of the MED system (microelectrode size, 50 mm × 50 mm; interelectrode distance, 450 mm), and the voltage signal was obtained by the MED64 system. Through stimulation of perforating fibers, the slope changes of the field excited synaptic potential (fEPSP) were recorded from the hippocampal DG area, and $30\%$ of the stimulus intensity that could induce the maximum slope of the fEPSP was selected as the basic stimulus intensity. In the case of a stable baseline, high-frequency stimulation (100 Hz, 200 pulses) was performed to induce long-term potential (LTP). An increase in fEPSP slope over $20\%$ of the basal level was employed as a criterion for induction success and was maintained for at least 30 min. In the experiment, fEPSP was recorded and observed for at least 120 min. ## 2.13. Statistical Analysis Data are mean ± SEM, and were assessed with SPSS 15.0 (SPSS, Chicago, IL, USA). Two-way analysis of variance (ANOVA) with post hoc Bonferroni’s test was carried out to assess the differences in escape latency measured during the place navigation trials. The remaining experimental data in each group were compared by one-way ANOVA with post hoc Bonferroni’s test. In all experiments, $p \leq 0.05$ indicated a statistically significant difference. ## 3.1. Minocycline Alleviates Sevoflurane-Induced Learning and Memory Impairments In order to examine the role of minocycline in sevoflurane-associated postoperative cognitive dysfunction, we established a sevoflurane-induced POCD model via inhalation of $3\%$ sevoflurane for 2 h daily for three days. During place navigation trials of the Morris water maze experiment, escape latency was markedly elevated in the Sev group compared with the Con and Sev + Min groups (Figure 1A). On the 6th day (spatial probe test), the data showed increased escape latency, fewer crossings of the initial platform, and decreased target quadrant time ratio, and the swimming path was disordered in the Sev group, which showed significant differences in comparison with control animals. However, mice intraperitoneally administered minocycline before sevoflurane exposure had a certain improvement in the number of original platform crossings, escape latency, target quadrant time ratio, and swimming path, with marked differences in comparison with the Sev group (Figure 1B–E). A similar cognitive impairment was found in the fear conditioning test. No significant differences in freezing percentage in each group were found in the cued fear conditioning test; however, the freezing level was decreased in the Sev group compared with the control and Sev + Min groups in the cued fear memory test (Figure 1G,H). The above results suggested that multiple sevoflurane exposures impaired cognitive function in aged mice, and minocycline pretreatment suppressed learning and memory impairments associated with sevoflurane. Figure 1: Minocycline alleviates learning and memory impairments associated with sevoflurane: (A) Escape latency (time required to find the platform) during the MWM training on days 1–5; (B) Escape latency in the space exploration experiment; (C) Platform crossing times during the space exploration test; (D) Target time ratio during the space exploration experiment; (E) Typical swimming track in each group; (F) Schematic diagram of the fear conditioning test; (G) Freezing percentage on day 1; (H) Freezing percentage on day 6. Data are mean ± SEM (n = 10). ns, no significance; p < 0.05, p < 0.01, * p < 0.001.* ## 3.2. Minocycline Attenuates Sevoflurane-Induced Synaptic Plasticity Impairment Synaptic plasticity is the basis of learning and memory; however, sevoflurane could impair synaptic plasticity by a variety of mechanisms [4,23]. To assess minocycline’s ameliorative effect on sevoflurane-associated synaptic plasticity impairment, synaptic plasticity was examined in acute brain slices by electrophysiological analysis (Figure 2A). The average fEPSP slope increased to 156.54 ± $2.86\%$ (control group) and 157.22 ± $1.83\%$ (sevoflurane + minocycline group) immediately after high-frequency stimulation (HFS) from baseline ($100\%$) and was maintained, indicating long-term potentiation (LTP) was successfully induced in the hippocampus. Meanwhile, the fEPSP slope was much lower in the sevoflurane group (123.8 ± $2.67\%$, Figure 2B,C). To explore the possible reasons for the above changes, synapse-associated proteins were detected in hippocampal samples and HT-22 cells. We observed that postsynaptic density protein 95 (PSD95) and synapsin 1 (Syn-1) were significantly downregulated after sevoflurane exposure in both hippocampal extracts and HT-22 cells according to Western blotting data; nevertheless, minocycline pretreatment significantly enhanced PSD95 and Syn-1 protein expression in the hippocampal region and HT-22 cells (Figure 2D–G). These findings suggested minocycline attenuated sevoflurane-related impairment of synaptic plasticity. ## 3.3. Minocycline Alleviates Sevoflurane-Induced Neurogenesis Dysfunction Neurogenesis is a process involving neural stem cells transforming into mature neurons and connecting with original neurons after proliferation, migration, and formation [24]. Thus, neurogenesis has a strong association with synaptic plasticity. Emerging evidence indicates sevoflurane has an inhibitory effect on hippocampal neurogenesis [25,26]. To confirm this notion and examine the therapeutic effect of minocycline, we used the nucleotide analog BrdU to label neural stem cells to observe the effects of sevoflurane on neurogenesis. As shown in Figure 3A,C, 24 h after sevoflurane exposure, BrdU-labeled cells were markedly decreased in the sevoflurane group in comparison with the control group, while cells pretreated with minocycline had remarkably increased BrdU-positive cells in comparison with the sevoflurane group. These findings suggested that $3\%$ sevoflurane could lead to neural stem cell proliferative dysfunction and that minocycline alleviated this impairment. Twenty-one days after sevoflurane exposure, BrdU and NeuN co-staining was carried out to observe neural stem cell formation. In comparison with control cells, sevoflurane exposure starkly reduced the amounts of BrdU- and NeuN-positive cells in the hippocampal dentate gyrus; meanwhile, minocycline pretreatment markedly elevated the amounts of BrdU- and NeuN-labeled cells (Figure 3B,D). The above results suggested that minocycline alleviated sevoflurane-induced neurogenesis dysfunction. ## 3.4. Minocycline Suppresses Hippocampal Apoptosis Induced by Sevoflurane Apoptosis is critical for multiple neurodegenerative processes [27]. Studies have suggested that sevoflurane can induce neuronal apoptosis [28,29]. To confirm this notion and examine the therapeutic effect of minocycline, we conducted TUNEL and immunoblot experiments. As shown in Figure 4A, in comparison with the control and minocycline groups, which were exposed to only $95\%$ O2 and $5\%$ CO2, the sevoflurane group had starkly elevated HT-22 cell apoptosis; however, upon minocycline pretreatment, the number of apoptotic HT-22 cells after sevoflurane exposure was reduced in comparison with the sevoflurane group, especially after treatment with high concentrations of minocycline (50 μM). Subsequently, apoptosis-associated proteins were detected. Similarly, in HT-22 cells, sevoflurane significantly upregulated proapoptotic proteins, including cleaved caspase 3, caspase 3, and Bax, and downregulated the antiapoptotic protein Bcl-2 compared with the control and minocycline groups. Treatment with moderate and high concentrations (10 μM and 50 μM) of minocycline before sevoflurane exposure significantly reversed the above changes in apoptosis-associated proteins (Figure 4B,D,F). Similar changes were also observed in hippocampal tissue specimens (Figure 4C,E,G). The above results suggested that sevoflurane exposure could induce hippocampal apoptosis and that minocycline suppressed hippocampal apoptosis induced by sevoflurane. ## 3.5. Minocycline Suppresses Sevoflurane-Associated Microglial Activation to the M1 Stage and Reduces Proinflammatory Cytokine Production Increasing evidence suggests that minocycline is the most effective tetracycline derivative for neuroprotection [30]. Due to the neuroprotective effect of minocycline, we investigated whether it exerts a neuroprotective effect on sevoflurane-induced neuroinflammation. CD68 is a biomarker of the activation (proinflammatory) state in microglia. As shown in Figure 5A, sevoflurane significantly upregulated CD68 in BV2 cells, with significantly brighter red fluorescence, compared with the control and minocycline groups. Nevertheless, minocycline pretreatment before sevoflurane exposure significantly downregulated CD68 expression in BV2, and the red fluorescence was significantly attenuated, especially in the groups pretreated with the middle and high concentrations of minocycline. Similar results were observed in Western blot experiments (Figure 5B,C). Then, real-time quantitative PCR was carried out to examine the mRNA expression of proinflammatory cytokines. Unsurprisingly, sevoflurane exposure significantly increased the mRNA amounts of M1-type cytokines (iNOS and IL-1β) in comparison with the control and minocycline groups, while minocycline pretreatment before sevoflurane exposure markedly decreased iNOS and IL-1β mRNA amounts, particularly in the groups pretreated with the middle and high concentrations of minocycline (Figure 5D,E). The above results indicated that sevoflurane exposure could activate microglia, leading to neuroinflammation, while minocycline exerted a neuroprotective effect by suppressing sevoflurane-induced microglial activation to the M1 stage and reducing proinflammatory cytokine production. ## 3.6. Minocycline Alleviates Sevoflurane-Induced Neuroinflammation via Notch Signaling Suppression According to the above data, we hypothesized that sevoflurane-induced neuroinflammation plays a dominant role in the observed changes. Thus, it is particularly important to unveil the underpinning molecular mechanisms behind sevoflurane-associated neuroinflammation. Recent studies have consistently suggested that the Notch signaling pathway has a strong association with inflammation [31,32]. To unveil the underlying molecular mechanisms of sevoflurane-induced neuroinflammation, immunofluorescence, and Western blot experiments were performed. As shown in Figure 6A, sevoflurane significantly upregulated Notch pathway-associated proteins in BV2 cells, including Notch1, cleaved Notch1, and Hes1, with significantly brighter red fluorescent signals compared with the control and minocycline groups. Nevertheless, minocycline pretreatment before sevoflurane exposure significantly downregulated Notch1, cleaved Notch1, and Hes1 in BV2 cells, and red fluorescent signals were significantly attenuated, especially in the groups pretreated with the middle and high concentrations of minocycline. Furthermore, Notch1, cleaved Notch1, and Hes1 protein amounts were dramatically decreased in hippocampal specimens and BV2 cells from the sevoflurane group according to Western blotting. Notably, minocycline treatment before sevoflurane exposure significantly downregulated Notch1, cleaved Notch1, and Hes1 proteins in the hippocampal region as well as BV2 cells (Figure 6B–G). These findings suggested that minocycline alleviated sevoflurane-associated neuroinflammation through inhibition of the Notch signaling pathway. ## 4. Discussion In this work, we examined whether minocycline injection attenuates sevoflurane-associated neuroapoptosis and neuroinflammation in aged mice to alleviate cognitive dysfunction and determined the role of Notch signaling in minocycline suppression of sevoflurane-associated neuroinflammation. As expected, we found that exposure to $3\%$ sevoflurane for 2 h a day for 3 consecutive days induced cognitive dysfunction in aged mice. Further investigation revealed that sevoflurane caused learning and memory impairments by inducing apoptosis, neuroinflammation, neurogenic dysfunction, and synaptic plasticity impairment. Minocycline significantly alleviated sevoflurane-induced cognitive dysfunction by exerting anti-inflammatory and antiapoptotic effects, which involved Notch signaling in aged mice. As early as 1955, Professor Bedford first reported the clinical phenomenon that patients show such features as reduced interest in family, forgetting things, and indifference to the outside world after surgery and anesthesia. He defined them as “adverse brain effects of anesthesia in the elderly” [33]. As a common postoperative neurological complication, POCD brings many adverse effects to patients. According to a study performed in the United States in 2020, cognitive dysfunction occurring within one year after surgery adds approximately USD 17,275 to the patient’s health care expenditures [34]. In addition, POCD increases the incidence rates of other complications and prolongs hospitalization, but also carries a risk of developing into dementia [35]. Indeed, POCD has always had a clinically significant prevalence. A high incidence of cognitive impairment is observed in common surgical procedures in the elderly such as coronary artery bypass graft surgery and total hip arthroplasty. A study performed in the year 2019 showed that one week after coronary artery bypass graft surgery, $71\%$ of patients experienced a decline in cognitive function and three months post-surgery, about $47\%$ of cases still experienced changes in cognitive behavior [36]. A review showed that in individuals administered elective hip or knee arthroplasty, POCD had incidence rates of $19.3\%$ and $10\%$ at one and three months after surgery, respectively [37]. Anesthetic administration may represent a major risk factor for POCD. In 1997, a clinical trial by Galinkin et al. demonstrated that sevoflurane can decrease learning and memory function [38]. Since then, multiple reports have examined the association of sevoflurane with cognition. With the deepening of research, the mechanism of sevoflurane-induced cognitive impairment is constantly being revealed. Liang et al. found that sevoflurane induces learning and memory deficits in aged mice by reducing the plasma Aβ1-40 concentration and upregulating RAGE at both the transcriptional and translational levels in the brain [39]. Dysregulated apoptosis contributes to multiple human disorders, e.g., neurodegenerative disorders, and multiple studies have revealed that inappropriate apoptosis is strongly related to Alzheimer’s disease (AD) [40], Parkinson’s disease (PD) [41], and amyotrophic lateral sclerosis (ALS) [42]. Chen et al. found that neuronal apoptosis mediated by endoplasmic reticulum (ER) stress might be involved in memory impairment associated with sevoflurane in aging rats [43], which is consistent with our findings. As demonstrated above, sevoflurane triggered apoptosis, upregulating apoptosis-associated proteins (cleaved caspase 3, caspase 3, and Bax) and downregulating antiapoptotic protein Bcl-2. Shen and collaborators demonstrated that sevoflurane induces neuroinflammation in young mice but not in adult mice by increasing calcium amounts to upregulate TNF-α and IL-6 through the nuclear factor-κB pathway [44]. Similarly, Dong et al. showed that sevoflurane causes cognitive decline by increasing microglia-regulated neuroinflammatory reactions in a rat model by decreasing PPAR-γ activity in the hippocampus [45], indicating that neuroinflammation is an important mechanism of sevoflurane-induced cognitive impairment. Notch signaling is expressed in the majority of cells. With the help of ADAM metalloprotease and the γ-secretase complex, Notch receptor cleavage occurs, and the Notch intracellular domain (cleaved Notch1, an activated fragment) is released. Then, cleaved Notch1 is translocated to the nuclear compartment and binds to CSL (RBP-J) for transcriptional regulation of downstream genes such as members of the HES family [46,47]. Recent studies have consistently suggested that Notch signaling has a strong relationship with inflammation. Qian et al. found that Notch signaling pathway activation is critical for the differentiation of A1 astrocytes after spinal cord injury (SCI) [31]. Similarly, Wu et al. found that Notch signaling regulates the activation of microglia and inflammatory responses in rats with experimental temporal lobe epilepsy [32]. In the current study, we found that sevoflurane polarized microglia to a proinflammatory state (M1 state) by enhancing Notch signaling, producing inflammatory cytokines, and causing neuroinflammation. Moreover, we found that neuroinflammation impaired hippocampal neurogenesis, leading to reduced synaptic plasticity and cognitive impairment, corroborating previous studies [48,49]. Minocycline was shown to protect the nervous system through anti-inflammatory effects [50]. Due to its anti-inflammatory properties, multiple reports have focused on minocycline’s therapeutic effect on cognitive dysfunction. However, in 2023, Takazawa et al. examined minocycline’s effect on POCD in elderly individuals administered total knee arthroplasty and found that 200 mg of minocycline daily from the day before surgery until the seventh day after surgery did not decrease POCD incidence, contradicting many preclinical studies [51]. The negative results of this study may be related to the short dosing period. Indeed, a study in 2021 showed that the duration of postoperative neuroinflammation in aged mice may be as long as 14 days [52]. In 2015, Tian and collaborators reported for the first time that minocycline reduces sevoflurane-associated neuroapoptosis and inflammation by suppressing sevoflurane-associated Aβ buildup and NF-κB pathway activation in a hippocampal sample from aged rats [16]. Two years later, Tian et al. again found that minocycline might protect from sevoflurane-associated cell damage through Nrf2-dependent antioxidation and NF-κB pathway suppression [53]. Both studies conducted by Tian et al. revealed that minocycline attenuates sevoflurane-related cognitive dysfunction through anti-inflammatory effects, but how minocycline suppresses sevoflurane-induced neuroinflammation remains undefined. The present study found that minocycline attenuates sevoflurane-induced POCD in aged mice by suppressing hippocampal apoptosis, neuroinflammatory, neurogenesis dysfunction, and synaptic plasticity impairment. Moreover, we found that Notch signaling participated in sevoflurane-associated neuroinflammation and that minocycline protected against neuroinflammation by regulating the Notch signaling pathway. The limitations of this study should be mentioned. First, we focused on minocycline’s protective effects on short-term sevoflurane-associated cognitive dysfunction. However, studies have shown that sevoflurane has the potential to cause long-term cognitive impairment as well [54,55], so it is still worth investigating whether minocycline improves long-term cognitive dysfunction induced by sevoflurane. Secondly, how sevoflurane activates Notch signaling also deserves further investigation. ## 5. Conclusions This study indicates that sevoflurane induces cognitive dysfunction in aged mice by causing apoptosis, neuroinflammation, neurogenesis dysfunction, and impaired synaptic plasticity. Minocycline attenuates sevoflurane-induced synaptic plasticity impairment through antiapoptotic and anti-neuroinflammatory effects, so as to improve cognitive dysfunction, and its anti-inflammatory effect is at least partially achieved via the Notch signaling pathway. In future studies, we will continue to explore the mechanism of how sevoflurane activates Notch signaling and whether minocycline improves long-term cognitive dysfunction induced by sevoflurane. In summary, our study indicates that minocycline use is still expected to become a new treatment method to prevent POCD induced by sevoflurane in clinics. ## References 1. Kotekar N., Shenkar A., Nagaraj R.. Postoperative cognitive dysfunction—Current preventive strategies. Clin. Interv. Aging (2018) 13 2267-2273. DOI: 10.2147/CIA.S133896 2. Miller D., Lewis S.R., Pritchard M.W., Schofield-Robinson O.J., Shelton C.L., Alderson P., Smith A.F.. Intravenous versus inhalational maintenance of anaesthesia for postoperative cognitive outcomes in elderly people undergoing non-cardiac surgery. Cochrane Database Syst. Rev. (2018) 8 CD012317. DOI: 10.1002/14651858.CD012317.pub2 3. Chen Y., Zhang P., Lin X., Zhang H., Miao J., Zhou Y., Chen G.. Mitophagy impairment is involved in sevoflurane-induced cognitive dysfunction in aged rats. Aging (2020) 12 17235-17256. DOI: 10.18632/aging.103673 4. Chai G., Wu J., Fang R., Liu Y., Wang X., Wang X., Zhang J., Zhou J., Jiang Z., Yi H.. Sevoflurane inhibits histone acetylation and contributes to cognitive dysfunction by enhancing the expression of ANP32A in aging mice. Behav. Brain Res. (2022) 431 113949. DOI: 10.1016/j.bbr.2022.113949 5. Yu Y., Yang Y., Tan H., Boukhali M., Khatri A., Yu Y., Hua F., Liu L., Li M., Yang G.. Tau Contributes to Sevoflurane-induced Neurocognitive Impairment in Neonatal Mice. Anesthesiology (2020) 133 595-610. DOI: 10.1097/ALN.0000000000003452 6. Wang X., Dong Y., Zhang Y., Li T., Xie Z.. Sevoflurane induces cognitive impairment in young mice via autophagy. PLoS ONE (2019) 14. DOI: 10.1371/journal.pone.0216372 7. Qin J.H., Zhang X.R., He L., Zhu J., Ma Q.J.. Effect of sevoflurane and halothane anesthesia on cognitive function and immune function in young rats. Saudi J. Biol. Sci. (2018) 25 47-51. DOI: 10.1016/j.sjbs.2016.08.002 8. Dong Y., Liang F., Huang L., Fang F., Yang G., Tanzi R.E., Zhang Y., Quan Q., Xie Z.. The anesthetic sevoflurane induces tau trafficking from neurons to microglia. Commun. Biol. (2021) 4 560. DOI: 10.1038/s42003-021-02047-8 9. Li G., Wang Y., Cao F., Wang D., Zhou L., Jin Y.. Sevoflurane Promotes Neurodegeneration Through Inflammasome Formation in APP/PS1 Mice. Front. Neurosci. (2021) 15 647136. DOI: 10.3389/fnins.2021.647136 10. Gui L., Lei X., Zuo Z.. Decrease of glial cell-derived neurotrophic factor contributes to anesthesia- and surgery-induced learning and memory dysfunction in neonatal rats. J. Mol. Med. (2017) 95 369-379. DOI: 10.1007/s00109-017-1521-9 11. Xu X., Zhao Y., Xu M., Dai Q., Meng W., Yang J., Qin R.. Activation of Notch signal pathway is associated with a poorer prognosis in acute myeloid leukemia. Med. Oncol. (2011) 28 S483-S489. DOI: 10.1007/s12032-010-9667-0 12. Tindemans I., van Schoonhoven A., KleinJan A., de Bruijn M.J., Lukkes M., van Nimwegen M., van den Branden A., Bergen I.M., Corneth O.B., van IJcken W.F.. Notch signaling licenses allergic airway inflammation by promoting Th2 cell lymph node egress. J. Clin. Investig. (2020) 130 3576-3591. DOI: 10.1172/JCI128310 13. Lavoz C., Poveda J., Marquez-Exposito L., Rayego-Mateos S., Rodrigues-Diez R.R., Ortiz A., Egido J., Mezzano S., Ruiz-Ortega M.. Gremlin activates the Notch pathway linked to renal inflammation. Clin. Sci. (2018) 132 1097-1115. DOI: 10.1042/CS20171553 14. Kobayashi K., Imagama S., Ohgomori T., Hirano K., Uchimura K., Sakamoto K., Hirakawa A., Takeuchi H., Suzumura A., Ishiguro N.. Minocycline selectively inhibits M1 polarization of microglia. Cell Death Dis. (2013) 4 e525. DOI: 10.1038/cddis.2013.54 15. Li J., Chen J., Mo H., Chen J., Qian C., Yan F., Gu C., Hu Q., Wang L., Chen G.. Minocycline Protects Against NLRP3 Inflammasome-Induced Inflammation and P53-Associated Apoptosis in Early Brain Injury After Subarachnoid Hemorrhage. Mol. Neurobiol. (2016) 53 2668-2678. DOI: 10.1007/s12035-015-9318-8 16. Tian Y., Guo S., Wu X., Ma L., Zhao X.. Minocycline alleviates sevoflurane-induced cognitive impairment in aged rats. Cell Mol. Neurobiol. (2015) 35 585-594. DOI: 10.1007/s10571-014-0154-6 17. Yang C., Gao J., Wu B., Yan N., Li H., Ren Y., Kan Y., Liang J., Jiao Y., Yu Y.. Minocycline attenuates the development of diabetic neuropathy by inhibiting spinal cord Notch signaling in rat. Biomed. Pharmacother. (2017) 94 380-385. DOI: 10.1016/j.biopha.2017.07.078 18. Lu Y., Zhou M., Li Y., Li Y., Hua Y., Fan Y.. Minocycline promotes functional recovery in ischemic stroke by modulating microglia polarization through STAT1/STAT6 pathways. Biochem. Pharmacol. (2021) 186 114464. DOI: 10.1016/j.bcp.2021.114464 19. Dai X.J., Li N., Yu L., Chen Z.-Y., Hua R., Qin X.. Activation of BV2 microglia by lipopolysaccharide triggers an inflammatory reaction in PC12 cell apoptosis through a toll-like receptor 4-dependent pathway. Cell Stress Chaperones (2015) 20 321-331. DOI: 10.1007/s12192-014-0552-1 20. Xu L., Shen J., Yu L., Sun J., McQuillan P.M., Hu Z., Yan M.. Role of autophagy in sevoflurane-induced neurotoxicity in neonatal rat hippocampal cells. Brain Res. Bull. (2018) 140 291-298. DOI: 10.1016/j.brainresbull.2018.05.020 21. Zhang D., Xue B., You J., Zhang B., Chai G.. Suberoylanilide hydroxamic acid reversed cognitive and synaptic plasticity impairments induced by sevoflurane exposure in adult mice. Neuroreport (2019) 30 274-279. DOI: 10.1097/WNR.0000000000001196 22. Zhao X.P., Zhong F., Luo R.Y., Zhang Y.L., Luo C., Li H., Dai R.P.. Early-life sevoflurane exposure impairs fear memory by suppressing extracellular signal-regulated kinase signaling in the bed nucleus of stria terminalis GABAergic neurons. Neuropharmacology (2021) 191 108584. DOI: 10.1016/j.neuropharm.2021.108584 23. Ju L.S., Jia M., Sun J., Sun X.R., Zhang H., Ji M.H., Yang J.J., Wang Z.Y.. Hypermethylation of Hippocampal Synaptic Plasticity-Related genes is Involved in Neonatal Sevoflurane Exposure-Induced Cognitive Impairments in Rats. Neurotox. Res. (2016) 29 243-255. DOI: 10.1007/s12640-015-9585-1 24. Toda T., Parylak S.L., Linker S.B., Gage F.H.. The role of adult hippocampal neurogenesis in brain health and disease. Mol. Psychiatry (2019) 24 67-87. DOI: 10.1038/s41380-018-0036-2 25. Jia J., Zhu J., Yang Q., Wang Y., Zhang Z., Chen C.. The Role of Histone Acetylation in the Sevoflurane-induced Inhibition of Neurogenesis in the Hippocampi of Young Mice. Neuroscience (2020) 432 73-83. DOI: 10.1016/j.neuroscience.2020.02.023 26. Xu L., Guo Y., Wang G., Sun G., Sun W., Li J., Li X., Wu J., Zhang M.. Inhibition of Adult Hippocampal Neurogenesis Plays a Role in Sevoflurane-Induced Cognitive Impairment in Aged Mice Through Brain-Derived Neurotrophic Factor/Tyrosine Receptor Kinase B and Neurotrophin-3/Tropomyosin Receptor Kinase C Pathways. Front. Aging Neurosci. (2022) 14 782932. DOI: 10.3389/fnagi.2022.782932 27. Stefanis L., Burke R.E., Greene L.A.. Apoptosis in neurodegenerative disorders. Curr. Opin. Neurol. (1997) 10 299-305. DOI: 10.1097/00019052-199708000-00004 28. Kang W., Lu D., Yang X., Ma W., Chen X., Chen K., Xu X., Zhou X., Zhou L., Feng X.. Sevoflurane Induces Hippocampal Neuronal Apoptosis by Altering the Level of Neuropeptide Y in Neonatal Rats. Neurochem. Res. (2020) 45 1986-1996. DOI: 10.1007/s11064-020-03028-9 29. Zhang X., Zhou Y., Xu M., Chen G.. Autophagy Is Involved in the Sevoflurane Anesthesia-Induced Cognitive Dysfunction of Aged Rats. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0153505 30. Garrido-Mesa N., Zarzuelo A., Gálvez J.. Minocycline: Far beyond an antibiotic. Br. J. Pharmacol. (2013) 169 337-352. DOI: 10.1111/bph.12139 31. Qian D., Li L., Rong Y., Liu W., Wang Q., Zhou Z., Gu C., Huang Y., Zhao X., Chen J.. Blocking Notch signal pathway suppresses the activation of neurotoxic A1 astrocytes after spinal cord injury. Cell Cycle (2019) 18 3010-3029. DOI: 10.1080/15384101.2019.1667189 32. Wu L., Li Y., Yu M., Yang F., Tu M., Xu H.. Notch Signaling Regulates Microglial Activation and Inflammatory Reactions in a Rat Model of Temporal Lobe Epilepsy. Neurochem. Res. (2018) 43 1269-1282. DOI: 10.1007/s11064-018-2544-5 33. Bedford P.D.. Adverse cerebral effects of anaesthesia on old people. Lancet (1955) 269 259-263. DOI: 10.1016/S0140-6736(55)92689-1 34. Boone M.D., Sites B., von Recklinghausen F.M., Mueller A., Taenzer A.H., Shaefi S.. Economic Burden of Postoperative Neurocognitive Disorders Among US Medicare Patients. JAMA Netw. Open (2020) 3 e208931. DOI: 10.1001/jamanetworkopen.2020.8931 35. Evered L.A., Silbert B.S., Scott D.A., Maruff P., Ames D.. Prevalence of Dementia 7.5 Years after Coronary Artery Bypass Graft Surgery. Anesthesiology (2016) 125 62-71. DOI: 10.1097/ALN.0000000000001143 36. Relander K., Hietanen M., Rantanen K., Rämö J., Vento A., Saastamoinen K.P., Roine R.O., Soinne L.. Postoperative cognitive change after cardiac surgery predicts long-term cognitive outcome. Brain Behav. (2020) 10 e01750. DOI: 10.1002/brb3.1750 37. Kitsis P., Zisimou T., Gkiatas I., Kostas-Agnantis I., Gelalis I., Korompilias A., Pakos E.. Postoperative Delirium and Postoperative Cognitive Dysfunction in Patients with Elective Hip or Knee Arthroplasty: A Narrative Review of the Literature. Life (2022) 12. DOI: 10.3390/life12020314 38. Galinkin J.L., Janiszewski D., Young C.J., Klafta J.M., Klock P.A., Coalson D.W., Apfelbaum J.L., Zacny J.P.. Subjective, psychomotor, cognitive, and analgesic effects of subanesthetic concentrations of sevoflurane and nitrous oxide. Anesthesiology (1997) 87 1082-1088. DOI: 10.1097/00000542-199711000-00012 39. Liang R., Ou S., Han Y., Xu J., Zhou S.. Plasma amyloid beta level changes in aged mice with cognitive dysfunction following sevoflurane exposure. Exp. Gerontol. (2020) 129 110737. DOI: 10.1016/j.exger.2019.110737 40. Shimohama S.. Apoptosis in Alzheimer’s disease—An update. Apoptosis (2000) 5 9-16. DOI: 10.1023/A:1009625323388 41. Lev N., Melamed E., Offen D.. Apoptosis and Parkinson’s disease. Prog. Neuropsychopharmacol. Biol. Psychiatry (2003) 27 245-250. DOI: 10.1016/S0278-5846(03)00019-8 42. Sathasivam S., Ince P.G., Shaw P.J.. Apoptosis in amyotrophic lateral sclerosis: A review of the evidence. Neuropathol. Appl. Neurobiol. (2001) 27 257-274. DOI: 10.1046/j.0305-1846.2001.00332.x 43. Chen G., Gong M., Yan M., Zhang X.. Sevoflurane induces endoplasmic reticulum stress mediated apoptosis in hippocampal neurons of aging rats. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0057870 44. Shen X., Dong Y., Xu Z., Wang H., Miao C., Soriano S.G., Sun D., Baxter M.G., Zhang Y., Xie Z.. Selective anesthesia-induced neuroinflammation in developing mouse brain and cognitive impairment. Anesthesiology (2013) 118 502-515. DOI: 10.1097/ALN.0b013e3182834d77 45. Dong P., Zhao J., Li N., Lu L., Li L., Zhang X., Yang B., Zhang L., Li D.. Sevoflurane exaggerates cognitive decline in a rat model of chronic intermittent hypoxia by aggravating microglia-mediated neuroinflammation via downregulation of PPAR-γ in the hippocampus. Behav. Brain Res. (2018) 347 325-331. DOI: 10.1016/j.bbr.2018.03.031 46. Siebel C., Lendahl U.. Notch Signaling in Development, Tissue Homeostasis, and Disease. Physiol. Rev. (2017) 97 1235-1294. DOI: 10.1152/physrev.00005.2017 47. Kopan R., Ilagan M.X.. The canonical Notch signaling pathway: Unfolding the activation mechanism. Cell (2009) 137 216-233. DOI: 10.1016/j.cell.2009.03.045 48. Zhang X., Wei X., Mei Y., Wang D., Wang J., Zhang Y., Li X., Gu Y., Peng G., Sun B.. Modulating adult neurogenesis affects synaptic plasticity and cognitive functions in mouse models of Alzheimer’s disease. Stem Cell Rep. (2021) 16 3005-3019. DOI: 10.1016/j.stemcr.2021.11.003 49. Sun G., Miao Z., Ye Y., Zhao P., Fan L., Bao Z., Tu Y., Li C., Chao H., Xu X.. Curcumin alleviates neuroinflammation, enhances hippocampal neurogenesis, and improves spatial memory after traumatic brain injury. Brain Res. Bull. (2020) 162 84-93. DOI: 10.1016/j.brainresbull.2020.05.009 50. Naderi Y., Sabetkasaei M., Parvardeh S., Zanjani T.M.. Neuroprotective effect of minocycline on cognitive impairments induced by transient cerebral ischemia/reperfusion through its anti-inflammatory and anti-oxidant properties in male rat. Brain Res. Bull. (2017) 131 207-213. DOI: 10.1016/j.brainresbull.2017.04.010 51. Takazawa T., Horiuchi T., Orihara M., Nagumo K., Tomioka A., Ideno Y., Hayashi K., Yashima H., Araki T., Hatayama K.. Prevention of Postoperative Cognitive Dysfunction by Minocycline in Elderly Patients after Total Knee Arthroplasty: A Randomized, Double-blind, Placebo-controlled Clinical Trial. Anesthesiology (2023) 138 172-183. DOI: 10.1097/ALN.0000000000004439 52. Oberman K., Hovens I., de Haan J., Falcao-Salles J., van Leeuwen B., Schoemaker R.. Acute pre-operative ibuprofen improves cognition in a rat model for postoperative cognitive dysfunction. J. Neuroinflamm. (2021) 18 156. DOI: 10.1186/s12974-021-02206-y 53. Tian Y., Wu X., Guo S., Ma L., Huang W., Zhao X.. Minocycline attenuates sevoflurane-induced cell injury via activation of Nrf2. Int. J. Mol. Med. (2017) 39 869-878. DOI: 10.3892/ijmm.2017.2908 54. Liao Z., Li J., Miao L., Huang Z., Huang W., Liu Y., Li Y.. Inhibition of RhoA Activity Does Not Rescue Synaptic Development Abnormalities and Long-Term Cognitive Impairment After Sevoflurane Exposure. Neurochem. Res. (2021) 46 468-481. DOI: 10.1007/s11064-020-03180-2 55. Sun Z., Satomoto M., Adachi Y.U., Kinoshita H., Makita K.. Inhibiting NADPH oxidase protects against long-term memory impairment induced by neonatal sevoflurane exposure in mice. Br. J. Anaesth. (2016) 117 80-86. DOI: 10.1093/bja/aew064
--- title: Ablation of GPR56 Causes β-Cell Dysfunction by ATP Loss through Mistargeting of Mitochondrial VDAC1 to the Plasma Membrane authors: - Israa Mohammad Al-Amily - Marie Sjögren - Pontus Duner - Mohammad Tariq - Claes B. Wollheim - Albert Salehi journal: Biomolecules year: 2023 pmcid: PMC10046417 doi: 10.3390/biom13030557 license: CC BY 4.0 --- # Ablation of GPR56 Causes β-Cell Dysfunction by ATP Loss through Mistargeting of Mitochondrial VDAC1 to the Plasma Membrane ## Abstract The activation of G Protein-Coupled Receptor 56 (GPR56), also referred to as Adhesion G-Protein-Coupled Ceceptor G1 (ADGRG1), by Collagen Type III (Coll III) prompts cell growth, proliferation, and survival, among other attributes. We investigated the signaling cascades mediating this functional effect in relation to the mitochondrial outer membrane voltage-dependent anion Channel-1 (VDAC1) expression in pancreatic β-cells. GPR56KD attenuated the Coll III-induced suppression of P70S6K, JNK, AKT, NFκB, STAT3, and STAT5 phosphorylation/activity in INS-1 cells cultured at 20 mM glucose (glucotoxicity) for 72 h. GPR56-KD also increased Chrebp, Txnip, and Vdac1 while decreasing Vdac2 mRNA expression. In GPR56-KD islet β-cells, Vdac1 was co-localized with SNAP-25, demonstrating its plasma membrane translocation. This resulted in ATP loss, reduced cAMP production and impaired glucose-stimulated insulin secretion (GSIS) in INS-1 and human EndoC βH1 cells. The latter defects were reversed by an acute inhibition of VDAC1 with an antibody or the VDAC1 inhibitor VBIT-4. We demonstrate that Coll III potentiates GSIS by increasing cAMP and preserving β-cell functionality under glucotoxic conditions in a GPR56-dependent manner by attenuating the inflammatory response. These results emphasize GPR56 and VDAC1 as drug targets in conditions with impaired β-cell function. ## 1. Introduction The prevalence of metabolic diseases, in particular obesity and Type 2 diabetes (T2D), is high among the older population in western countries and is also increasing among the young around the globe. A resistance to insulin in its target organs, together with pancreatic β-cell dysfunction, plays a central role in the development of the metabolic syndrome and T2D [1,2,3]. Many studies have shown that genetic and environmental factors are interconnected in promoting the development of the disease through the failure of the β-cells to increase insulin secretion in compensation for resistance to the hormone [1,3]. In most of these conditions, suboptimal blood glucose control and dyslipidemia during years of prediabetes lead to β-cell dysfunction, dedifferentiation, and, eventually, apoptosis [1,4,5]. Thus, the prevention of β-cell dysfunction would have a great impact on the prevention of the metabolic syndrome and T2D development [5,6,7]. Glucotoxicity induces cellular stress in β-cells, such as the activation of inflammatory signals, oxidative stress, and endoplasmic reticulum stress, ultimately impairing the functions of vital subcellular organelles, including mitochondria [8,9]. Since β-cell mitochondria play a central role in the coupling of glucose metabolism to insulin secretion, their dysfunction has been implicated in the defective hormone release in T2D [8,9,10]. Mitochondrial metabolism requires an inward and outward flux of hydrophilic metabolites, including ATP, ADP, and respiratory substrates, through voltage-dependent anion channels (VDACs), also known as mitochondrial porins, in the mitochondrial outer membrane [11]. There are three VDAC isoforms (VDAC1, VDAC2, and VDAC3) differentially expressed in mammalian tissues [11,12]. We previously presented evidence that pancreatic β-cells express all three isoforms [13,14]. We described the overexpression of VDAC1 and downregulation of VDAC2 in islets of T2D organ donors. This causes a mis-targeting of VDAC1 to the β-cell surface, resulting in ATP loss and impaired glucose-stimulated insulin secretion (GSIS). This paradigm is mimicked by the culture of human islets in the presence of high glucose and can be prevented by VDAC1 knock-down, or by an acute addition of VDAC1 inhibitors [14]. These results show that T2D β-cells are dysfunctional, and that GSIS can be recovered both in vitro [14] and in vivo [7,15]. G-Protein-Coupled Receptors (GPCRs) constitute the largest group of cell surface receptors in men and are also the targets of ~$35\%$ of all prescription medicines [16,17]. G-Protein-Coupled Receptor 56 (GPR56), also known as Adhesion G-Protein-Coupled Receptor G1 (ADGRG1), is highly expressed in many tissues, including pancreatic β-cells [18,19,20,21]. Functionally, GPR56 has been linked to a myriad of biological and physiological processes within the cells, including the development and differentiation of cerebral cortex, innate immunity, muscle cell and oligodendrocyte development, and carcinogenesis, as well as affecting pancreatic β-cell function [18,19,20,21,22,23]. The GPR56 receptor is an adhesion receptor characterized by an extremely long N-terminal extracellular domain (NTD), which is cleaved off from the seven-membrane-spanning C-terminal domain (CTD) by auto-proteolysis upon binding to its principal ligand Collagen Type III. The CTD initiates signaling, the nature of which depends on the cellular context. Several studies have shown GPR56 coupling to the Gα$\frac{12}{13}$ class of heterotrimeric G-proteins to promote RhoA activation and alterations of the actin filamentous network [22,23]. GPR56 has also been reported to signal via Gαs to stimulate adenylate cyclase [18,24], as well as raising [Ca2+]i through Gαq [19], although this was not clearly linked to Gαq since it is in contrast to ATP-induced [Ca2+]i rise, which was abolished in the absence of extracellular Ca2+ [19], suggesting the activation of calcium channels [20,21]. We have previously shown that GPR56 activation promotes β-cell viability and function both in human and rodents; in human pancreatic β-cells, GPR56 expression correlates with the expression levels of transcripts, vital for the function of β-cells and GPR56 mRNA, which is indeed reduced in islets of Type 2 diabetic organ donors [18]. We have also shown that the knock-down of GPR56 (GPR56-KD) leads to β-cell dysfunction, reduced cell viability, and attenuates the beneficial effect of Collagen Type III (Coll III) on β-cell function [18]. Our observations thus corroborate reported findings showing that, although GPR56-KO mice have normal islet vascularization and only mildly impaired glucose tolerance, the activation of GPR56 by Coll III increases islet insulin secretion and enhances cell viability [19,25]. GPR56 has, thus, been mainly linked to protecting β-cells from apoptosis, but it is similarly important for the insulin secretory function of β-cells. GPR56 is linked to the activation of the cAMP/Protein Kinase A (PKA) system [18], while it is well-known that the cAMP- and calcium-signaling pathways are coupled, i.e., a rise in [Ca2+]i can also activate adenylate cyclase [15,26]. These results show the complexity of GPR56-induced signaling pathways, which are not completely elucidated. By virtue of its transmembrane expression and its capability of also having adhesive GPCR functions, GPR56 seems not only to regulate secretory capacity but also interlinking many intracellular-signaling pathways important for β-cell structure and viability. The aim of the present investigation was to study to what extent GPR56-KD could interfere with the effect of Coll III on the activity of several important stress kinases, and whether GPR56-KD is associated with increased VDAC1 expression and its mistargeting to the cell surface in rodent and human β-cells. ## 2.1. Animals Female mice (C57/bl) (Janvier Laboratory, Saint Isle, France) weighing 25–30 g were used under standard conditions (12 h light/dark cycle, 22 °C) with access to standard pellet diet (B&K) and water ad libitum. The study protocol was approved by the Ethics Committee for Animal Research at Lund University ($\frac{1057}{2020}$). Isolation of pancreatic islets was performed by retrograde injection of a collagenase solution via the pancreatic duct and islets, which were then collected under a stereomicroscope at room temperature [27]. ## 2.2. Reagents Fatty acid-free bovine serum albumin (Boehringer, Ingelheim, Germany), rabbit polyclonal and mouse monoclonal anti-VDAC1 antibody (N-terminal) (Abcam, Cambridge, UK and Santa Cruz Biotechnology Inc, Santa Cruz, CA, USA respectively), insulin ELISA kit (Mercordia, Uppsala, Sweden), primers and qPCR materials were from Applied Biosystems (Waltham, MA, USA). Cell Signaling Multiplex Assay from Merck Millipore, VBIT-4 and AKOS022075291 (AKOS) were from Glixx Laboratories (Hopkinton, MA, USA). All other chemicals were from Merck AG (Darmstadt, Germany) or Sigma (Saint Louis, MO, USA). ## 2.3. INS-1 832/13 Cell Culture INS-1 $\frac{832}{13}$ cells (kindly donated by Dr. C. B. Newgaard, Duke University, Durham, NC, USA) were cultured in RPMI-1640 containing 11.1 mM of D-glucose and supplemented with $10\%$ fetal bovine serum, 100 U/mL of penicillin (Gibco, BRL, Gaithersburg, MD, USA), 100 μg/mL of streptomycin (PAA Laboratories, Toronto, Ontario, Canada), 10 mM of N-2 hydroxyethylpiperazine-N’-2-ethanesulfonic acid (HEPES), 2 mM of glutamine, 1 mM of sodium pyruvate, and 50 μM of β-mercaptoethanol (Sigma Aldrich, Saint Louis, MO, USA) at 37 °C in a humidified atmosphere containing $95\%$ air and $5\%$ CO2. For an assessment of the effects of elevated glucose (glucotoxicity), cells were treated with 20 mM of D-glucose (20G) for 72 h. Controls were maintained in 5 mM of glucose (5G) media. For GPR56 knockdown by siRNA, INS1 $\frac{832}{13}$ cells were cultured to $75\%$ confluency and then subjected to GPR56-KD as previously described [18]. The siRNAs used for Gpr56-KD are shown in Supplementary Table S1. Each treatment was carried out in three biological replicates, along with scramble controls. Thereafter, the cells were washed and cultured for 6 h in a normal RPMI medium (recovery period) before being cultured in RPMI-1640 with 5 or 20 mM of glucose (5 G or 20 G) in the presence or absence of 20 ug/mL of Collagen-III (Col III) for 72 h (Figure 1), or they were subjected to incubation with indicated agents (see results Section 3.4). ## 2.4. EndoC βH1 Cell Culture The human clonal β-cell line EndoC-βH1 (EndoCells, Paris, France) was seeded in 24 well plates at a density of 1.8 × 105 cells/well and maintained in DMEM culture medium (5.5 mM glucose) $2\%$ BSA fraction V (Roche, Basel, Switzerland), 10 mM of nicotinamide (Merck, Darmstadt, Germany), 50 µM of 2-mercaptoethanol, 5.5 µg/mL of transferrin, 6.7 ng/mL of sodium selenite (Sigma), 100 U/mL of penicillin, and 100 µg/mL of streptomycin (PAA Laboratories, Toronto, Ontario, Canada) as described in [28]. For measurement of insulin secretion, the cells were cultured for 12h in the culture medium containing 2.8 mM glucose before incubation at 1 or 20 mM of glucose. ## 2.5. Knockdown of GPR56 in Mouse Islets, INS-1 832/13, and EndoC βH1 Cells GPR56-KD in mouse islets was performed using a cocktail of three different Lentivirus delivered by shRNAs targeting the GPR56 gene (Santa Cruz, CA, USA), as described previously in [18]. For downregulation of GPR56 in INS-1 $\frac{832}{13}$ cells and EndoC βH1 cells, siRNA from ThermoFisher Scientific (Wilmington, DE, USA), with an appropriate scrambled control, were used according to the manufacturer’s recommendations. Validation of GPR56-KD was determined by qPCR and confocal microscopy, as also described previously [18]. The siRNAs, or shRNAs used for GPR56-KD, are shown in Supplementary Table S1. ## 2.6. Determination of Intracellular Pathways The Effect of Collagen Type III (Col III) on the major signaling pathways was determined in scramble control (Scr control) and Gpr56-KD INS-1 $\frac{832}{13}$ cells cultured at 5 or 20 mM of glucose for 72 h. Detection of phosphorylated P70S6K, JNK, AKT, NFκb, STAT3, and STAT5 was assayed on the cell extracts by Luminex, according to the manufacturer’s protocol. ## 2.7. Immunostaining and Confocal Imaging Isolated mouse islets, as well as EndoC-βH1 cells, were seeded on glass-bottom dishes cultured overnight. They were then washed twice and fixed with $3\%$ paraformaldehyde for 10 min, followed by permeabilization with $0.1\%$ Triton-X 100 for 15 min. The blocking solution contained $5\%$ normal donkey serum in PBS and was used for 15 min. Primary antibodies against VDAC1 (Abcam, 1:200), SNAP-25 (Abcam, 1:100), Na+/K+ ATPase (Abcam, 1:100), and Guinea pig insulin (Eurodiagnostica, 1:300) were diluted in blocking buffer and incubated overnight at 4 °C. Immunoreactivity was quantified using fluorescently labeled secondary antibodies (1:200) and visualized by confocal microscopy (Carl Zeiss, Germany). ## 2.8. Quantitative Polymerase Chain Reaction (qPCR) Total RNA was extracted from INS1 $\frac{832}{13}$ and EndoC-βH1 cells using RNAeasy (Qiagen, Hilden, Germany) before complementary DNA (cDNA) was synthesized using SuperScript (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s protocol. Concentration and purity of total RNA were measured with a NanoDrop ND-1000 spectrophotometer (A260/A280 > 1.9 and A260/A23 0 > 1.4) (NanoDrop Technologies LLC, Wilmington, DE, USA). RNA Quality Indicator (RQI) higher than 8.0 (Experion Automated Electrophoresis, Bio-Rad, USA) was considered to be high-quality total RNA preparation. TaqMan mastermix from Applied Biosystems (Foster City, CA, USA) was used for qPCR and performed following manufacturer’s protocol and was run in a 7900 HT Fast Real-Time System (Applied Biosystems). The qPCR was carried out as follows: 50 °C for 2 min, 95 °C for 10 min, 40 cycles of 95 °C for 15 s, and 60 °C for 1 min. Changes in gene expression were calculated using the ΔΔCt method with a fold-change cut-off at ≥ 1.5 and $p \leq 0.05$ considered significant. All samples were run in duplicate, and relevant negative controls were run on each plate. qPCR results were normalized to housekeeping genes (PPIA or HPRT). Primer sequences used in the qPCR assays are provided in Supplementary Table S2. ## 2.9. Western Blots EndoC βH1 cells were homogenized in ice-cold RIPA buffer and kept shaking on ice for 30 min. Extracted total protein content from homogenates was measured by Pierce BCA Protein Assay Kit (Thermo Scientific, Waltham, MA USA). Homogenate samples (10 µg) from scramble control (Scr) or GPR56-KD cells were electrophoresed on $7.5\%$ SDS-polyacrylamide gel (Bio-Rad, Hercules, CA, USA). After electrophoresis, proteins were transferred to a nitrocellulose membrane (Bio-Rad). The membrane was blocked in LS-buffer (10 mmol/L Tris, pH 7.4, 100 mmol/L NaCl, $0.1\%$ Tween-20) containing $5\%$ non-fat dry milk powder for 60 min at room temperature. Subsequently, the membranes were incubated overnight with the same VDAC1 antibody used for the confocal experiments (1:10,000) at 4 °C. After washing (three times) in LS-buffer, the membrane was finally incubated with a horseradish peroxidase-conjugated anti-rabbit antibody (1:5000) (Bio-Rad, Hercules, CA, USA). Detection of α-tubulin was with rabbit-anti-α-tubulin (Sigma, USA) and secondary anti-rabbit anti-body (1:1000). Immunoreactivity was detected using an enhanced chemiluminescence reaction (Pierce, Rockford, IL, USA). The blots were scanned with ChemiDocTM MP Imaging System (Bio-Rad), and bands corresponding to the ~37-kDa (protein marker) were identified as VDAC1 protein. A typical Western blot image of entire gel, performed on the homogenates from two Scr controls and four GPR56-KD cells, is shown in Supplementary Figure S2E. ## 2.10. Insulin Secretion For functional studies after recovery, the siRNA-treated INS-1 $\frac{832}{13}$, or EndoC βH1 cells, were washed and pre-incubated for 120 min at 37 °C in SAB buffer, pH 7.4, and supplemented with 10 mM of HEPES, 0.1 % bovine serum albumin, and 2.8 mM of glucose. After pre-incubation, the buffer was changed and INS-1 $\frac{832}{13}$, or EndoC βH1 cells, were incubated at 1 or 20 mM of glucose with indicated agents for 60 min at 37 °C. Immediately after incubation, an aliquot of the medium was removed and frozen for subsequent assay of insulin. The cells were then washed with PBS and stored in 100 mM of HCl containing IBMX (100 μM) for subsequent analysis of cyclic AMP. ## 2.11. cAMP Determination The cAMP content in the cell lysate was measured using a direct cAMP ELISA kit (Enzo Life Sciences, Farmingdale, NY, USA) according to the manufacturer’s instructions, and the values were related to protein content. The protein concentrations of the cell lysates were measured by a BCA kit (Thermo Fisher Scientific, Wilmington, DE, USA). ## 2.12. ATP Determination ATP content (INS-1 $\frac{832}{13}$ and EndoC βH1 cells) and release (INS-1 $\frac{832}{13}$ cells) in incubated cells after GPR56-KD were determined using a luminometric assay kit according to manufacturer’s recommendation (BioVision, Milpitas, CA, USA) and normalized to protein content. After incubation, the cells were washed with PBS buffer (three times) in Ripa buffer containing protease inhibitors and stored at −80 °C for subsequent measurements of cellular ATP, while the released ATP was measured in the 1-h incubation medium. The protein contents of each sample were analyzed by BCA protein kit (Thermo Scientific, IL, USA). ## 2.13. Statistics The results are expressed as means ± SEM for the indicated number of observations or illustrated by an observation representative of the results obtained from different experiments (confocal microscopy). The significance of random differences was analyzed by Student’s t-test or, where applicable, an analysis of variance was performed, followed by Tukey–Kramers’ multiple comparisons test. p-value < 0.05 was considered significant. ## 3.1. The Consequence of GPR56-KD on the Activation of Several Intracellular Pathways The impact of long-term high glucose (20 mM, 72 h) culture on the key signaling molecules (pathways) involved in the physiology/pathophysiology of β-cells in scramble control INS-1 cells in the presence or absence of the naturally occurring GPR56 agonist Coll III was studied. The effect of Coll III on these signaling molecules was only investigated in Gpr56-KD cells under the two glucose culture conditions. Cell-signaling analysis revealed an increase in p-P70S6K, p-JNK, and p-NFκB induced by high glucose, while p-AKT, p-STAT3, and p-STAT5 were not significantly altered in Scr control INS-1 cells (Figure 1A–C). The presence of Coll III (20 µg/mL) during the culture period suppressed the phosphorylation of P70S6K, JNK, NFκB, AKT, STAT3, and STAT5, both at 5 or 20 mM of glucose in Scr control cells (Figure 1A–F). This effect of Coll III was abolished in Gpr56-KD INS-1 cells regardless of ambient glucose levels (Figure 1A–F). While high glucose did not appreciably affect p-CREB, it was increased in scramble control INS-1 $\frac{832}{13}$ cells by the presence of Coll III (not shown). Coll III actions are thus mediated through GPR56 activation. ## 3.2. The Effect of GPR56-KD on the Expression of Chrebp, Txnip, Vdac1 and Vdac2 Since we have reported that GPR56-KD is associated with β-cell dysfunction and decreased viability reminiscent of diabetic β-cells [18], we evaluated the impact of GPR56-KD on mitochondrial VDAC1 and VDAC2 expression, as we have demonstrated that the altered expression of VDAC1 and VDAC2 is associated with mitochondrial dysfunction, leading to impaired β-cell function [13,14]. We also assessed the consequence of GPR56-KD on Chrebp and txnip, two transcriptional factors of importance for the glucotoxicity-induced increase in VDAC1 expression in pancreatic β-cells [14]. As seen in Figure 2, GPR56-KD significantly increased mRNA expression of Chrebp (A), Txnip (B), and Vdac1 (C), while Vdac2 mRNA was reduced (D). The calculated efficiency of GPR56-KD showed a reduction of almost $75\%$ of GPR56 mRNA compared to the scramble control group (Supplementary Figure S1A). To evaluate whether GPR56-KD would have any off-target effect, we have also analyzed the mRNA level of a highly expressed GPCR in β-cells i.e., Gprc5b [29,30]. The mRNA level of Gprc5b was not affected by Gpr56-KD, revealing no off-target effect (Supplementary Figure S1B). ## 3.3. The Impact of Gpr56-KD on the Vdac1 Protein Expression and its Mistargeting to the Cell Surface in Mouse Islets Since increased Vdac1 is associated with its mistargeting to the β-cell membrane contributing to β-cell decompensation [14], we next investigated the impact of Gpr56-KD on protein expression and sub-cellular localization of Vdac1 in isolated mouse pancreatic islets by confocal microscopy. Compared with the very low Vdac1 expression in scramble control islets (Figure 3A), Gpr56-KD markedly increased Vdac1 protein expression in the islets (Figure 3B). We studied the co-localization of Vdac1 with the plasma membrane-associated SNARE protein SNAP-25 (cell membrane marker) in islets after Gpr56-KD. Remarkably, confocal microscopy revealed Vdac1 co-localization with SNAP-25 in insulin positive cells, indicating Vdac1 surface localization in β-cells of Gpr56-KD islets (Figure 3B). ## 3.4. The Impact of Gpr56-KD on cAMP, ATP Content, and Insulin Secretion Vdac1 is an ATP-conducting anion channel normally allowing the transport of ATP from mitochondria to the cytoplasm in cells [11,14]. Since immunohistochemical experiments showed an increased VDAC1 expression in pancreatic β-cells upon Gpr56-KD, we next evaluated the effect of two different VDAC1 blockers on the cellular cAMP and ATP content in relation to GSIS in scramble control and Gpr56-KD INS-1 $\frac{832}{13}$ cells. In the following experiments, the cAMP and ATP content, as well as the GSIS, were measured after 60 min incubation of INS-1 cells in the presence or absence of Coll III and VDAC1 inhibitors, i.e., VBIT-4 and VDAC1 antibodies (VD1ab) [14,31]. As shown in Figure 4A, the increased cellular cAMP content induced by high glucose was further augmented by Coll III, but not by the presence of VBIT-4 or VD1-ab during the short-term incubation of scramble control cells. In Gpr56-KD cells, both basal- and glucose-stimulated increases in cAMP were diminished, and the presence of Coll III did not alter cAMP generation. In contrast, the inhibition of VDAC1 with VBIT-4 and VDAC1 antibodies further increased glucose-stimulated cAMP generation in the KD cells. Our data also show that a glucose-stimulated increase in cellular ATP level was not affected by Coll III or by VBIT-4 and VDAC1 antibodies in scramble control INS-1 $\frac{832}{13}$ cells. In Gpr56-KD cells, the diminished glucose-induced rise in ATP level was restored only in the presence of VBIT-4 and VDAC1 antibodies (Figure 4B). To evaluate the impact of altered cellular signaling, we monitored GSIS. Insulin secretion was potentiated by Coll III, while, as expected, VBIT-4 or VDAC1-ab did not alter secretion in scramble control INS-1 $\frac{832}{13}$ cells. In Gpr56-KD cells, the GSIS was markedly reduced and Coll III potentiation was abolished. The presence of VBIT-4 and VDAC1-ab during the final incubation improved GSIS (Figure 4C). As mistargeting of VDAC1 to the plasma membrane causes loss of ATP [14], we next investigated the impact of GPR56-KD on the ATP release in INS-1 cells at 1 mM glucose to avoid high glucose-induced ATP generation [14]. After GPR56-KD, the INS-1 cells were incubated for 60 min at 1 mM of glucose in the presence or absence of Coll III, VBIT-4, and AKOS, which is another VDAC1 inhibitor [14,31]. The basal ATP release from scramble control INS-1 cells was not affected by Coll III, VBIT-4, or AKOS (Figure 5). However, Gpr56-KD was associated with a markedly increased ATP loss from INS-1 $\frac{832}{13}$ cells (Figure 5), which was not affected by the presence of Coll III. It is noteworthy that the ATP release was markedly prevented by VBIT-4 and AKOS (Figure 5). Finally, we extended the observations in INS-1 cells to human EndoC βH1 cells, studying the impact of GPR56-KD on VDAC1 expression, apoptosis, and the release of inflammatory cytokines, as well as cAMP generation and insulin secretion. Among a panel of cytokines (IL-2, IL-6, IL-10, IFNγ, IL-12bp40, IL-12p70, and IL-17) that were analyzed, as seen in Figure 6A–C, GPR56-KD was associated with an increased release of MCP-1 (CCL2), IL-2, and TNF-α from EndoC βH1 cells, while the release of other measured cytokines was undetectable (not shown) in scramble control or GPR56-KD cells. Western blot and immunohistochemical analysis by confocal microscopy revealed that GPR56-KD was associated with increased VDAC1 protein expression (Supplementary Figure S2D). GPR56-KD clearly caused VDAC1 mistargeting to the cell surface, as revealed by co-staining with Na+/K+ATPase in EndoC βH1 cells (Figure 6D,E). GPR56-KD also resulted in an increased intensity of nuclear Hoechst staining, indicating an increased apoptotic rate [32] (Figure 6F). Interestingly, GPR56-KD (Supplementary Figure S2A–C) in EndoC βH1 cells attenuated both GSIS and glucose-induced increase in cAMP when the cells were incubated for 60 min at 20 mM glucose, while basal insulin release or cAMP content was not influenced (Figure 6G,H). The potentiation of GSIS concomitant with the cAMP generation by Coll III was also markedly attenuated by GPR56-KD (Figure 6G,H). These results link β-cell stress after GPR56 loss-of-function to VDAC1-mediated abrogation of ATP formation and the subsequent impairment of stimulus-secretion coupling. ## 4. Discussion The rationale for the present investigations is the documented role of GPR56 in pancreatic β-cell survival and secretory function [18,19,21,25,33]. The expression level of GPR56 is positively correlated with the transcript level of a great number of genes with a beneficial impact on the β-cell fate in human pancreatic islets [18]. Moreover, GPR56 is downregulated in islets from T2D organ donors [18]. Herein, we define the mechanism by which GPR56 loss-of-function causes pancreatic β-cell dysfunctionality. As GPR56 is the most abundant GPR both in mouse and human β-cells [29], it is of interest that not only Coll III derived from islet endothelial cells [19], but also the insulinotropic amino acid L-phenylalanine has been identified as a GPR56 agonist [34]. The stimulation of insulin secretion by L-phenylalanine, a ligand for GPR142, is preserved in GPR142 KO islets, which could be explained by GPR56-mediated signaling [19,35]. The cumulating evidence makes GPR56 an interesting drug target in T2D. We and others have presented evidence that GPR56 is indeed capable of acting as a G-protein-coupling (Gαs type) to elicit downstream signaling cascades via the cAMP/PKA system in pancreatic β-cells [18,21,33], which was confirmed in the present work. A similar signal transduction has been shown for GpR56 activation by testosterone in prostate cells [24]. Although signaling via Gαq has also been reported in β-cells and neuronal cells [19,20,21,36], the cAMP/PKA system could act by lowering the threshold level of the exocytosis for intracellular Ca2+ ([Ca2+]i) elevation, increasing the secretory response of β-cell to even small increases in [Ca2+]i [15]. In the current study, we present mechanistic data revealing GPR56 as a mediator of the suppressive effect of Coll III on cellular stress-related signals, such as P70S6K, JNK, AKT, NFκb, STAT3, and STAT5 signaling, thereby providing the pathway(s) by which GPR56 activation prevents β-cell dysfunction. While it is well-established that stress kinases, NFκB, STAT3, and STAT5 signaling is required for an array of physiological/pathophysiological events, they are mostly involved in stress- or inflammation-induced β-cell dysfunction [6,37]. We show here that GPR56-KD causes cellular stress that also results in the release of certain inflammatory cytokines, such as MCP-1, IL-2, and TNFα by the β-cells. This observation confirms previous studies that GPR56 plays a role in inflammation and has been identified as an inhibitory receptor, suppressing the pro-inflammatory activity of cytotoxic lymphocytes [38]. Likewise, cytokine-induced β-cell apoptosis is prevented by Coll III, but not by a simple overexpression of GPR56 [25]. However, the delineation of the inflammatory mechanism in GPR56-deficient β-cells extends beyond the scope of the present work and merits further investigation. It should also be mentioned that the altered regulation of NFκB, STAT3, and STAT5 plays a critical role in inducing/maintaining the chronic, low-grade inflammation that conveys both β-cell dysfunction [37] and complications of diabetes, including atherosclerotic vascular lesions [39] by influencing diverse cellular gene expression programs. Moreover, we link GPR56-KD to increased expression of the transcription factors Chrebp and Txnip that initiate the overexpression of VDAC1 with a consequent reduction of VDAC2 expression in INS-1 $\frac{832}{13}$ cells. Chrebp and Txnip are highly increased in pancreatic β-cells in glucotoxic condition [40] and in islets from T2D organ donors [14,41]. TXNIP is known to activate the NLRP3 inflammasome, generating interleukin-1β, thereby contributing to impaired β-cell function [37,42]. An increased cAMP formation exerts β-cell protection in part by promoting TXNIP proteosomal degradation [42], an effect that could explain coll III protection from cytokine-induced apoptosis in human islets [25]. The overexpression of VDAC1 and its mistargeting to the β-cell plasma membrane leads to a loss of ATP, the crucial metabolic coupling factor in GSIS [8]. We have shown previously that the prevention of ATP loss by the acute addition of VDAC1 antibodies and inhibitors completely restored the defective GSIS in islets from human T2D organ donors and diabetic db/db mice [14]. Remarkably, confocal microscopy revealed that VDAC1 surface mistargeting occurs in GPR56-KD mouse islet β-cells, as shown by the co-localization with the plasma membrane-associated SNARE protein SNAP-25. A similar VDAC1 mistargeting was observed after GPR56 KD in human EndoC βH1 cells. It is noteworthy that VBIT-4 and AKOS, two chemical VDAC1 inhibitors [14,31], as well as the VDAC1 antibody efficiently restored glucose-induced rises in cellular ATP levels, the generation of cAMP, and the stimulation of insulin secretion in the GPR56 KD cells. We show in INS-1 GPR56-KD cells that this is mainly due to the attenuation of the high rate of ATP leakage. The most plausible explanation for the acute restoration of cAMP generation is the increase in cellular ATP, which drives the cAMP formation during glucose stimulation [43]. In contrast, the stimulatory effects of Coll III are not restored, further validating GPR56 as the collagen receptor. Plasma membrane-resident gated VDAC1 has been documented in various mouse and human tissues with the mitochondrial surface residues facing the extracellular space [44,45,46]. Of note, oxidative stress in neurons activates the conductance of the neurolemmal VDAC1, initiating apoptosis, which is prevented by antibodies directed against the extracellular N-terminus of VDAC1 [45]. Interestingly, the association of GPR56 deletion with the mistargeting of mitochondrial VDAC1 to the cell surface further emphasizes the importance of GPR56 in the regulation of insulin secretion. Loss of GRP56 function in T2D [18] thus participates in altered gene expression and β-cell dysfunction. At first sight, this conclusion seems to be at variance with the only mild glucose intolerance of the GPR56 KO mouse [19]. We speculate that the deletion of GPR56 during fetal development may upregulate other adhesion GPRs, which would not necessarily occur during short-term KD of the receptor. VDAC1 upregulation and oligomerization is caused by oxidative and nitrosative stress, not only in β-cells in T2D but also in neurodegenerative diseases, in particular, Alzheimer’s disease [47,48]. It is, therefore, of great interest that VBIT-4, which inhibits VDAC1 conductance and oligomerization, prevents onset of diabetes in db/db mice [14] and markedly improves the phenotype in a mouse model of Alzheimer’s disease [48]. VBIT-4, thus, has both acute effects on cell signaling by preventing ATP loss through VDAC1 expressed in the plasma membrane and long-term actions on gene expression and cell function in diseases linked to oxidative stress and mitochondrial dysfunction. Taken together, the present data show that GPR56 activation by Coll Type III is associated with the suppression of P70S6K, JNK, AKT, NFκb, STAT3, and STAT5 phosphorylation/activity and increased CREB signaling. Since GPR56 positively modulates the activity of the cAMP-PKA system in the β-cell ([18] and present work), the Coll Type III-mediated beneficial effects on the β-cell function seem to be through cAMP signaling concomitant with the suppression of the aforementioned stress kinases and transcriptional factors. A further signaling pathway implicating GPR56 is its link to integrin function [20,21]. This is relevant for β-cell senescence in which, among others, STAT3 is upregulated [49]. ## 5. Conclusions In the current work, we have linked the dysregulation and mistargeting of the diabetes executer protein VDAC1 [14] to the suppression of GPR56, thereby linking this adhesion GPCR to β-cell mitochondrial dysfunction with impaired ATP accumulation and compromised insulin secretion. GPR56 could, therefore, constitute a novel drug target preventing the loss of β-cell function in prediabetes and diabetes. ## References 1. Weir G.C., Bonner-Weir S.. Five stages of evolving beta-cell dysfunction during progression to diabetes. Diabetes (2004) 53 S16-S21. DOI: 10.2337/diabetes.53.suppl_3.S16 2. Reed J., Bain S., Kanamarlapudi V.. A Review of Current Trends with Type 2 Diabetes Epidemiology, Aetiology, Pathogenesis, Treatments and Future Perspectives. Diabetes Metab. Syndr. Obes. (2021) 14 3567-3602. DOI: 10.2147/DMSO.S319895 3. Wigger L., Barovic M., Brunner A.D., Marzetta F., Schöniger E., Mehl F., Kipke N., Friedland D., Burdet F., Kessler C.. Multi-omics profiling of living human pancreatic islet donors reveals heterogeneous beta cell trajectories towards type 2 diabetes. Nat. Metab. (2021) 3 1017-1031. DOI: 10.1038/s42255-021-00420-9 4. Zhang Z., Gao Y., Meng Z.X.. Transcriptional control of pancreatic beta-cell identity and plasticity during the pathogenesis of type 2 diabetes. J. Genet. Genom. (2022) 49 316-328. DOI: 10.1016/j.jgg.2022.03.002 5. Rohm T.V., Meier D.T., Olefsky J.M., Donath M.Y.. Inflammation in obesity, diabetes, and related disorders. Immunity (2022) 55 31-55. DOI: 10.1016/j.immuni.2021.12.013 6. Wajchenberg B.L.. Beta-cell failure in diabetes and preservation by clinical treatment. Endocr. Rev. (2007) 28 187-218. DOI: 10.1210/10.1210/er.2006-0038 7. Retnakaran R., Choi H., Ye C., Kramer C.K., Zinman B.. Two-year trial of intermittent insulin therapy vs metformin for the preservation of beta-cell function after initial short-term intensive insulin induction in early type 2 diabetes. Diabetes Obes. Metab. (2018) 20 1399-1407. DOI: 10.1111/dom.13236 8. Wiederkehr A., Wollheim C.B.. Mitochondrial signals drive insulin secretion in the pancreatic beta-cell. Mol. Cell. Endocrinol. (2012) 353 128-137. DOI: 10.1016/j.mce.2011.07.016 9. Las G., Oliveira M.F., Shirihai O.S.. Emerging roles of beta-cell mitochondria in type-2-diabetes. Mol. Aspects Med. (2020) 71 100843. DOI: 10.1016/j.mam.2019.100843 10. Madec A.M., Perrier J., Panthu B., Dingreville F.. Role of mitochondria-associated endoplasmic reticulum membrane (MAMs) interactions and calcium exchange in the development of type 2 diabetes. Int. Rev. Cell. Mol. Biol. (2021) 363 169-202. PMID: 34392929 11. Shoshan-Barmatz V., Shteinfer-Kuzmine A., Verma A.. VDAC1 at the Intersection of Cell Metabolism, Apoptosis, and Diseases. Biomolecules (2020) 10. DOI: 10.3390/biom10111485 12. Raghavan A., Sheiko T., Graham B.H., Craigen W.J.. Voltage-dependant anion channels: Novel insights into isoform function through genetic models. Biochim. Biophys. Acta (2012) 1818 1477-1485. DOI: 10.1016/j.bbamem.2011.10.019 13. Ahmed M., Muhammed S.J., Kessler B., Salehi A.. Mitochondrial proteome analysis reveals altered expression of voltage dependent anion channels in pancreatic beta-cells exposed to high glucose. Islets (2010) 2 283-292. DOI: 10.4161/isl.2.5.12639 14. Zhang E., Al-Amily I.M., Mohammed S., Luan C., Asplund O., Ahmed M., Ye Y., Ben-Hail D., Soni A., Vishnu N.. Preserving Insulin Secretion in Diabetes by Inhibiting VDAC1 Overexpression and Surface Translocation in beta Cells. Cell. Metab. (2019) 29 64-77.e6. DOI: 10.1016/j.cmet.2018.09.008 15. Carvalho D.S., de Almeida A.A., Borges A.F., Campos D.V.. Treatments for diabetes mellitus type II: New perspectives regarding the possible role of calcium and cAMP interaction. Eur. J. Pharmacol. (2018) 830 9-16. DOI: 10.1016/j.ejphar.2018.04.002 16. Amisten S., Salehi A., Rorsman P., Jones P.M., Persaud S.J.. An atlas and functional analysis of G-protein coupled receptors in human islets of Langerhans. Pharmacol. Ther. (2013) 139 359-391. DOI: 10.1016/j.pharmthera.2013.05.004 17. Amisten S., Neville M., Hawkes R., Persaud S.J., Karpe F., Salehi A.. An atlas of G-protein coupled receptor expression and function in human subcutaneous adipose tissue. Pharmacol. Ther. (2015) 146 61-93. DOI: 10.1016/j.pharmthera.2014.09.007 18. Duner P., Al-Amily I.M., Soni A., Asplund O., Safi F., Storm P., Groop L., Amisten S.. Adhesion G Protein-Coupled Receptor G1 (ADGRG1/GPR56) and Pancreatic beta-Cell Function. J. Clin. Endocrinol. Metab. (2016) 101 4637-4645. DOI: 10.1210/jc.2016-1884 19. Olaniru O.E., Pingitore A., Giera S., Piao X., Gonzalez R.C., Jones P.M., Persaud S.J.. The adhesion receptor GPR56 is activated by extracellular matrix collagen III to improve beta-cell function. Cell. Mol. Life Sci. (2018) 75 4007-4019. DOI: 10.1007/s00018-018-2846-4 20. Singh A.K., Lin H.H.. The role of GPR56/ADGRG1 in health and disease. Biomed. J. (2021) 44 534-547. DOI: 10.1016/j.bj.2021.04.012 21. Su T., Guan Q., Cheng H., Zhu Z., Jiang C., Guo P., Tai Y., Sun H., Wang M., Wei W.. Functions of G protein-coupled receptor 56 in health and disease. Acta Physiol. (2022) 236 e13866. DOI: 10.1111/apha.13866 22. Giera S., Deng Y., Luo R., Ackerman S.D., Mogha A., Monk K.R., Ying Y., Jeong S.J., Makinodan M., Bialas A.R.. The adhesion G protein-coupled receptor GPR56 is a cell-autonomous regulator of oligodendrocyte development. Nat. Commun. (2015) 6 6121. DOI: 10.1038/ncomms7121 23. Piao X., Hill R.S., Bodell A., Chang B.S., Basel-Vanagaite L., Straussberg R., Dobyns W.B., Qasrawi B., Winter R.M., Innes A.M.. G protein-coupled receptor-dependent development of human frontal cortex. Science (2004) 303 2033-2036. DOI: 10.1126/science.1092780 24. Singh J.P., Dagar M., Dagar G., Kumar S., Rawal S., Sharma R.D., Tyagi R.K., Bagchi G.. Activation of GPR56, a novel adhesion GPCR, is necessary for nuclear androgen receptor signaling in prostate cells. PLoS ONE (2020) 15. DOI: 10.1371/journal.pone.0226056 25. Olaniru O.E., Cheng J., Ast J., Arvaniti A., Atanes P., Huang G.C., King A.J., Jones P.M., Broichhagen J., Hodson D.J.. SNAP-tag-enabled super-resolution imaging reveals constitutive and agonist-dependent trafficking of GPR56 in pancreatic beta-cells. Mol. Metab. (2021) 53 101285. DOI: 10.1016/j.molmet.2021.101285 26. Raoux M., Vacher P., Papin J., Picard A., Kostrzewa E., Devin A., Gaitan J., Limon I., Kas M.J., Magnan C.. Multilevel control of glucose homeostasis by adenylyl cyclase 8. Diabetologia (2015) 58 749-757. DOI: 10.1007/s00125-014-3445-z 27. Al-Amily I.M., Duner P., Groop L., Salehi A.. The functional impact of G protein-coupled receptor 142 (Gpr142) on pancreatic beta-cell in rodent. Pflugers Arch. (2019) 471 633-645. DOI: 10.1007/s00424-019-02262-7 28. Ravassard P., Hazhouz Y., Pechberty S., Bricout-Neveu E., Armanet M., Czernichow P., Scharfmann R.. A genetically engineered human pancreatic beta cell line exhibiting glucose-inducible insulin secretion. J. Clin. Investig. (2011) 121 3589-3597. DOI: 10.1172/JCI58447 29. Amisten S., Atanes P., Hawkes R., Ruz-Maldonado I., Liu B., Parandeh F., Zhao M., Huang G.C., Salehi A., Persaud S.J.. A comparative analysis of human and mouse islet G-protein coupled receptor expression. Sci. Rep. (2017) 7 46600. DOI: 10.1038/srep46600 30. Soni A., Amisten S., Rorsman P., Salehi A.. GPRC5B a putative glutamate-receptor candidate is negative modulator of insulin secretion. Biochem. Biophys. Res. Commun. (2013) 441 643-648. DOI: 10.1016/j.bbrc.2013.10.099 31. Ben-Hail D., Begas-Shvartz R., Shalev M., Shteinfer-Kuzmine A., Gruzman A., Reina S., De Pinto V., Shoshan-Barmatz V.. Novel Compounds Targeting the Mitochondrial Protein VDAC1 Inhibit Apoptosis and Protect against Mitochondrial Dysfunction. J. Biol. Chem. (2016) 291 24986-25003. PMID: 27738100 32. Majtnerova P., Capek J., Petira F., Handl J., Rousar T.. Quantitative spectrofluorometric assay detecting nuclear condensation and fragmentation in intact cells. Sci. Rep. (2021) 11 11921. DOI: 10.1038/s41598-021-91380-3 33. Wang Q., Henry T.A.N., Pronin A.N., Jang G.F., Lubaczeuski C., Crabb J.W., Bernal-Mizrachi E., Slepak V.Z.. The regulatory G protein signaling complex, Gbeta5-R7, promotes glucose- and extracellular signal-stimulated insulin secretion. J. Biol. Chem. (2020) 295 7213-7223. DOI: 10.1074/jbc.RA119.011534 34. Chen H., Nwe P.K., Yang Y., Rosen C.E., Bielecka A.A., Kuchroo M., Cline G.W., Kruse A.C., Ring A.M., Crawford J.M.. A Forward Chemical Genetic Screen Reveals Gut Microbiota Metabolites That Modulate Host Physiology. Cell. (2019) 177 1217-1231.e18. DOI: 10.1016/j.cell.2019.03.036 35. Lin H.V., Efanov A.M., Fang X., Beavers L.S., Wang X., Wang J., Valcarcel I.C.G., Ma T.. GPR142 Controls Tryptophan-Induced Insulin and Incretin Hormone Secretion to Improve Glucose Metabolism. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0157298 36. Little K.D., Hemler M.E., Stipp C.S.. Dynamic regulation of a GPCR-tetraspanin-G protein complex on intact cells: Central role of CD81 in facilitating GPR56-Galpha q/11 association. Mol. Biol. Cell. (2004) 15 2375-2387. DOI: 10.1091/mbc.e03-12-0886 37. Eizirik D.L., Pasquali L., Cnop M.. Pancreatic beta-cells in type 1 and type 2 diabetes mellitus: Different pathways to failure. Nat. Rev. Endocrinol. (2020) 16 349-362. DOI: 10.1038/s41574-020-0355-7 38. Chang G.W., Hsiao C.C., Peng Y.M., Braga F.A.V., Kragten N.A., Remmerswaal E.B., van de Garde M.D., Straussberg R., König G.M., Kostenis E.. The Adhesion G Protein-Coupled Receptor GPR56/ADGRG1 Is an Inhibitory Receptor on Human NK Cells. Cell. Rep. (2016) 15 1757-1770. DOI: 10.1016/j.celrep.2016.04.053 39. Banes-Berceli A.K., Ketsawatsomkron P., Ogbi S., Patel B., Pollock D.M., Marrero M.B.. Angiotensin II and endothelin-1 augment the vascular complications of diabetes via JAK2 activation. Am. J. Physiol. Heart Circ. Physiol. (2007) 293 H1291-H1299. DOI: 10.1152/ajpheart.00181.2007 40. Shalev A.. Minireview: Thioredoxin-interacting protein: Regulation and function in the pancreatic beta-cell. Mol. Endocrinol. (2014) 28 1211-1220. DOI: 10.1210/me.2014-1095 41. Poungvarin N., Lee J.K., Yechoor V.K., Li M.V., Assavapokee T., Suksaranjit P., Thepsongwajja J.J., Saha P.K., Oka K., Chan L.. Carbohydrate response element-binding protein (ChREBP) plays a pivotal role in beta cell glucotoxicity. Diabetologia (2012) 55 1783-1796. DOI: 10.1007/s00125-012-2506-4 42. Shao W., Yu Z., Fantus I.G., Jin T.. Cyclic AMP signaling stimulates proteasome degradation of thioredoxin interacting protein (TxNIP) in pancreatic beta-cells. Cell. Signal. (2010) 22 1240-1246. DOI: 10.1016/j.cellsig.2010.04.001 43. Takahashi N., Kadowaki T., Yazaki Y., Ellis-Davies G.C., Miyashita Y., Kasai H.. Post-priming actions of ATP on Ca. Proc. Natl. Acad. Sci. USA (1999) 96 760-765. DOI: 10.1073/pnas.96.2.760 44. De Pinto V., Messina A., Lane D.J., Lawen A.. Voltage-dependent anion-selective channel (VDAC) in the plasma membrane. FEBS Lett. (2010) 584 1793-1799. PMID: 20184885 45. Akanda N., Tofighi R., Brask J., Tamm C., Elinder F., Ceccatelli S.. Voltage-dependent anion channels (VDAC) in the plasma membrane play a critical role in apoptosis in differentiated hippocampal neurons but not in neural stem cells. Cell. Cycle (2008) 7 3225-3234. DOI: 10.4161/cc.7.20.6831 46. Okada S.F., O’Neal W.K., Huang P., Nicholas R.A., Ostrowski L.E., Craigen W.J., Lazarowski E.R., Boucher R.C.. Voltage-dependent anion channel-1 (VDAC-1) contributes to ATP release and cell volume regulation in murine cells. J. Gen. Physiol. (2004) 124 513-526. DOI: 10.1085/jgp.200409154 47. Atlante A., Valenti D., Latina V., Amadoro G.. Dysfunction of Mitochondria in Alzheimer’s Disease: ANT and VDAC Interact with Toxic Proteins and Aid to Determine the Fate of Brain Cells. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23147722 48. Verma A., Shteinfer-Kuzmine A., Kamenetsky N., Pittala S., Paul A., Nahon Crystal E., Ouro A., Chalifa-Caspi V., Pandey S.K., Monsengo A.. Targeting the overexpressed mitochondrial protein VDAC1 in a mouse model of Alzheimer’s disease protects against mitochondrial dysfunction and mitigates brain pathology. Transl. Neurodegener. (2022) 11 58. DOI: 10.1186/s40035-022-00329-7 49. Midha A., Pan H., Abarca C., Andle J., Carapeto P., Bonner-Weir S., Aguayo-Mazzucato C.. Unique Human and Mouse beta-Cell Senescence-Associated Secretory Phenotype (SASP) Reveal Conserved Signaling Pathways and Heterogeneous Factors. Diabetes (2021) 70 1098-1116. DOI: 10.2337/db20-0553
--- title: TNFR2 as a Potential Biomarker for Early Detection and Progression of CKD authors: - Irina Lousa - Flávio Reis - Sofia Viana - Pedro Vieira - Helena Vala - Luís Belo - Alice Santos-Silva journal: Biomolecules year: 2023 pmcid: PMC10046457 doi: 10.3390/biom13030534 license: CC BY 4.0 --- # TNFR2 as a Potential Biomarker for Early Detection and Progression of CKD ## Abstract The inflammatory pathway driven by TNF-α, through its receptors TNFR1 and TNFR2, is a common feature in the pathogenesis of chronic kidney disease (CKD), regardless of the initial disease cause. Evidence correlates the chronic inflammatory status with decreased renal function. Our aim was to evaluate the potential of TNF receptors as biomarkers for CKD diagnosis and staging, as well as their association with the progression of renal lesions, in rat models of early and moderate CKD. We analyzed the circulating levels of inflammatory molecules—tumor necrosis factor-alpha (TNF-α), tumor necrosis factor receptor 1 (TNFR1) and 2 (TNFR2) and tissue inhibitor of metalloproteinase-1 (TIMP-1)—and studied their associations with TNFR1 and TNFR2 renal expression, glomerular and tubulointerstitial lesions, and with biomarkers of renal (dys)function. An increase in all inflammatory markers was observed in moderate CKD, as compared to controls, but only circulating levels of both TNFR1 and TNFR2 were significantly increased in the early disease; TNFR2 serum levels were negatively correlated with eGFR. However, only TNFR2 renal expression increased with CKD severity and showed correlations with the score of mild and advanced tubular lesions. Our findings suggest that renal TNFR2 plays a role in CKD development, and has potential to be used as a biomarker for the early detection and progression of the disease. Still, the potential value of this biomarker in disease progression warrants further investigation. ## 1. Introduction According to the Global Burden of Disease (GBD) studies, chronic kidney disease (CKD) has emerged as a major cause of global morbidity and mortality, affecting more than 800 million individuals worldwide and imposing major socio-economic costs [1,2]. CKD is an heterogenous condition, with a broad range of underlying etiologies, clinical manifestations and variable progression rates [3]. Inflammation has been implicated in the progression and outcome of CKD, regardless of the initial disease cause. Emerging data suggest an association between biomarkers of inflammation and decreased renal function, as a result of the underlying kidney injury mechanisms [4,5,6]. Therefore, it has been suggested that more sensitive measurements of inflammation could outperform the classical kidney function estimation tools currently used for CKD diagnosis and prognosis [7,8], namely glomerular filtration rate (GFR) and albuminuria assessment. Moreover, the identification of early biomarkers of the inflammatory milieu underlying CKD would allow an early detection and intervention to prevent disease worsening, and reduce the associated socioeconomic costs. Circulating levels of molecules involved in the tumor necrosis factor alpha (TNF-α) pathway, such as TNF receptors 1 and 2 (TNFR1 and TNFR2), were shown to be increased in CKD, in several cohorts of patients, age-groups and races, and with different disease etiologies [9,10,11,12,13]. Recently, our group reported that the increase in the circulating levels of TNFR2 were associated with renal function decline, suggesting that the TNF-α inflammation pathway reflects disease progression [14]; moreover, TNFR2 appeared to be useful for an early detection of CKD, presenting a significantly higher value, as compared to the control group. The tissue inhibitor of metalloproteinase-1 (TIMP-1), an extracellular matrix remodeling regulator, is upregulated in renal interstitial fibrosis [15] and has been linked to the occurrence of renal inflammation in CKD [16]. However, in a mice model with overload proteinuria, the severity of interstitial fibrosis was not changed when the TIMP-1 gene was knocked out [17]. The multiple drug therapies (such as statins, antihypertensive, diuretics, etc.), as well as the prevalence and complexity of comorbidities in CKD patients, represent a constraint when analyzing inflammatory parameters in these populations. Moreover, most of the studies on inflammation biomarkers are limited to individuals with CKD without a biopsy confirmation of cause [9,18,19,20], and therefore, we cannot infer whether those biomarkers provide specificity for kidney histopathologic lesions. Despite being invasive procedures with associated risks, kidney biopsies provide insights on glomerular and tubulointerstitial histology, which are associated with the risk of CKD progression and death [21]. Due to the limitations and ethical issues for these procedures, cellular and tissue renal studies, using appropriate animal models, would be useful to evaluate the sensitivity and specificity of new, potential, earlier and more sensitive CKD biomarkers, than the traditional ones, by searching for correlations between the circulating levels of a biomarker, its renal cell expression and associated kidney lesions. The characterization of novel CKD biomarkers and its association with underlying histopathologic lesions might allow the identification of non-invasive early disease and/or prognostic biomarkers, and also provide information on the clinical phenotyping of kidney diseases. Most published animal studies highlight the importance of TNF-α and its receptors in the development of kidney diseases by demonstrating reduced disease activity with a TNF-α blockade or using knockout models [22,23,24,25]; however, their association with clinical and histopathological findings has not been subject of investigation. In this study, we used rat models of early and moderate CKD, induced by nephrectomy, to characterize inflammatory biomarkers in CKD staging. We measured the serum levels of TNF-α, TNFR1 and TNFR2, and their renal cell expression, and performed kidney histological studies, in rats with different disease staging. Since chronic inflammation has been implicated in the progression and outcome of CKD, we hypothesized that these inflammatory biomarkers might reflect early renal cell and histological changes and, therefore, might be useful for early CKD detection and/or for staging. ## 2.1. Animal Welfare and Experimental Design All rodent experiments were conducted according to the National and European Community Council directives on animal care. The project received approval ($\frac{7}{2020}$) from the local Organization Responsible for Animal Welfare (ORBEA, from Coimbra Institute for Clinical and Biomedical Research, Faculty of Medicine, University of Coimbra) and complied with the Animal Care National and European Directives and with ARRIVE guidelines [26]. Male Wistar rats (Charles River Laboratories, Barcelona, Spain) were housed two per cage and maintained in rooms with a 12 h light/dark cycle, under a controlled temperature (22 °C) and humidity (50–$60\%$). Animals received free access to standard rat laboratory chow (4RF21 Mucedola, Milan, Italy) and water. Body weight was monitored every week. After an adaptation period, rats at 12 weeks of age were randomly divided in three groups (sham, mild CRF and moderate CRF) and submitted to a 5-week protocol. Mild chronic renal failure (Mild CRF, $$n = 8$$) was induced by the complete removal of the left kidney (day 0), and one week later (day 7), a right flank incision was performed without renal mass reduction. Moderate chronic renal failure (moderate CRF, $$n = 7$$) was induced by a two-step $\frac{5}{6}$ nephrectomy, with surgical incision of both poles of the left kidney ($\frac{2}{3}$ nephrectomy) at day 0, and, one week later (day 7), the complete removal of the right kidney. The sham operated group ($$n = 8$$) was subjected to surgical process without kidney mass reduction (days 0 and 7) and used as control. To perform the surgical procedures, rats were subjected to intraperitoneal anesthesia with 75 mg/Kg ketamine (Nimatek, Eurovet Animal Health BV, Bladel, The Netherlands) and 1 mg/Kg medetomidine (Sedator, Dechra, Barcelona, Spain). In the post-surgical period, the animals were kept on recovery blankets with a controlled temperature, and analgesia (0.05 mg/Kg of buprenorphine, Bupaq®, Richter Pharma AG, Austria). Anesthesia was reversed with 2.5 mg/Kg atipamezole (Revazol, Dechra, Barcelona, Spain). ## 2.2. Samples Collection Rats were enclosed in metabolic cages for 24 h before the sacrifice (day 35), with free access to laboratory chow and water, for the collection of 24 h urine. Urine volume and water consumption were recorded, and urine aliquots were stored at −80 °C. At the end of the protocol (day 36), rats were sacrificed using an overdose of intraperitoneal 80 mg/Kg pentobarbital (Sigma-Adrich, Saint-Louis, MO, USA), and blood was collected through cardiac puncture into tubes with K3EDTA or without anticoagulant for hematological and biochemical studies. Serum and plasma aliquots were immediately stored at −80 °C until they were assayed. Kidneys were immediately removed, cleaned and weighed, and stored for further analysis. ## 2.3. Hematological and Biochemical Analysis Red blood cells (RBC), leukocyte, platelet and reticulocytes counts, hemoglobin concentration and hematocrit values were evaluated using an automated cell counter (HORIBA ABX, Amadora, Portugal). The reticulocyte production index (RPI) was calculated as previously described [27]. Blood urea nitrogen (BUN) and serum creatinine were evaluated using validated automated methods and equipment (Hitachi 717 Chemical analyser, Roche Diagnostics, Basel, Switzerland). The urinary levels of creatinine and urea were analyzed in the 24 h urine using automatic methods (Cobas Integra 400 Plus, Roche Diagnostics, Basel, Switzerland). GFR, urea and creatinine clearances were calculated as previously described by Pestel et al. [ 28]. Serum or plasma biomarkers levels were measured with rat-specific ELISA kits, in accordance with the manufacturer’s instructions (TIMP-1 Rat ELISA Kit and Rat TNF-alpha ELISA Kit, Life Technologies, Carlsband, CA, USA; Rat TNFRSF1A ELISA Kit and Rat TNFRSF1B ELISA kit, MyBiosource, San Diego, CA, USA). ## 2.4. Histopathological Analysis Renal tissue samples were formalin-fixed, embedded in paraffin wax, and 4 µm thick sections were stained with Periodic acid–Shiff (PAS) (Sigma Aldrich, Saint Louis, MO, USA). All samples were examined under light microscopy, using a Zeiss microscope Axioplan 2 (Carl Zeiss Microscopy, LLC, NY, USA) and images were captured using a digital microscope camera (Leica DFC450, Leica Microsystems, Wetzlar, Germany) at 400× magnification. Mild and advanced glomerular and tubulointerstitial lesions were identified and evaluated in the total tissue of the slide. The severity of lesions was semi-quantitatively rated according to the extension occupied by the lesion (assessed % over total cortical area or glomeruli affected): 0—absent/normal (<$5\%$); 1—mild (5–$25\%$); 2—moderate (25–$50\%$); 3—severe (>$50\%$). Glomerular hypertrophy was analyzed by measuring the area of ten cortical glomeruli and ten juxtamedullary glomeruli, randomly selected, in each rat (ImageJ processing software, U.S. National Institute of Health, Bethesda, MD, USA). The final score of each lesion was obtained after averaging the individual scores of the animals [29]. Renal pathology evaluation was confirmed by a senior pathologist in a blinded fashion. ## 2.5. Immunohistochemistry Analysis Renal paraffin sections were used for immunohistochemical staining, after incubation with xylene and rehydration with graded ethanol series to water. A mouse- and rabbit-specific horseradish peroxidase (HRP)/diaminobenzidine (DAB) detection kit (ab80436, Abcam Inc, Cambridge, UK) was used according to the manufacturer’s protocol. To retrieve antigen exposure, samples were treated with 10 mM citrate buffer solution at 95 °C, for 30 min. Sections were incubated with primary antibodies at 4 °C overnight in a humidified chamber. Primary antibodies were used for detection of TNF-α (dilution 1:100, PA1-40281), TNFR1 (dilution 1:1000, PA5-95585) and TNFR2 (dilution 1:800, MA5-32618) (Invitrogen, Carlsband, CA, USA). After washing, tissues were dehydrated and counterstained with hematoxylin and mounted with DPX (Merck, Darmstadt, Germany). Negative controls were included, via omission of the primary antibodies. ## 2.6. Protein Analysis For Western blot analysis, kidneys were immediately frozen with liquid nitrogen and stored at −80 °C. Kidney proteins were extracted through homogenization via ultrasonication, using ice-cold RIPA buffer. The homogenates were centrifuged, and the protein concentration of the supernatant was assayed using the bicinchoninic acid (BCA) protein assay kit (Pierce™ BCA, ThermoScientific, Rockford, IL, USA). Aliquots of the extract containing 150–200 µg of proteins were separated via electrophoresis, using $12\%$ SDS-PAGE gels and, afterwards, were transferred into nitrocellulose membranes. Membranes were blocked with $5\%$ non-fat milk and incubated overnight at 4 °C with the same primary antibodies used for immunohistochemistry analysis. Thereafter, membranes were washed and incubated with anti-rabbit secondary antibody (1:1000) HRP-conjugated (sc-2004, Santa Cruz Biotechnology Inc., Dallas, TX, USA). Immunoreactive proteins were detected using the enhanced chemiluminescence method (ECL; WesternBright, Advansta, San Jose, CA, USA), and visualized on a ChemiDoc™ Imaging System (Bio-Rad Laboratories). Finally, the optical density of the bands was determined using Image Lab™ Software (Bio-Rad Laboratories, Hercules, CA, USA). Results were normalized against beta-Actin (1:500, SICGEN, Cantanhede, Portugal) concentrations and an internal control was loaded in all gels. ## 2.7. Statistical Analysis Descriptive statistics are presented as mean ± standard errors of the mean (SEM) or as a count with percentages for categorical variables. Comparison between groups were performed using Kruskal–Wallis or one-way ANOVA. The strength of the association between continuous variables was estimated using Spearman’s correlation coefficient. Statistical significance was accepted at $p \leq 0.05.$ Statistical analysis was performed using the IBM Statistical Package for Social Sciences (SPSS) for Windows, version 26.0 (IBM, Armonk, NY, USA) or GraphPad Prism® software, version 9.4.1 (GraphPad Software, Inc., San Diego, CA, USA). ## 3.1. Body and Kidney Weight At the beginning of the study, all groups showed a similar body weight (BW). During the 5-week protocol, BW variation was negative in animals with moderate CRF, while the other study groups showed a similar increase in BW (Table 1). At the end of the protocol, we observed a significant increase in the right kidney weight (KW), as well as in the relative right KW in the mild CRF group, when compared to the sham group. Despite having just $\frac{1}{3}$ of the left kidney, the left KW and relative left KW in the moderate CRF group were similar to sham (Table 1). ## 3.2. Hematological and Renal Function Data The moderate CRF group presented a consistent decrease in RBC count, hemoglobin and hematocrit values, compared to the sham group. However, only the reduction in hemoglobin concentration reached statistical significance (Table 2). Reticulocyte count and percentage, as well as RPI, were significantly decreased in rats with moderate CKD, compared to rats with mild disease. Serum and urinary renal function data are illustrated in Table 3. At the end of the protocol, there were no significant differences regarding serum and urine measures of creatinine and urea, as well as concerning their clearances, between the sham and mild CRF group. However, all renal function parameters evaluated were aggravated in the moderate CRF group versus the sham group. Even though there was a steady decrease in eGFR alongside CKD severity, only the moderate CRF group’s decline was significant, compared to sham. ## 3.3. Kidney Histomorphology Glomerular and tubulointerstitial lesions were evaluated and quantified (Supplementary Tables S1 and S2, respectively) in kidney sections stained with PAS. At the end of the protocol, we observed the existence of glomerular crescent-like structures in the sham group, a feature that was common to all group animals ($\frac{8}{8}$) at a similar rate of frequency. These structures were previously identified as an early signature of nephropathy, in a rat model of prediabetes [30], but no significant changes between groups were relevant. The overall observation suggested the absence of any other changes in both glomerular and tubulointerstitial structures, despite the presence of inflammatory infiltrate in one out of eight of the sham operated animals, without signs of interstitial fibrosis and atrophy. Regarding the mild CRF group, several mild glomerular and tubulointerstitial lesions were found: all rats ($$n = 8$$) presented some rate of glomerular hypertrophy, dilatation of Bowman’s space and tubular atrophy; however, a significantly increased total score was only observed for mild glomerular lesions, when compared to the sham group. Some advanced glomerular lesions (Figure 1(B1–B4)) were also found in mild CRF, but no significant changes were found in the total score, when compared to sham. The moderate CRF group presented the highest total score of all kidney lesions under study. All rats of this group ($$n = 7$$) presented with mild glomerular and tubulointerstitial lesions, such as glomerular basement membrane thickening, glomerular hypertrophy and dilatation of Bowman’s space. Advanced glomerular lesions, including glomerular “blebbing” ($\frac{4}{7}$), glomerular atrophy ($\frac{7}{7}$) and glomerulosclerosis ($\frac{7}{7}$) were found in the moderate CRF rats. In addition, the moderate CRF group presented hyaline cylinders ($\frac{4}{7}$), vacuolar tubular degeneration ($\frac{4}{7}$), as well as interstitial fibrosis and tubular atrophy (IFTA) ($\frac{7}{7}$), which were absent in all other groups. None of the rats under study presented tubular calcification or necrosis. ## 3.4. Inflammatory Markers Both TNFR1 and TNFR2 showed significantly higher values for the earlier disease stage, the mild CRF group, compared to controls, while the other markers, TNF-α and TIMP-1, presented similar values. Moreover, the serum levels of TNFR2 and TIMP-1 increased with renal function worsening, but only TIMP-1 reached statistical significance (Figure 2). Actually, for both TNF-α and TIMP-1, the highest circulating levels were observed in the more advanced disease stages, showing significantly higher values than those presented by the sham and mild CRF groups. The TNFR2 serum levels were negatively and significantly correlated with eGFR (r = −0.506, $$p \leq 0.014$$) and positively with serum creatinine ($r = 0.675$, $p \leq 0.01$). Mild and advanced glomerular and tubulointerstitial lesions observed were significantly and negatively correlated with eGFR, and significantly and positively correlated with serum levels of all biomarkers, except for TNFR1, which showed no correlation with advanced tubular lesions. Supplementary Figures S1–S4 show the differences in the circulating levels of each biomarker by each histopathologic lesion evaluated. Serum TNFR1 and TNFR2 levels increase with glomerular hypertrophy, dilatation of Bowman space and tubular atrophy ($p \leq 0.01$ for all). However, in more advanced lesions, such as glomerulosclerosis and IFTA, an inverse tendency was observed. No other relevant and clear differences were found in this analysis. No staining was observed on kidney sections when the primary antibodies were omitted (Figure 3(A1,B1,C1)). In the histologically normal kidney, there was occasional weak immunostaining for TNFR1 (Figure 3(B2)) and TNFR2 (Figure 3(C2)) on epithelial tubular cells, but in general, no TNF receptors were detected on renal tubules or glomeruli. TNF-α immunostaining in glomeruli and tubulointerstitium was negative for both sham rats and mild CRF rats, compared with the negative controls (Figure 3(A1–A3)). In the $\frac{5}{6}$ nephrectomy group, the labelling was strongly positive for TNF-α (Figure 3(A4)). A moderate signal for TNFR1 was only evident in tubular epithelial cells, in both CRF groups (Figure 3(B3,B4)). Considering TNFR2 immunolabeling, we found an increased intensity on renal tubules with a worsening disease (Figure 3(C3,C4)). These results from the immunohistochemistry were supported by Western blotting analysis (Figure 4). Accordingly, there was a steady increase in TNFR2 renal expression, but only reaching statistical significance in moderate disease, when compared to controls. No significant changes in TNFR1 renal expression were detected. TNFR2 renal expression levels showed correlations with the score of mild and advanced tubular lesions ($r = 0.501$, $$p \leq 0.015$$ and $r = 0.5518$, $$p \leq 0.011$$, respectively), and also with eGFR (r = −0.442, $$p \leq 0.035$$). ## 4. Discussion During the last few decades, several advances have occurred in the management of CKD patients; however, morbidity and mortality rates are still high, compared to the general population [1,31]. Chronic inflammation is the driving force in the progression of kidney diseases and play a key role in the development of its associated comorbidities [32,33]. The inflammatory process begins in the early stages of the disease, regardless of its cause, eventually leading to renal fibrosis and end-stage renal disease (ESRD) [33]. In this study, we evaluated several inflammatory markers related to TNF signaling (TNF-α, TNFR1 and TNFR2) and the inhibition of metalloproteinases (TIMP-1), and analyzed kidney histopathologic lesions in rats with mild and moderate disease. After 5 weeks of nephrectomy, the mild CRF group rats developed an early degree of renal insufficiency. The surgical removal of the left kidney induced a compensatory hypertrophy of the right kidney, confirmed by the increased RKW/BW ratio, which resulted in non-significant alterations of serum and urinary measures of renal function, and by the absence of anemia. However, several mild glomerular and tubulointerstitial lesions were found, with a significantly increased total score only for mild glomerular lesions, when compared to the sham group. The absence of differences in creatinine, and subsequently eGFR, is a known limitation of these classical CKD biomarkers in early disease detection, since it does not change until kidney function is substantially impaired [34,35]. The surgical $\frac{5}{6}$ nephrectomy model is a well-established model of moderate, but sustained CKD [36]. Despite having only $\frac{1}{3}$ of the left kidney, the moderate CRF group showed a trend to increased KW, consistent with a compensational status, since the ratio of KW/BW was similar to the sham group. This increase in KW was accompanied by a deterioration in kidney function, as we found a significant increase in serum urea and creatinine, alongside decreased urinary levels and clearance, as well as decreased eGFR. Additionally, several advanced renal lesions were observed in both glomeruli and the tubulointerstitial area. In this group, the negative change in BW may be related to a loss of appetite, due to uremic milieu and anemia. Anemia is a known complication of CKD and results primarily from decreased erythropoietin synthesis in the kidney. This is in line with our results, demonstrating a lower hemoglobin concentration and a reduction in RPI in moderate CKD. The TNF-α signaling pathway, through its receptors TNFR1 and TNFR2, seems to be essential in renal function deterioration [37]. However, the mechanisms through which the TNFRs initiate and perpetuate renal damage are not completely understood [38]. In addition to the various studies addressing the association of TNFRs with human kidney disease and related clinical outcomes [18,39,40,41], several animal studies highlight their role in the pathophysiology of kidney diseases [24,25,42]. In clinical studies, soluble TNF receptors 1 and 2 were showed to be increased in several cohorts of diabetic patients. Accordingly, our group also reported increased levels of TNFR2 in patients with decreased eGFR and suggested their potential as a biomarker of early CKD, as well as disease staging/worsening in a cohort of CKD patients with diverse etiologies [14]. We found a similar pattern for this biomarker in the present study. The circulating levels of both TNFR1 and TNFR2 were significantly increased in early disease, the mild CRF group, while the other markers, TNF-α and TIMP-1, were not increased at this stage. However, only TNFR2 serum levels seem to increase steadily with the worsening of renal function, as documented by the inverse correlation with eGFR. A steady increase in TNFR2 renal expression was also found, but only reaching statistical significance in moderate disease, when compared to controls, and no significant changes in TNFR1 renal expression were detected. In addition, TNFR2 renal expression showed correlations with the score of mild and advanced tubular lesions. These differences are likely explained by the different biological actions of the two receptors, which engage shared and different downstream signaling pathways [37]. In accordance with our findings, renal TNFR2 expression, not TNFR1, seems to mediate the development of glomerulonephritis [24] and cisplatin-induced acute renal failure [42] in mice models. As reported by Al-Lamki et al., TNFR1 seems to be basally expressed in the normal kidney, while TNFR2 expression is upregulated after renal injury and is also mostly expressed in immune cells [43]. TNFR1 is mostly implicated in tissue inflammation and injury; however, TNFR2 promotes epithelial-to-mesenchymal transition and cell proliferation, and has been mainly associated with the nuclear factor kappa B (NF-кB) pathway, being implicated in immune modulation and tissue regeneration [44]. As we previously hypothesized [14], the greater expression of renal TNFR2 by the activated leukocytes and injured renal cells activates the NF-кB mediated pathway, through TNF-α signaling, to promote injury resolution. Thus, TNFR2 seems to act not only as an early biomarker of renal damage but also as a mediator of the disease. Furthermore, we hypothesized that these inflammatory biomarkers might correlate with histological findings, since most of the published literature on TNFR1 and TNFR2 has focused on CKD patients without biopsy confirmation. Our cross-sectional analysis showed that serum levels of TNFR1 and TNFR2 are correlated with the total score of mild glomerular lesions, particularly glomerular hypertrophy and dilatation of Bowman’s space, and also tubular atrophy. Plasma TNFR1 and TNFR2 have been previously associated with early glomerular lesions in type 2 diabetic patients [45] and with tubulointerstitial lesions in patients across a diverse set of kidney diseases [39,46]. However, Niewczas et al. reported no correlations of renal mRNA expression of TNFR1 and TNFR2 in advanced lesions, such as glomerular sclerosis and IFTA, in patients with diabetic nephropathy [40]. The diverse range of kidney diseases, the prevalence and complexity of associated comorbidities, and the multiple drug therapies are major obstacles when assessing inflammatory parameters in CKD patients. In this study, we used rat models of mild and moderate CKD, which allowed us to study these disease stages without the heterogeneity and individual variability that is observed in CKD patients. However, our study has some limitations that should be addressed. First, the cross-sectional design does not allow us to fully characterize biomarkers with renal function deterioration. Additionally, the results were obtained with a specific model of CKD, and thus, might not be reproducible in other conditions. In conclusion, our findings suggest that TNFR2 seems to play an early role in the renal cell damage underlying CKD development, and to have potential as biomarker for an earlier detection, as compared to classic biomarkers of renal (dys)function; moreover, it has potential as a biomarker of disease progression, as reflected by the circulating levels. Therefore, TNFR2 is indicated as a new therapeutic target and its blockade may be a promising strategy in the treatment of CKD. Further mechanistic studies would be necessary to understand the prognostic value of this biomarker. ## References 1. Kovesdy C.P.. Epidemiology of chronic kidney disease: An update 2022. Kidney Int. Suppl. (2022) 12 7-11. DOI: 10.1016/j.kisu.2021.11.003 2. Jager K.J., Kovesdy C., Langham R., Rosenberg M., Jha V., Zoccali C.. A single number for advocacy and communication-worldwide more than 850 million individuals have kidney diseases. Nephrol. Dial. Transpl. (2019) 34 1803-1805. DOI: 10.1093/ndt/gfz174 3. Levin A., Stevens P.E., Bilous R.W., Coresh J., De Francisco A.L.M., De Jong P.E., Griffith K.E., Hemmelgarn B.R., Iseki K., Lamb E.J.. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int. Suppl. (2012) 3 5-14. DOI: 10.1111/j.1600-6143.2009.02834.x 4. Bash L.D., Erlinger T.P., Coresh J., Marsh-Manzi J., Folsom A.R., Astor B.C.. Inflammation, Hemostasis, and the Risk of Kidney Function Decline in the Atherosclerosis Risk in Communities (ARIC) Study. Am. J. Kidney Dis. (2009) 53 596-605. DOI: 10.1053/j.ajkd.2008.10.044 5. Gupta J., Mitra N., Kanetsky P.A., Devaney J., Wing M.R., Reilly M., Shah V.O., Balakrishnan V.S., Guzman N.J., Girndt M.. Association between Albuminuria, Kidney Function, and Inflammatory Biomarker Profile in CKD in CRIC. Clin. J. Am. Soc. Nephrol. (2012) 7 1938. DOI: 10.2215/CJN.03500412 6. Tinti F., Lai S., Noce A., Rotondi S., Marrone G., Mazzaferro S., Di Daniele N., Mitterhofer A.P.. Chronic Kidney Disease as a Systemic Inflammatory Syndrome: Update on Mechanisms Involved and Potential Treatment. Life (2021) 11. DOI: 10.3390/life11050419 7. Amdur R.L., Feldman H.I., Dominic E.A., Anderson A.H., Beddhu S., Rahman M., Wolf M., Reilly M., Ojo A., Townsend R.R.. Use of Measures of Inflammation and Kidney Function for Prediction of Atherosclerotic Vascular Disease Events and Death in Patients with CKD: Findings from the CRIC Study. Am. J. Kidney Dis. (2019) 73 344-353. DOI: 10.1053/j.ajkd.2018.09.012 8. Puthumana J., Thiessen-Philbrook H., Xu L., Coca S.G., Garg A.X., Himmelfarb J., Bhatraju P.K., Ikizler T.A., Siew E.D., Ware L.B.. Biomarkers of inflammation and repair in kidney disease progression. J. Clin. Investig. (2021) 131 e139927. DOI: 10.1172/JCI139927 9. Keller C.R., Odden M.C., Fried L.F., Newman A.B., Angleman S., Green C.A., Cummings S.R., Harris T.B., Shlipak M.G.. Kidney function and markers of inflammation in elderly persons without chronic kidney disease: The health, aging, and body composition study. Kidney Int. (2007) 71 239-244. DOI: 10.1038/sj.ki.5002042 10. Zwiech R.. Predictive value of conjointly examined IL-1ra, TNF-R I, TNF-R II, and RANTES in patients with primary glomerulonephritis. J. Korean Med. Sci. (2013) 28 261-267. DOI: 10.3346/jkms.2013.28.2.261 11. Murakoshi M., Gohda T., Sonoda Y., Suzuki H., Tomino Y., Horikoshi S., Suzuki Y.. Effect of tonsillectomy with steroid pulse therapy on circulating tumor necrosis factor receptors 1 and 2 in IgA nephropathy. Clin. Exp. Nephrol. (2017) 21 1068-1074. DOI: 10.1007/s10157-017-1408-7 12. Liu G., Deng Y., Sun L., Ye X., Yao P., Hu Y., Wang F., Ma Y., Li H., Liu Y.. Elevated plasma tumor necrosis factor-α receptor 2 and resistin are associated with increased incidence of kidney function decline in Chinese adults. Endocrine (2016) 52 541-549. DOI: 10.1007/s12020-015-0807-3 13. Martin W.P., Conroy C., Naicker S.D., Cormican S., Griffin T.P., Islam M.N., McCole E.M., McConnell I., Lamont J., FitzGerald P.. Multiplex Serum Biomarker Assays Improve Prediction of Renal and Mortality Outcomes in Chronic Kidney Disease. Kidney360 (2021) 2 1225-1239. DOI: 10.34067/KID.0007552020 14. Lousa I., Belo L., Valente M.J., Rocha S., Preguiça I., Rocha-Pereira P., Beirão I., Mira F., Alves R., Reis F.. Inflammatory biomarkers in staging of chronic kidney disease: Elevated TNFR2 levels accompanies renal function decline. Inflamm. Res. (2022) 71 591-602. DOI: 10.1007/s00011-022-01574-2 15. Yang B., Vohra P.K., Janardhanan R., Misra K.D., Misra S.. Expression of profibrotic genes in a murine remnant kidney model. J. Vasc. Interv. Radiol. (2011) 22 1765-1772.e1761. DOI: 10.1016/j.jvir.2011.08.026 16. Cheng Z., Limbu M.H., Wang Z., Liu J., Liu L., Zhang X., Chen P., Liu B.. MMP-2 and 9 in Chronic Kidney Disease. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18040776 17. Eddy A.A., Kim H., López-Guisa J., Oda T., Soloway P.D.. Interstitial fibrosis in mice with overload proteinuria: Deficiency of TIMP-1 is not protective. Kidney Int. (2000) 58 618-628. DOI: 10.1046/j.1523-1755.2000.00208.x 18. Coca S.G., Nadkarni G.N., Huang Y., Moledina D.G., Rao V., Zhang J., Ferket B., Crowley S.T., Fried L.F., Parikh C.R.. Plasma Biomarkers and Kidney Function Decline in Early and Established Diabetic Kidney Disease. J. Am. Soc. Nephrol. (2017) 28 2786-2793. DOI: 10.1681/ASN.2016101101 19. Barr E.L.M., Barzi F., Hughes J.T., Jerums G., Hoy W.E., O’Dea K., Jones G.R.D., Lawton P.D., Brown A.D.H., Thomas M.. High Baseline Levels of Tumor Necrosis Factor Receptor 1 Are Associated with Progression of Kidney Disease in Indigenous Australians with Diabetes: The eGFR Follow-up Study. Diabetes Care (2018) 41 739-747. DOI: 10.2337/dc17-1919 20. Gohda T., Maruyama S., Kamei N., Yamaguchi S., Shibata T., Murakoshi M., Horikoshi S., Tomino Y., Ohsawa I., Gotoh H.. Circulating TNF Receptors 1 and 2 Predict Mortality in Patients with End-stage Renal Disease Undergoing Dialysis. Sci. Rep. (2017) 7 43520. DOI: 10.1038/srep43520 21. Schmidt I.M., Sarvode Mothi S., Wilson P.C., Palsson R., Srivastava A., Onul I.F., Kibbelaar Z.A., Zhuo M., Amodu A., Stillman I.E.. Circulating Plasma Biomarkers in Biopsy-Confirmed Kidney Disease. Clin. J. Am. Soc. Nephrol. (2022) 17 27-37. DOI: 10.2215/CJN.09380721 22. Meldrum K.K., Misseri R., Metcalfe P., Dinarello C.A., Hile K.L., Meldrum D.R.. TNF-alpha neutralization ameliorates obstruction-induced renal fibrosis and dysfunction. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2007) 292 R1456-R1464. DOI: 10.1152/ajpregu.00620.2005 23. Guo G., Morrissey J., McCracken R., Tolley T., Klahr S.. Role of TNFR1 and TNFR2 receptors in tubulointerstitial fibrosis of obstructive nephropathy. Am. J. Physiol. (1999) 277 F766-F772. DOI: 10.1152/ajprenal.1999.277.5.F766 24. Vielhauer V., Stavrakis G., Mayadas T.N.. Renal cell-expressed TNF receptor 2, not receptor 1, is essential for the development of glomerulonephritis. J. Clin. Investig. (2005) 115 1199-1209. DOI: 10.1172/JCI200523348 25. Chen C.C., Pedraza P.L., Hao S., Stier C.T., Ferreri N.R.. TNFR1-deficient mice display altered blood pressure and renal responses to ANG II infusion. Am. J. Physiol. Ren. Physiol. (2010) 299 F1141-F1150. DOI: 10.1152/ajprenal.00344.2010 26. Percie du Sert N., Hurst V., Ahluwalia A., Alam S., Avey M.T., Baker M., Browne W.J., Clark A., Cuthill I.C., Dirnagl U.. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. PLoS Biol. (2020) 18. DOI: 10.1371/journal.pbio.3000410 27. Ribeiro S., Garrido P., Fernandes J., Vala H., Rocha-Pereira P., Costa E., Belo L., Reis F., Santos-Silva A.. Renal risk-benefit determinants of recombinant human erythropoietin therapy in the remnant kidney rat model—Hypertension, anaemia, inflammation and drug dose. Clin. Exp. Pharm. Physiol. (2016) 43 343-354. DOI: 10.1111/1440-1681.12541 28. Pestel S., Krzykalla V., Weckesser G.. Measurement of glomerular filtration rate in the conscious rat. J. Pharm. Toxicol. Methods (2007) 56 277-289. DOI: 10.1016/j.vascn.2007.03.001 29. Garrido P., Ribeiro S., Fernandes J., Vala H., Bronze-da-Rocha E., Rocha-Pereira P., Belo L., Costa E., Santos-Silva A., Reis F.. Iron-hepcidin dysmetabolism, anemia and renal hypoxia, inflammation and fibrosis in the remnant kidney rat model. PLoS ONE (2015) 10. DOI: 10.1371/journal.pone.0124048 30. Nunes S., Alves A., Preguiça I., Barbosa A., Vieira P., Mendes F., Martins D., Viana S.D., Reis F.. Crescent-Like Lesions as an Early Signature of Nephropathy in a Rat Model of Prediabetes Induced by a Hypercaloric Diet. Nutrients (2020) 12. DOI: 10.3390/nu12040881 31. Crews D.C., Bello A.K., Saadi G.. Burden, access, and disparities in kidney disease. Kidney Int. (2019) 95 242-248. DOI: 10.1016/j.kint.2018.11.007 32. Andrade-Oliveira V., Foresto-Neto O., Watanabe I.K.M., Zatz R., Câmara N.O.S.. Inflammation in Renal Diseases: New and Old Players. Front. Pharm. (2019) 10 1192. DOI: 10.3389/fphar.2019.01192 33. Akchurin O.M., Kaskel F.. Update on inflammation in chronic kidney disease. Blood Purif. (2015) 39 84-92. DOI: 10.1159/000368940 34. Waikar S.S., Betensky R.A., Bonventre J.V.. Creatinine as the gold standard for kidney injury biomarker studies?. Nephrol. Dial. Transpl. (2009) 24 3263-3265. DOI: 10.1093/ndt/gfp428 35. Ferguson M.A., Waikar S.S.. Established and emerging markers of kidney function. Clin. Chem. (2012) 58 680-689. DOI: 10.1373/clinchem.2011.167494 36. Garrido P., Reis F., Costa E., Teixeira-Lemos E., Parada B., Alves R., Piloto N., Sereno J., Figueiredo A., Pinto R.. Characterization of a Rat Model of Moderate Chronic Renal Failure—Focus on Hematological, Biochemical, and Cardio-Renal Profiles. Ren. Fail. (2009) 31 833-842. DOI: 10.3109/08860220903151435 37. Lousa I., Reis F., Santos-Silva A., Belo L.. The Signaling Pathway of TNF Receptors: Linking Animal Models of Renal Disease to Human CKD. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23063284 38. Al-Lamki R.S., Mayadas T.N.. TNF receptors: Signaling pathways and contribution to renal dysfunction. Kidney Int. (2015) 87 281-296. DOI: 10.1038/ki.2014.285 39. Srivastava A., Schmidt I.M., Palsson R., Weins A., Bonventre J.V., Sabbisetti V., Stillman I.E., Rennke H.G., Waikar S.S.. The Associations of Plasma Biomarkers of Inflammation with Histopathologic Lesions, Kidney Disease Progression, and Mortality-The Boston Kidney Biopsy Cohort Study. Kidney Int. Rep. (2021) 6 685-694. DOI: 10.1016/j.ekir.2020.12.025 40. Niewczas M.A., Pavkov M.E., Skupien J., Smiles A., Md Dom Z.I., Wilson J.M., Park J., Nair V., Schlafly A., Saulnier P.J.. A signature of circulating inflammatory proteins and development of end-stage renal disease in diabetes. Nat. Med. (2019) 25 805-813. DOI: 10.1038/s41591-019-0415-5 41. Gutiérrez O.M., Shlipak M.G., Katz R., Waikar S.S., Greenberg J.H., Schrauben S.J., Coca S., Parikh C.R., Vasan R.S., Feldman H.I.. Associations of Plasma Biomarkers of Inflammation, Fibrosis, and Kidney Tubular Injury with Progression of Diabetic Kidney Disease: A Cohort Study. Am. J. Kidney Dis. (2022) 79 849-857.e841. DOI: 10.1053/j.ajkd.2021.09.018 42. Ramesh G., Reeves W.B.. TNFR2-mediated apoptosis and necrosis in cisplatin-induced acute renal failure. Am. J. Physiol. Ren. Physiol. (2003) 285 F610-F618. DOI: 10.1152/ajprenal.00101.2003 43. Al-Lamki R.S., Wang J., Skepper J.N., Thiru S., Pober J.S., Bradley J.R.. Expression of Tumor Necrosis Factor Receptors in Normal Kidney and Rejecting Renal Transplants. Lab. Investig. (2001) 81 1503-1515. DOI: 10.1038/labinvest.3780364 44. Gough P., Myles I.A.. Tumor Necrosis Factor Receptors: Pleiotropic Signaling Complexes and Their Differential Effects. Front. Immunol. (2020) 11 585880. DOI: 10.3389/fimmu.2020.585880 45. Pavkov M.E., Weil E.J., Fufaa G.D., Nelson R.G., Lemley K.V., Knowler W.C., Niewczas M.A., Krolewski A.S.. Tumor necrosis factor receptors 1 and 2 are associated with early glomerular lesions in type 2 diabetes. Kidney Int. (2016) 89 226-234. DOI: 10.1038/ki.2015.278 46. Sonoda Y., Gohda T., Suzuki Y., Omote K., Ishizaka M., Matsuoka J., Tomino Y.. Circulating TNF receptors 1 and 2 are associated with the severity of renal interstitial fibrosis in IgA nephropathy. PLoS ONE (2015) 10. DOI: 10.1371/journal.pone.0122212
--- title: Prognostic Value of Combined Hematological/Biochemical Indexes and Tumor Clinicopathologic Features in Colorectal Cancer Patients—A Pilot Single Center Study authors: - Vladica Cuk - Aleksandar Karamarkovic - Jovan Juloski - Dragana Arbutina - Radosav Radulovic - Ljiljana Milic - Bojan Kovacevic - Silvio De Luka - Jelena Grahovac journal: Cancers year: 2023 pmcid: PMC10046459 doi: 10.3390/cancers15061761 license: CC BY 4.0 --- # Prognostic Value of Combined Hematological/Biochemical Indexes and Tumor Clinicopathologic Features in Colorectal Cancer Patients—A Pilot Single Center Study ## Abstract ### Simple Summary Colorectal cancer (CRC) is a growing health burden in Serbia and worldwide. Surgical resection is the main modality for CRC treatment, and adjuvant treatment can further reduce the frequency of disease relapse and improve overall survival. Our study presents evidence that standard laboratory parameters, which do not present any additional cost for the health system, may provide additional information on the CRC patient outcome and lay the groundwork for a larger prospective examination. In our patient cohort, Clavien–Dindo classification of post-operative complications, modified Glasgow prognostic score, lymph node ratio, tumor deposits and peritumoral lymphocyte response were factors that were significantly associated with survival of operated patients. ### Abstract Colorectal cancer (CRC) is a significant public health problem. There is increasing evidence that the host’s immune response and nutritional status play a role in the development and progression of cancer. The aim of our study was to examine the prognostic value of clinical markers/indexes of inflammation, nutritional and pathohistological status in relation to overall survival and disease free-survival in CRC. The total number of CRC patients included in the study was 111 and they underwent laboratory analyses within a week before surgery. Detailed pathohistological analysis and laboratory parameters were part of the standard hospital pre-operative procedure. Medical data were collected from archived hospital data. Data on the exact date of death were obtained by inspecting the death registers for the territory of the Republic of Serbia. All parameters were analyzed in relation to the overall survival and survival period without disease relapse. The follow-up median was 42 (24−48) months. The patients with the III, IV and V degrees of the Clavien–Dindo classification had 2.609 (HR: 2.609; $95\%$ CI: 1.437−4.737; $$p \leq 0.002$$) times higher risk of death. The modified Glasgow prognostic score (mGPS) 2 and higher lymph node ratio carried a 2.188 (HR: 2.188; $95\%$ CI: 1.413−3.387; $p \leq 0.001$) and 6.862 (HR: 6.862; $95\%$ CI: 1.635−28.808; $$p \leq 0.009$$) times higher risk of death in the postoperative period, respectively; the risk was 3.089 times higher (HR: 3.089; $95\%$ CI: 1.447−6.593; $$p \leq 0.004$$) in patients with verified tumor deposits. The patients with tumor deposits had 1.888 (HR: 1.888; $95\%$ CI: 1024−3481; $$p \leq 0.042$$) and 3.049 (HR: 3.049; $95\%$ CI: 1.206−7.706; $$p \leq 0.018$$) times higher risk of disease recurrence, respectively. The emphasized peritumoral lymphocyte response reduced the risk of recurrence by $61\%$ (HR: 0.391; $95\%$ CI: 0.196−0.780; $$p \leq 0.005$$). Standard perioperative laboratory and pathohistological parameters, which do not present any additional cost for the health system, may provide information on the CRC patient outcome and lay the groundwork for a larger prospective examination. ## 1. Introduction Colorectal cancer (CRC) is a significant public health problem. Worldwide, in 2020, there were 1.9 million newly diagnosed CRC patients and 935,000 deaths from colorectal cancer [1]. Globally, every tenth newly ill and deceased patient with a malignant tumor had colorectal cancer. CRC is one of the most common cancers (after breast and lung cancer), with a $10\%$ share of all malignancies, and the second leading cause of cancer death (after lung cancer) with a $9.4\%$ share of all cancer-related deaths [1]. Surgical treatment is the main modality of CRC treatment [2]. Although it is generally known that tumor grade, degree of differentiation, histological (sub)type of tumor, tumor localization, number of positive lymph nodes and disease stage are good indicators of one-year, three-year and five-year survival rate, there are still controversies about prediction of colorectal cancer outcomes. There is no completely clear explanation for why patients with stage IIIa disease have better three-year and five-year survival rate than those in stage IIb (3-year survival: $91.4\%$ vs. $80.2\%$; 5-year survival: $83.4\%$ vs. $72.2\%$), which calls into question current treatment protocols [3]. Answers to these questions can be offered by recent research whose results go beyond generally accepted understanding of the biology of colorectal cancer [3,4]. There is increasing evidence that the host’s immune response and the nutritional status play a role in development and progression of cancer [5]. Taking into account the mechanisms of inflammation and the nutritional status of patients, great efforts are being made to find effective, easily accessible and cheap predictors of colorectal cancer prognosis in order to facilitate identification of critically ill patients, their postoperative monitoring and timely treatment. Several studies indicate that preoperative values of hematological/biochemical parameters (e.g., absolute values of lymphocytes, neutrophils, monocytes, platelets, values of serum albumin, C-reactive protein) and their mutual integration into indexes and scores are good indicators of the prognosis of colorectal cancer [6,7,8,9]. The LANR- specific ratio of absolute values of lymphocytes, neutrophils and albumin [6], PNI- prognostic nutritional index that integrates absolute values of lymphocytes and albumin [7], CAR- ratio of serum values of C-reactive protein and albumin [8], and mGPS- modified Glasgow prognostic score based on serum C-reactive protein and albumin [9], have been promising in predicting the patient outcome. The majority of these findings stem from large cohorts of patients of the Asian population [6,7,8,10,11]. In the Republic of Serbia, the TNM stage of CRC and basic pathohistological findings, are taken as dominant indicators of prognosis and are often the only factors considered in deciding on adjuvant therapy. Preoperative nutritional status, standard hematological/biochemical parameters, and their integration into indexes in CRC patients have not been taken into account in CRC therapy decision making in Serbia to date. Our assumption was that the clinical laboratory results and the pathohistological status of tumor, which are generally available in every health institution around the world, present clinically “tangible” evidence of the subtle mechanisms of the immune system at the cellular and subcellular level that can help more precise prognosis for CRC patients and better-informed therapy decision making. In this regard, the aim of our study was to examine the clinical markers/indexes of inflammation, nutritional and pathohistological status in relation to overall survival (OS) and disease-free survival (DFS) in Serbian colorectal cancer patients. ## 2.1. Patient Population A database of 120 patients with diagnosed colorectal adenocarcinoma, confirmed by pathohistological findings, was retrospectively examined. All patients were operated on at the Surgical Clinic “Nikola Spasic” of Zvezdara University Medical Centre, in the period from January 2017 to December 2017. Excluding criteria were: [1] Presence of other malignancies treated in a period of less than 5 years preceding the time of surgery due to CRC; [2] Incomplete medical documentation; [3] Inadequate postoperative follow-up. Taking into account the exclusion criteria, after final processing a total of 111 patients met the study requirements. The stage of the disease was determined on the basis of the Eighth Edition of the American Joint Committee on Cancer (AJCC) Cancer Staging Manual [12]. The research was approved by the ethics board of the Zvezdara University Medical Centre in Belgrade, and all patients gave their written consent to participate in the research. ## 2.2. Patient Characteristics Data on gender, age of patients, total length of hospitalization and length of hospitalization after surgical treatment were collected. The ASA (American Society of Anesthesiology) score was calculated for all patients, and postoperative complications were expressed through the Clavien–Dindo classification (C-D classification). ## 2.3. Preoperative Laboratory Measurements and Other Prognostic Scores Data from standard biochemical and hematological analyses, performed in the hospital laboratory on Roche analyzers Cobas 6000 (c501 and e601) (Roche Diagnostics GmbH, Mannheim, Germany) and Sysmex XN-1000 (Sysmex Europe SE, Norderstedt, Germany) were issued by the hospital laboratory. In addition to the standard laboratory analyses, data on the preoperative values of sex hormones (Estradiol and Testosterone) and morning Cortisol as well as tumor markers (CEA, CA 19-9) were collected. All analyses were performed within a week before the operative treatment. Integrated hematological and biochemical parameters were calculated based on formulas used in previous studies: NLR (Neutrophile to Lymphocyte Ratio); MLR (Monocyte to Lymphocyte Ratio); PLR (Platelets to Lymphocyte Ratio); RLR (RBC to Lymphocyte Ratio); MPR (MPV to Platelets Ratio); CAR = CRP (mg/dL)/Serum albumin (g/dL); modified Glasgow Prognostic Score (mGPS): 0 (CRP ≤ 10, Alb ≥ 35), 1 (CRP ≤ 10, Alb < 35; CRP > 10, Alb ≥ 35), 2 (CRP > 10, Alb < 35); PNI (prognostic nutritional index) = Albumin value(g/L) + 5 × Lymphocyte (109/L); LANR = Lymphocyte (109/L) × Albumin (g/L)/Neutrophil (109/L) [6,7,9,13]. ## 2.4. Tumor Characteristics Data on tumor localization, disease dissemination and type of surgery were obtained from the operative findings. Surgical preparations were analyzed by one pathologist and each pathohistological finding included: macroscopic description of the tumor, TNM classification, Astler–Coller classification, stage of the disease, tumor configuration, tumor size, macroscopic perforation of the tumor, macroscopic intactness of mesorectal fascia, histological type, histological grade, TIL (tumor-infiltrating lymphocytes), peritumoral lymphocyte (PTL) response of the tumor, mucinous component of the tumor, circumferential resection margin (in rectal cancer), lymphovascular, venous and perineural invasion, tumor deposits, total number of lymph nodes, number of positive lymph nodes, lymph node ratio, tumor budding. ## 2.5. Follow-Up At the Zvezdara University Medical Centre, postoperative follow-up was based on The National Guide of Good Clinical Practice for Colorectal Cancer issued by The Ministry of Health of the Republic of Serbia [14]. The national guide was modeled after the guide designed by Pfister et al. and represents a compromise between more and less intensive follow-up of patients after surgical treatment of CRC [15]. The overall patient survival (OS) was calculated as the date of diagnosis to the date of death from any cause. The disease-free patient survival (DFS) was defined as the time interval from cancer primary treatment until tumor recurrence or death from any cause. Tumor recurrence was defined as any clinically, biochemically or radiologically inferred relapse of the disease. Data on neoadjuvant, adjuvant chemo/radiation therapy and the disease recurrence for the included patients were collected from archived data at the Institute for Oncology and Radiology of Serbia in Belgrade and postoperative follow-up at the primary care health institution. In addition to in-hospital mortality, data on the exact date of death were obtained by inspecting the death registers for the territory of the Republic of Serbia. ## 2.6. Statistical Analysis In statistical analysis, continuous variables were expressed as mean ± standard deviation (X¯ ± SD) or as median (interquartile range), while categorical variables were presented by number of cases (percentage). In order to assess the normality of the used data, the Kolmogorov–Smirnov test was used. The statistical computations for significance were two-tailed. The Mann–Whitney U test for continuous and the Chi-square test for categorical variables were used to assess differences between groups. All variables that showed a statistically significant correlation with disease survival and progression ($p \leq 0.05$) were analyzed by the Cox Hazard Ratio (Cox HR) model. The Cox HR model was used for univariate and multivariate regression analysis. Statistically significant differences between the analyzed variables ($p \leq 0.05$) in univariate analysis were included in multivariate regression analysis in order to assess good predictors of OS and DFS. Final outcomes were analyzed using the Log Rank test with the Kaplan–Meier survival curve. Statistical analysis was performed using IBM SPSS statistical software (SPSS for Windows, release 25.0, SPSS, Chicago, IL, USA). ## 3.1. Patient Characteristics Out of the 111 CRC patients that fulfilled the inclusion criteria, 56.8 % were male; predominantly the ASA (American Society of Anesthesiology) scores were 2 and 3. Among the 111 patients, $59.5\%$ patients had a modified Glasgow Prognostic Score of 0. More than $85\%$ of patients were older than 60 years. Clinical, biochemical and hematological laboratory characteristics of the patients are shown in Table 1. Differences in estradiol levels between male and female patients were statistically significant ($p \leq 0.001$) (median (interquartile range): 30.465 (18.35–60.015) vs. 18.35 (5–18.35), respectively), presumably due to the age of the majority of patients. Differences in estradiol levels between females younger than 60 and older than 60 years were statistically significant ($p \leq 0.001$). Clinical characteristics of patients and pathohistological characteristics of tumors in relation to overall survival and disease progression after surgical treatment are shown in Table 2. Out of the total number of operated patients, 34 ($30.9\%$) patients had rectal cancer, while two ($1.8\%$) had synchronous colon adenocarcinomas. Almost all patients, 106 ($95.5\%$), underwent elective surgical treatment. Intrahospital mortality was observed in 6 ($5.4\%$) patients. The median duration of postoperative treatment was 10 (8−12) days, and the total length of hospitalization was 13 (11−21) days. In our study, the follow-up median was 42 (24−48) months after surgery, while the median of survival without recurrence of the disease was 39 (10−45) months. During the mentioned follow-up period, 73 ($65.8\%$) patients were still alive, while the recurrence of the disease was verified in 27 ($27.6\%$) patients. ## 3.2. Overall Patient Survival (OS) Univariate and multivariate Cox regression analysis for overall survival period are shown in Table 3. Taking into account the clinical characteristics of patients, with univariate and multivariate Cox regression analysis, we found that patients with III, IV and V degrees of the Clavien–Dindo classification had 2.609 ($95\%$ CI: 1.437−4.737; $$p \leq 0.002$$) times higher risk of death during a follow-up period of 42 (24−48) months. By analysis of gender, age, comorbidity and the received therapy, there was no statistically significant association with the total survival time ($p \leq 0.05$). By Cox regression univariate and multivariate analysis of hematological/biochemical characteristics, we found a statistically significant association with overall survival in patients with CRP > 10 mg/mL and serum albumin <35 g/L. Namely, patients with mGPS 2 had a 2.188 ($95\%$ CI: 1.413−3.387; $p \leq 0.001$) times higher risk of death during the period of our postoperative follow-up. In the univariate analysis, statistically significant association with the total survival time was also shown for the values of hemoglobin ($$p \leq 0.031$$), hematocrit ($$p \leq 0.030$$), CRP ($$p \leq 0.006$$), serum albumin ($$p \leq 0.001$$), CEA ($$p \leq 0.041$$), PLR ($$p \leq 0.045$$), CAR ($$p \leq 0.003$$) and PNI ($$p \leq 0.003$$), LANR ($$p \leq 0.035$$); while in the multivariate analysis none of them retained the statistical significance. The remaining analyzed hematological/biochemical parameters: NLR, MLR, RLR, MPR did not show a statistically significant association with the overall survival time. Regarding the pathohistological characteristics of colorectal cancer, univariate analysis showed a statistically significant association with the overall survival time in patients with stage III/IV disease ($$p \leq 0.009$$), LNR ($p \leq 0.001$), PTL response ($$p \leq 0.016$$), perineural invasion ($$p \leq 0.001$$), tumor deposits ($$p \leq 0.001$$) and tumor budding ($$p \leq 0.052$$). Taking into account all of these parameters, using multivariate Cox regression analysis, we found that patients with higher lymph node ratio had 6.862 ($95\%$ CI: 1.635−28.808; $$p \leq 0.009$$) times higher risk of death in the postoperative period; while this risk was 3.089 times higher ($95\%$ CI: 1.447−6.593; $$p \leq 0.004$$) in patients with verified tumor deposits. As expected, patients with grade III, IV and V complications of the Clavien–Dindo classification lived for a shorter period than patients with grade I and II complications (20.892 ± 6.318 vs. 43.688 ± 1.592 vs. 34,577 ± 2.415, respectively) (Figure 1A). Patients with mGPS 2 had significantly shorter average survival (21.45 ± 5.304 months) compared to patients with mGPS 0 (33.939 ± 3.365 months) and mGPS 1 (41.308 ± 1.681 months) (Figure 1B). Patients with tumor deposits had a significantly shorter survival (25.405 ± 4.162 months) compared to the patients without tumor deposits (39.240 ± 1.686 months) (Figure 1C). ## 3.3. Disease-Free Patient Survival (DFS) Univariate and multivariate Cox regression analyses for DFS are shown in Table 4. The age of the patients posed a risk for disease relapse in univariate regression analysis (HR: 0.961; $95\%$ CI: 0.926−0.998; $$p \leq 0.041$$). Younger patients had a higher risk of disease recurrence, but this statistical significance was lost in multivariate regression analysis ($p \leq 0.05$). None of the previously mentioned analyzed hematological/biochemical parameters and indexes had a significant statistical association with DFS. Using univariate Cox regression analysis, a statistically significant association was found with survival time without relapse in stage III/IV disease ($$p \leq 0.002$$), LNR ($$p \leq 0.076$$), PTL response ($$p \leq 0.014$$), lymphovascular invasion ($$p \leq 0.069$$), perineural invasion ($$p \leq 0.035$$) and tumor deposits ($$p \leq 0.001$$). Multivariate regression analysis showed that patients with stage III/IV and tumor deposits had 1.888 ($95\%$ CI: 1024−3481; $$p \leq 0.042$$) and 3.049 ($95\%$ CI: 1.206−7.706; $$p \leq 0.018$$) times higher risk of disease recurrence, respectively. The emphasized peritumoral lymphocyte response reduced the risk of recurrence by $61\%$ (HR: 0.391; $95\%$ CI: 0.196−0.780; $$p \leq 0.005$$). Patients with clinical stage III/IV relapsed earlier than patients in stages I and II (31.506 ± 2.848 vs. 44.475 ± 2.010 vs. 40.828 ± 2.922 months, respectively) (Figure 2A). The same trend was observed in patients with tumor deposits (Figure 2B). Patients with high peritumoral lymphocyte response had longer disease-free survival time compared to patients without and to patients with mild to moderate peritumoral lymphocyte response (46.118 ± 1.826 vs. 33.429 ± 3.902 vs. 37.230 ± 2.244, respectively) (Figure 2C). ## 4. Discussion In this study, the association of hematological/biochemical parameters, their indexes and pathohistological characteristics of the tumor with the CRC patient outcome were examined in a pilot single center cohort in Serbia. Colorectal cancer is a growing health burden in Serbia and worldwide. Although a Serbian national screening program for colorectal cancer in the population aged 50 to 74 was announced in 2013, to date the only screening method is colonoscopy and in practice no structured screening exists. Genetic testing for Lynch syndrome in Serbia was established in 2018. This has an impact on the characteristics of patient cohorts treated at the secondary and tertiary points of care. Nevertheless, examination of these groups provides real world insight into the clinical utility of prognostic parameters validated in settings with more advanced screening procedures. The most important findings of our research were that postoperative complications of grade III, IV and V according to Clavien–Dindo classification, mGPS 2, higher LNR and tumor deposits were statistically significantly associated with worse OS. Moreover, III/IV TNM disease stage and tumor deposits were statistically significantly associated with worse DFS, while the presence of peritumoral lymphocyte response was statistically significantly associated with better DFS. Our examined group consisted mainly of patients older than 60 years ($85.57\%$) and hence we did not find significant influence of age on the overall survival. Patient sex was also not significantly associated with overall survival, unlike in larger cohort studies that show that women have substantial survival advantage [16]. We speculate that, given the age of the patients in our cohort, the protective effect of estrogen in women was lost, as was previously documented [17,18]. Indeed, the estradiol levels in female patients older than 60 years ($$n = 43$$) were significantly lower than in patients younger than 60 years ($$n = 5$$). Estrogen confers survival advantage in females through estrogen-regulated genes and cell signaling [19], and can control tumor growth by regulating the tumor immune microenvironment [20]. The level of estrogen receptor expression in most females 50 years of age and older is less than $10\%$ [21] and out of the total number of women in our study, $95.83\%$ were postmenopausal women over the age of 50. With regards to age, by univariate regression analysis we noticed that younger patients had a higher risk of disease recurrence during three years of follow-up, but this significance was not confirmed in the multivariate regression analysis. Our results showed that postoperative complications based on the Clavien–Dindo classification were an independent risk factor in relation to total survival time, but not in disease-free survival time (DFS). Clavien–Dindo classification is a simple and feasible grading system of postoperative complications. Grades I and II represent surgical complications that can be solved by conservative treatment. Grade III complications require surgical, endoscopic or radiological intervention with or without anesthesia. Grade IV complications mean life-threatening complication (including CNS complications) requiring intermediate care/intensive care unit management. Grade V implies a complication that ends in death [22]. In our cohort, patients with grades III, IV and V of the Clavien–Dindo classification had a HR for total survival of 2.609 ($95\%$ CI: 1.437−4.737; $$p \leq 0.002$$), slightly higher than in the previously reported large observational studies [23,24]. Postoperative complications are partly due to the involvement of the immune system. Systemic inflammation is an indicator of poor prognosis in 21−$41\%$ of patients with colorectal cancer [25]. Many markers of systemic inflammation are based on the number, ratios or scores of circulating leukocytes or acute phase proteins, or serum albumins, such as NLR, LANR, MLR, CAR, mGPS [6,7,8,9]. Several studies have shown that NLR and LMR were good predictors of prognosis of overall survival, cancer-specific survival, and disease-free survival when considering cohorts of patients with rectal cancer or colon and rectal cancer together [26,27]. In our patient cohort, consisting of patients with both colon and rectal cancer, we did not find such associations for NRL and MRL. We also did not confirm the findings of Liang et al. [ 6], who were the first to report the LANR—the relationship between lymphocytes, serum albumin and neutrophils—as a good indicator of overall survival and relapse-free survival in resectable colorectal cancer. In our cohort, univariate analysis showed that LANR was associated with longer overall survival only (HR: 0.946; 95 CI: 0.898−0.996; $$p \leq 0.035$$), while multivariate analysis showed no statistical significance in terms of overall and disease-free survival. The platelet to lymphocyte ratio (PLR) in our cohort had a significant statistical association with overall survival in univariate regression analysis only (HR: 1.002; $95\%$ CI: 1.00−1.005; $$p \leq 0.045$$), while this statistical significance was lost in multivariate regression analysis. Ozawa et al. have shown that high values of PLR alone are an independent prognostic factor for disease-free survival and cancer-specific survival in stage II disease for CRC cancer [28]. We did not find that this variable was associated with survival when all the stages of CRC were analyzed together. Systemic inflammation is a very important factor in cachexia related to malignancy. Cachexia not only reduces the quality of life of patients and the response to treatment, but is also an indirect cause of death in about $20\%$ of patients who eventually die from cancer [29]. Nutritional status plays a significant role in the overall survival in colorectal cancer patients [9,30,31]. In our patient cohort, significant survival indicators related to the nutritional status of patients were CAR, PNI, and mGPS. Modified Glasgow prognostic score (mGPS) is a prognostic score based on C-reactive protein (CRP) and serum albumin concentrations. The mGPS score ranges from 0 to 2. Patients with both an elevated CRP (>10 mg/L) and decreased albumin (<35 g/L) are assigned a score of 2, whereas those with either an elevated CRP or decreased albumin alone are assigned a score of 1. Patients with a normal CRP concentration and albumin level are assigned a score of 0 [9]. In our study, $12.6\%$ of patients had mGPS 2. Univariate and multivariate regression analysis showed that mGPS 2 increased the risk of death more than 2-fold during 42 months of the postoperative follow-up. This was in line with the study by Proctor et al. [ 9] where mGPS was found to be an independent prognostic indicator in multi-cancer analysis for overall survival and tumor-specific survival, including colorectal cancer. A recent study with a significant focus on patient nutritional status and chronic inflammation, conducted by Son et al. [ 30] implied that mGPS and CAR taken together had better prognostic value than individually considered mGPS and CAR in patients with CRC. On the other hand, in the cohort with mismatch repair-deficient colorectal cancer there was no statistically significant difference between the High CAR Group vs. Low CAR Group, either in terms of overall survival or in terms of survival time without disease relapse [32]. The results of the study by Nagashima et al. indicated that the mGPS was a good predictor not only of 60-day mortality, but also of the overall survival of patients with late-stage cancer and malignant bowel obstruction [33]. In our patient cohort, univariate regression analysis indicated that the ratio of CRP and serum albumin (CAR) was associated with longer overall survival (HR: 1335; $95\%$ CI: 1102−1617; $$p \leq 0.003$$), without statistical significance in multivariate regression analysis. The same trend was observed in PNI (HR: 0.924; $95\%$ CI: 0.877−0.974; $$p \leq 0.003$$), with neither CAR nor PNI values being a risk factor for disease recurrence. In contrast to our results, the results of the study by Tamai et al. propose CAR as an independent indicator of OS and imply that CAR is a useful and promising prognostic marker in elderly patients undergoing curative surgery for CRC [34]. CRP, an inflammatory marker, is an acute-phase reactant synthesized by liver cells [35]. An elevated CRP level reflects the inflammatory response caused by tumor necrosis and is significantly higher in metastatic colorectal cancer liver disease [36]. CRP is mediated by many pro-inflammatory cytokines [37], which suppress the synthesis of albumin under inflammatory conditions [38]. In our patient cohort, the number of patients with metastatic liver disease of CRC ($$n = 9$$), which leads to the highest oscillations in the values of CRP and serum albumin, was smaller than in other studies [36], so the results related to CAR were slightly different. In terms of pathohistological characteristics, univariate regression analysis of the TNM classification, lymph node ratio, peritumor lymphocyte response, perineural invasion, presence of tumor deposits and tumor budding showed statistical significance; while multivariate regression analysis showed that only higher lymph node ratio and presence of tumor deposits were strong indicators of a poor prognosis of the overall patient survival. Univariate analysis also showed that TNM stage, lymph node ratio, peritumor lymphocyte response, lymphovascular invasion, perineural invasion and tumor deposits were associated with the disease-free survival in our cohort of CRC patients. In multivariate regression analysis the TNM stage, presence of peritumoral lymphocyte response and absence of tumor deposits were the only parameters associated with longer disease-free survival. In contrast to the systemic inflammatory response, which is associated with a poor prognosis [6,7,8,9], verified intense immune cell infiltration in and around the tumor is often associated with better survival in colorectal cancer, regardless of disease stage or other prognostic parameters [39]. This is attributed to the ability of immune cells to recognize transformed malignant cells and limit tumor growth (immunosuppression hypothesis) [39]. Peritumoral lymphocyte response reduced the risk of disease relapse by $61\%$ in our patient cohort, underlying the importance of the immune surveillance in cancer management [39]. The prognostic value of tumor deposits in our CRC patient cohort was in line with previous studies [40,41,42], confirming their importance as indicators of poor prognosis. While the disadvantages of our study were its single-center retrospective design, relatively small sample size and combination of colon and rectal cancer patients, as well as inclusion of patients with neoadjuvant chemoradiotherapy that would potentially alter their preoperative immune response, our results present a real world perspective on the utility of hematological/biochemical parameters and pathohistological characteristics of a tumor for prediction of survival of CRC patients in a middle-income Eastern European country. Insights that not all previously reported parameters hold prognostic value when examined in an age-skewed cohort lays the groundwork for examination of these specific parameters in a larger prospective study. ## 5. Conclusions To the best of our knowledge, this is the first study of this type on the Serbian colorectal cancer patient population. It provides an important insight into the single center real-world scenario utility of the hematological/biochemical parameters, their indexes and pathohistological tumor characteristics as indicators of the prognosis in patients that were undergoing CRC surgery. These parameters—that are a part of any standard hospital pre-operative procedure and do not present any additional cost burden for the health system—may provide additional information on the patient outcome. We found that the Clavien–Dindo classification of post-operative complications, mGPS, lymph node ratio, tumor deposits and peritumoral lymphocyte response were factors worth taking into consideration when predicting survival of operated patients. In combination with the reported genetic studies performed on the same population [43], our results may also be useful for future meta-analyses of CRC patient populations. ## References 1. Sung H., Ferlay J., Siegel R.L., Laversanne M., Soerjomataram I., Jemal A., Bray F.. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA A Cancer J. Clin. (2021.0) 71 209-249. DOI: 10.3322/caac.21660 2. Hashiguchi Y., Muro K., Saito Y., Ito Y., Ajioka Y., Hamaguchi T., Hasegawa K., Hotta K., Ishida H., Ishiguro M.. Japanese Society for Cancer of the Colon and Rectum (JSCCR) guidelines 2019 for the treatment of colorectal cancer. Int. J. Clin. Oncol. (2020.0) 25 1-42. DOI: 10.1007/s10147-019-01485-z 3. O’Connell J.B., Maggard M.A., Ko C.Y.. Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging. J. Natl. Cancer Inst. (2004.0) 96 1420-1425. DOI: 10.1093/jnci/djh275 4. Naxerova K., Reiter J.G., Brachtel E., Lennerz J.K., van de Wetering M., Rowan A., Cai T., Clevers H., Swanton C., Nowak M.A.. Origins of lymphatic and distant metastases in human colorectal cancer. Science (2017.0) 357 55-60. DOI: 10.1126/science.aai8515 5. Percario R., Panaccio P., di Mola F.F., Grottola T., Di Sebastiano P.. The Complex Network between Inflammation and Colorectal Cancer: A Systematic Review of the Literature. Cancers (2021.0) 13. DOI: 10.3390/cancers13246237 6. Liang X., Yao S., Lu P., Ma Y., Xu H., Yin Z., Hu J., Liu Y., Wei S.. The Prognostic Value of New Index (LANR) Composed of Pre-operative Lymphocytes, Albumin, and Neutrophils in Patients with Resectable Colorectal Cancer. Front. Oncol. (2021.0) 11 610264. DOI: 10.3389/fonc.2021.610264 7. Hayama T., Ozawa T., Asako K., Kondo R., Ono K., Okada Y., Tsukamoto M., Fukushima Y., Shimada R., Nozawa K.. Impact of Colon Cancer Location on the Prognostic Significance of Nutritional Indexes and Inflammatory Markers. In Vivo (2021.0) 35 1261-1269. DOI: 10.21873/invivo.12377 8. Ishizuka M., Nagata H., Takagi K., Iwasaki Y., Shibuya N., Kubota K.. Clinical Significance of the C-Reactive Protein to Albumin Ratio for Survival After Surgery for Colorectal Cancer. Ann. Surg. Oncol. (2016.0) 23 900-907. DOI: 10.1245/s10434-015-4948-7 9. Proctor M.J., Morrison D.S., Talwar D., Balmer S.M., O’Reilly D.S., Foulis A.K., Horgan P.G., McMillan D.C.. An inflammation-based prognostic score (mGPS) predicts cancer survival independent of tumour site: A Glasgow Inflammation Outcome Study. Br. J. Cancer (2011.0) 104 726-734. DOI: 10.1038/sj.bjc.6606087 10. Man W., Lin H., Liu Z., Jin L., Wang J., Zhang J., Bai Z., Yao H., Zhang Z., Deng W.. Usefulness of Inflammation-Based Prognostic Scores for Predicting the Risk of Complications After Radical Resection of Colorectal Carcinoma. Cancer Manag. Res. (2020.0) 12 1029-1038. DOI: 10.2147/CMAR.S234448 11. Sato R., Oikawa M., Kakita T., Okada T., Abe T., Yazawa T., Tsuchiya H., Akazawa N., Sato M., Ohira T.. Preoperative change of modified Glasgow prognostic score after stenting predicts the long-term outcomes of obstructive colorectal cancer. Surg. Today (2020.0) 50 232-239. DOI: 10.1007/s00595-019-01862-1 12. Amin M.B., Greene F.L., Edge S.B., Compton C.C., Gershenwald J.E., Brookland R.K., Meyer L., Gress D.M., Byrd D.R., Winchester D.P.. The Eighth Edition AJCC Cancer Staging Manual: Continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging. CA A Cancer J. Clin. (2017.0) 67 93-99. DOI: 10.3322/caac.21388 13. Roxburgh C.S., McMillan D.C.. Role of systemic inflammatory response in predicting survival in patients with primary operable cancer. Future Oncol. (2010.0) 6 149-163. DOI: 10.2217/fon.09.136 14. Ministry of Health of the Republic of Serbia 15. Pfister D.G., Benson A.B., Somerfield M.R.. Clinical practice. Surveillance strategies after curative treatment of colorectal cancer. N. Engl. J. Med. (2004.0) 350 2375-2382. DOI: 10.1056/NEJMcp010529 16. Majek O., Gondos A., Jansen L., Emrich K., Holleczek B., Katalinic A., Nennecke A., Eberle A., Brenner H.. Sex differences in colorectal cancer survival: Population-based analysis of 164,996 colorectal cancer patients in Germany. PLoS ONE (2013.0) 8. DOI: 10.1371/journal.pone.0068077 17. Clendenen T.V., Koenig K.L., Shore R.E., Levitz M., Arslan A.A., Zeleniuch-Jacquotte A.. Postmenopausal levels of endogenous sex hormones and risk of colorectal cancer. Cancer epidemiology, biomarkers & prevention: A publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. Cancer Epidemiol. Biomark. Prev. (2009.0) 18 275-281 18. Barzi A., Lenz A.M., Labonte M.J., Lenz H.J.. Molecular pathways: Estrogen pathway in colorectal cancer. Clinical cancer research. Off. J. Am. Assoc. Cancer Res. (2013.0) 19 5842-5848. DOI: 10.1158/1078-0432.CCR-13-0325 19. Abancens M., Bustos V., Harvey H., McBryan J., Harvey B.J.. Sexual Dimorphism in Colon Cancer. Front. Oncol. (2020.0) 10 607909. DOI: 10.3389/fonc.2020.607909 20. Jiang L., Fei H., Yang A., Zhu J., Sun J., Liu X., Xu W., Yang J., Zhang S.. Estrogen inhibits the growth of colon cancer in mice through reversing extracellular vesicle-mediated immunosuppressive tumor microenvironment. Cancer Lett. (2021.0) 520 332-343. DOI: 10.1016/j.canlet.2021.08.011 21. Salehi Far S., Soltani M., Zardast M., Zhu J., Sun J., Liu X., Xu W., Yang J., Zhang S.. Investigating the Factors Associated with the Level of Expression of Estrogen and Progesterone Receptors in Patients Suffering from Colorectal Cancer. J. Cancer Epidemiol. (2021.0) 2021 4478155. DOI: 10.1155/2021/4478155 22. Dindo D., Demartines N., Clavien P.A.. Classification of surgical complications: A new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann. Surg. (2004.0) 240 205-213. DOI: 10.1097/01.sla.0000133083.54934.ae 23. Duraes L.C., Stocchi L., Steele S.R., Kalady M.F., Church J.M., Gorgun E., Liska D., Kessler H., Lavryk O.A., Delaney C.P.. The Relationship between Clavien-Dindo Morbidity Classification and Oncologic Outcomes After Colorectal Cancer Resection. Ann. Surg. Oncol. (2018.0) 25 188-196. DOI: 10.1245/s10434-017-6142-6 24. Aoyama T., Oba K., Honda M., Sadahiro S., Hamada C., Mayanagi S., Kanda M., Maeda H., Kashiwabara K., Sakamoto J.. Impact of postoperative complications on the colorectal cancer survival and recurrence: Analyses of pooled individual patients’ data from three large phase III randomized trials. Cancer Med. (2017.0) 6 1573-1580. DOI: 10.1002/cam4.1126 25. Tuomisto A.E., Mäkinen M.J., Väyrynen J.P.. Systemic inflammation in colorectal cancer: Underlying factors, effects, and prognostic significance. World J. Gastroenterol. (2019.0) 25 4383-4404. DOI: 10.3748/wjg.v25.i31.4383 26. Xia L.J., Li W., Zhai J.C., Yan C.W., Chen J.B., Yang H.. Significance of neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, lymphocyte-to-monocyte ratio and prognostic nutritional index for predicting clinical outcomes in T1-2 rectal cancer. BMC Cancer (2020.0) 20. DOI: 10.1186/s12885-020-6698-6 27. Mizuno H., Yuasa N., Takeuchi E., Miyake H., Nagai H., Yoshioka Y., Miyata K.. Blood cell markers that can predict the long-term outcomes of patients with colorectal cancer. PLoS ONE (2019.0) 14. DOI: 10.1371/journal.pone.0220579 28. Ozawa T., Ishihara S., Nishikawa T., Tanaka T., Tanaka J., Kiyomatsu T., Hata K., Kawai K., Nozawa H., Kazama S.. The preoperative platelet to lymphocyte ratio is a prognostic marker in patients with stage II colorectal cancer. Int. J. Color. Dis. (2015.0) 30 1165-1171. DOI: 10.1007/s00384-015-2276-9 29. Argilés J.M., Busquets S., Stemmler B., López-Soriano F.J.. Cancer cachexia: Understanding the molecular basis. Nat. Rev. Cancer (2014.0) 14 754-762. DOI: 10.1038/nrc3829 30. Son W., Shin S.J., Park S.H., Lee S.K., Park E.J., Baik S.H., Lee K.Y., Kang J.. Clinical Impact of Combined Modified Glasgow Prognostic Score and C-Reactive Protein/Albumin Ratio in Patients with Colorectal Cancer. Diagnostics (2020.0) 10. DOI: 10.3390/diagnostics10110859 31. Sun G., Li Y., Peng Y., Lu D., Zhang F., Cui X., Zhang Q., Li Z.. Impact of the preoperative prognostic nutritional index on postoperative and survival outcomes in colorectal cancer patients who underwent primary tumor resection: A systematic review and meta-analysis. Int. J. Color. Dis. (2019.0) 34 681-689. DOI: 10.1007/s00384-019-03241-1 32. Climent M., Ryan É.J., Stakelum Á., Khaw Y.L., Creavin B., Lloyd A., Alhassan D., Mohan H.M., Kennelly R., Sheahan K.. Systemic inflammatory response predicts oncological outcomes in patients undergoing elective surgery for mismatch repair-deficient colorectal cancer. Int. J. Color. Dis. (2019.0) 34 1069-1078. DOI: 10.1007/s00384-019-03274-6 33. Nagashima Y., Funahashi K., Kagami S., Ushigome M., Kaneko T., Miura Y., Yoshida K., Koda T., Kurihara A.. Which preoperative immunonutritional index best predicts postoperative mortality after palliative surgery for malignant bowel obstruction in patients with late-stage cancer? A single-center study in Japan comparing the modified Glasgow prognostic score (mGPS), the prognostic nutritional index (PNI), and the controlling nutritional status (CONUT). Surg. Today (2023.0) 53 22-30. PMID: 35781553 34. Tamai K., Okamura S., Makino S., Yamamura N., Fukuchi N., Ebisui C., Inoue A., Yano M.. C-reactive protein/albumin ratio predicts survival after curative surgery in elderly patients with colorectal cancer. Updates Surg. (2022.0) 74 153-162. DOI: 10.1007/s13304-021-01011-9 35. Morris-Stiff G., Gomez D., Prasad K.R.. C-reactive protein in liver cancer surgery. Eur. J. Surg. Oncol. J. Eur. Soc. Surg. Oncol. Br. Assoc. Surg. Oncol. (2008.0) 34 727-729. DOI: 10.1016/j.ejso.2008.01.016 36. Zhou J., Wei W., Hou H., Ning S., Li J., Huang B., Liu K., Zhang L.. Prognostic Value of C-Reactive Protein, Glasgow Prognostic Score, and C-Reactive Protein-to-Albumin Ratio in Colorectal Cancer. Front. Cell Dev. Biol. (2021.0) 9 637650. DOI: 10.3389/fcell.2021.637650 37. Nakazaki H.. Preoperative and postoperative cytokines in patients with cancer. Cancer (1992.0) 70 709-713. DOI: 10.1002/1097-0142(19920801)70:3<709::AID-CNCR2820700328>3.0.CO;2-O 38. McMillan D.C., Watson W.S., O’Gorman P., Preston T., Scott H.R., McArdle C.S.. Albumin concentrations are primarily determined by the body cell mass and the systemic inflammatory response in cancer patients with weight loss. Nutr. Cancer (2001.0) 39 210-213. DOI: 10.1207/S15327914nc392_8 39. Finn O.J.. A Believer’s Overview of Cancer Immunosurveillance and Immunotherapy. J. Immunol. (2018.0) 200 385-391. DOI: 10.4049/jimmunol.1701302 40. Liu F., Zhao J., Li C., Wu Y., Song W., Guo T., Chen S., Cai S., Huang D., Xu Y.. The unique prognostic characteristics of tumor deposits in colorectal cancer patients. Ann. Transl. Med. (2019.0) 7 769. DOI: 10.21037/atm.2019.11.69 41. Lord A.C., D’Souza N., Pucher P.H., Moran B.J., Abulafi A.M., Wotherspoon A., Rasheed S., Brown G.. Significance of extranodal tumour deposits in colorectal cancer: A systematic review and meta-analysis. Eur. J. Cancer (2017.0) 82 92-102. DOI: 10.1016/j.ejca.2017.05.027 42. Nagtegaal I.D., Knijn N., Hugen N., Marshall H.C., Sugihara K., Tot T., Ueno H., Quirke P.. Tumor Deposits in Colorectal Cancer: Improving the Value of Modern Staging-A Systematic Review and Meta-Analysis. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. (2017.0) 35 1119-1127. DOI: 10.1200/JCO.2016.68.9091 43. Jakovljevic K., Malisic E., Cavic M., Krivokuca A., Dobricic J., Jankovic R.. KRAS and BRAF mutations in Serbian patients with colorectal cancer. J. BUON Off. J. Balk. Union Oncol. (2012.0) 17 575-580
--- title: Integrative Analyses Reveal the Anticancer Mechanisms and Sensitivity Markers of the Next-Generation Hypomethylating Agent NTX-301 authors: - Byungho Lim - Dabin Yoo - Younghwa Chun - Areum Go - Ji Yeon Kim - Ha Young Lee - Rebecca J. Boohaker - Kyung-Jin Cho - Sunjoo Ahn - Jin Soo Lee - DooYoung Jung - Gildon Choi journal: Cancers year: 2023 pmcid: PMC10046470 doi: 10.3390/cancers15061737 license: CC BY 4.0 --- # Integrative Analyses Reveal the Anticancer Mechanisms and Sensitivity Markers of the Next-Generation Hypomethylating Agent NTX-301 ## Abstract ### Simple Summary We thoroughly investigated the experimental and preclinical efficacy of the novel hypomethylating agent (HMA) NTX-301 through comparative analyses with conventional HMAs. By performing multiomics data analyses, we demonstrated the mechanisms of action underlying the anticancer activity of NTX-301 and identified molecular markers that are associated with sensitivity to NTX-301. ### Abstract Epigenetic dysregulation characterized by aberrant DNA hypermethylation is a hallmark of cancer, and it can be targeted by hypomethylating agents (HMAs). Recently, we described the superior therapeutic efficacy of a novel HMA, namely, NTX-301, when used as a monotherapy and in combination with venetoclax in the treatment of acute myeloid leukemia. Following a previous study, we further explored the therapeutic properties of NTX-301 based on experimental investigations and integrative data analyses. Comprehensive sensitivity profiling revealed that NTX-301 primarily exerted anticancer effects against blood cancers and exhibited improved potency against a wide range of solid cancers. Subsequent assays showed that the superior efficacy of NTX-301 depended on its strong effects on cell cycle arrest, apoptosis, and differentiation. Due to its superior efficacy, low doses of NTX-301 achieved sufficiently substantial tumor regression in vivo. Multiomics analyses revealed the mechanisms of action (MoAs) of NTX-301 and linked these MoAs to markers of sensitivity to NTX-301 and to the demethylation activity of NTX-301 with high concordance. In conclusion, our findings provide a rationale for currently ongoing clinical trials of NTX-301 and will help guide the development of novel therapeutic options for cancer patients. ## 1. Introduction Epigenetic homeostasis is tightly controlled by various epigenetic regulators to maintain normal physiological homeostasis [1]. Increased somatic mutations and transcriptional deregulation in epigenetic regulators impair epigenetic homeostasis, driving tumor development and progression [2]. Epigenetic dysregulation characterized by aberrations in DNA methylation and histone modification is a major hallmark of cancer [3]. Therefore, targeting epigenetic regulators is a well-established therapeutic approach for the treatment of multiple cancers. DNA methyltransferase I (DNMT1) is an epigenetic ‘writer’ that is responsible for the maintenance of cytosine methylation. DNMT1 transfers a methyl group to the 5′ carbon of cytosine rings in newly synthesized DNA [4], thus ensuring the high-fidelity inheritance of DNA methylation patterns during replication [5]. Emerging evidence suggests that the deregulation of DNMT1 and the resulting aberrant DNA hypermethylation are involved in malignant transformation. Indeed, the reversal of aberrant hypermethylation by inhibiting DNMT1 promoted anticancer activity, especially in hematologic malignancies [6]; this finding partially explains the relevance of DNMT1 as a therapeutic target. Hypomethylating agents (HMAs) are the most widely used DNMT1 inhibitors. Mechanistically, HMAs, which are cytidine analogs, are incorporated into newly replicated DNA, and DNMT1 becomes sequestered by its formation of covalent complexes with HMAs, eventually resulting in the proteasomal degradation of DNMT1 [7]. The first-generation HMAs decitabine (DAC) and azacitidine (AZA) have been approved for clinical use in the treatment of hematologic malignancies [8]. However, their low response rate, poor bioavailability, and dose-limiting toxicity highlight the need for further treatment improvements [9,10]. The representative effort to overcome the limitations of HMAs is the development of reversible DNMT1-selective inhibitors. GSK3685032, a first-in-class DNMT1-selective noncovalent inhibitor, achieved improved in vivo tolerability, higher antitumor efficacy, and greater DNA demethylation, providing clinical benefits over conventional HMAs [11]. As another effort, we recently described the superior therapeutic potential of NTX-301 (5-aza-4′-thio-2′-deoxycytidine) as a next-generation HMA using six mouse models [12]. This 4′-thio-modified nucleoside analog showed preclinical efficacy, tolerability, and survival outcomes that were superior to those of conventional HMAs when used as a monotherapy or in combination with venetoclax in the treatment of acute myeloid leukemia (AML) [12]. Consistent with our findings, Thottassery et al. showed pharmacological advantages of NTX-301, namely, that NTX-301 exhibited improved chemical stability, DNA incorporation rates, and preclinical activity compared with conventional HMAs [13]. Notably, NTX-301 is better tolerated than DAC, with an at least 10-fold higher selectivity index (ratio of the maximum tolerated dose to the minimal dose required to deplete DNMT1) [13]. Although these findings provide a rationale for the currently ongoing clinical trials that are investigating NTX-301 (NCT04167917, NCT03366116, and NCT04851834), the anticancer activity and mechanisms of action (MoAs) of NTX-301, as well as biomarkers that can be used to predict its efficacy, are not yet elucidated. Herein, we aimed to thoroughly investigate the experimental/preclinical efficacy of NTX-301 through comparative analyses with the conventional agents DAC and AZA. To understand the widespread therapeutic potential of NTX-301, we determined the sensitivity profiles of 199 cancer cell lines (CCLs) after NTX-301 treatment. By integrating these sensitivity profiles with multiomics data, we investigated the MoAs underlying the anticancer activity of NTX-301 and identified molecular determinants that are associated with sensitivity to NTX-301. ## 2.1. Cell Lines and Reagents Five human leukemia cell lines (the MV4-11, HL-60, MOLM-13, KG-1, and THP-1 cell lines) and HEK293T cells were cultured in RPMI 1640 and DMEM (Thermo, Waltham, MA, USA), respectively. CRISPR/Cas9 technology was used to knock out TP53 in MV4-11 cells (Cyagen, Santa Clara, CA, USA). NTX-301 was provided by MercachemSyncom, and DAC, AZA, and mTOR inhibitors (Torin1 and AZD-8055) were purchased from Selleckchem (Houston, TX, USA). ## 2.2. Cell-Based Phenotypic Assays The sensitivity profiling of 199 CCLs was performed using the OncoPanel™ Multiplex Cytotoxicity Assay (Eurofins, Luxembourg). Briefly, cells grown in RPMI 1640 were seeded into 384-well plates, and NTX-301 was added the following day. NTX-301 was serially diluted and assayed at 10 concentrations with a maximum assay concentration of $0.1\%$ DMSO. After a 3-day incubation, the cells were fixed and stained with nuclear dye to measure cell proliferation by fluorescence intensity. Automated fluorescence microscopy was carried out using a Molecular Devices ImageXpress Micro XL high-content imager, and images were collected with a 4× objective. Sixteen-bit TIFF images were acquired and analyzed with MetaXpress 5.1.0.41 software. To assess the anticancer effect of NTX-301 at the cellular level, we performed the following assays: [1] A CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA) was used to determine IC50 values. For 2, 4, and 6 days, NTX-301 or DAC was added to 2 × 103 cells plated into 96-well plates in triplicate. Treatment doses ranged from 1.5 nM to 10 µM. [2] For the cell cycle assay, NTX-301 or DAC (500 nM) was added to 5 × 105 MV4-11 cells plated into six-well plates for 2 and 3 days, and the cells were stained with phospho-histone H3 (CST, Danvers, MA, USA) and propidium iodide (Sigma, St. Louis, MO, USA). [ 3] To assay apoptosis, NTX-301 or DAC (500 nM for 2 and 3 days) was added to 5 × 105 MV4-11 cells plated into six-well plates, and two different methods were used (Annexin V Apoptosis Detection Kit (Thermo) and the Cleaved Caspase-3 Staining Kit (Abcam, Cambridge, UK)). [ 4] For AML cell differentiation, NTX-301 or DAC (60 and 200 nM) was added to 5 × 105 MV4-11 cells plated into six-well plates for 6 days, and the cells were analyzed with APC-conjugated anti-human CD14 and Alexa Fluor® 488-conjugated anti-human CD11b antibodies (BD Pharmingen™, Franklin Lakes, NJ, USA). The differentiated cell morphology was inspected by Giemsa staining (Thermo). All assays were performed according to the manufacturers’ protocols, and the results were captured using BD FACSCalibur™. ## 2.3. Mouse Study The mouse study was conducted in accordance with the recommendations of the Guide for Care and Use of Laboratory Animals. All experiments were performed by Charles River Discovery Services (CR Discovery Services, Worcester, MA, USA), which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. A systemic NOD/SCID model bearing luciferase-labeled MV4-11 tumors ($$n = 8$$ per group, 4 groups) was established to assess the efficacy of low doses of NTX-301 (0.15, 0.30, and 0.45 mg/kg (p.o.)) in combination with tetrahydrouridine (42.0 mg/kg). Briefly, the mice were administered cyclophosphamide ((100 mg/kg (i.p.) once a day for two days) to ablate their bone marrow three days prior to tumor implantation. After the injection of 5.0 × 106 luciferase-labeled MV4-11 cells, treatment was initiated on day 20 by preparing NTX-301 in N-methyl-2-pyrrolidone (NMP) and then PEG400, followed by the addition of saline with vortexing and sonication. Tumor growth was measured by in vivo bioluminescence imaging on day 42. This study is an unpublished part of a previously reported mouse study [12]. ## 2.4. Western Blotting The cells were lysed with NP-40 buffer (Thermo), and 30 µg of protein was loaded into Mini-PROTEAN precast gels (Bio-Rad, Hercules, CA, USA). Protein transfer was conducted using a Trans-Blot Turbo Transfer System (Bio-Rad). Immunoreactions were detected with SuperSignal™ West Pico or Femto substrates (Thermo) using an iBright System (Thermo). All antibodies were purchased from CST. ## 2.5.1. DNA Methylation Analysis For 2 days, 5 × 105 cells plated into six-well plates in triplicate were treated with NTX-301 or DAC (30 nM for HL-60 and 60 nM for MV4-11). After purifying genomic DNA with a QIAamp DNA Mini Kit (Qiagen, Venlo, The Netherlands), methylome analysis was performed by Macrogen using Illumina Infinium Methylation EPIC BeadChip kits and Illumina GenomeStudio v2011.1 (San Diego, CA, USA). Methylation data points were represented as fluorescent signals of the methylated (M) and unmethylated (U) alleles by subtracting each data point from background intensity calculated from a set of negative controls [14]. The β-value that reflects the methylation level of each CpG site was calculated with the following formula: (max(M, 0))/(\|U\| + \|M\| + 100) [14]. To annotate histone modification features at the genome positions of each CpG site, we used H3K27ac, H3K36me3, H3K9me3, and H3K27me3 ChIP-seq data from ENCODE (https://www.encodeproject.org/ (accessed on 24 September 2020)). To annotate replication timing at the genomic locations of each CpG site, we used Repli-seq data from ReplicationDomain (https://www2.replicationdomain.com/ (accessed on 4 November 2020)). ‘ Replication timing < 0’, ‘0 ≤ Replication timing < 1’, and ‘1 ≤ Replication timing’ were classified as ‘late’, ‘mid-early’, and ‘early’ replication regions, respectively. The normalized demethylation activity of NTX-301 (versus DAC) was calculated as follows: (the number of CpGs within a given genomic feature that were demethylated by NTX-301)/(the number of CpGs within a given genomic feature that were demethylated by DAC). ## 2.5.2. Transcriptome Analysis A total of 5 × 105 cells were plated into six-well plates the day before drug treatment, and NTX-301 or DAC (60 nM for MV4-11, 30 nM for HL-60, and 15 nM for MOLM-13) was administered in triplicate for 2 days (2 and 4 days for MV4-11). After extracting total RNA using an RNeasy mini kit (Qiagen), RNA purity and integrity were assessed by an ND-1000 spectrophotometer. Transcriptome data were produced using Illumina RNA sequencing (for HL-60 and MOLM-13) and Affymetrix GeneChip® Human Gene 2.0 ST Array (for MV4-11). Sequencing libraries were prepared using a TruSeq Stranded Total RNA LT Sample Prep Kit (Gold), and paired-end sequencing was conducted by Macrogen using the Illumina platform. After completing quality control using FastQC, sequencing reads were preprocessed by trimming with Trimmomatic 0.38 [15]. Sequencing reads were then mapped with HISAT2 [16], and transcript assembly and quantification were conducted using StringTie [17]. We used Enrichr [18], GSEA [19,20], IPA (Qiagen), and Gene Ontology analyses to characterize the biological pathways associated with gene sets. ## 2.5.3. Integrative Data Analysis To perform integrative multiomics analyses across cancer cell lines, we downloaded Cancer Cell Line Encyclopedia (CCLE) datasets including somatic mutations, gene expression, RPPA, and DNA methylation from cBioPortal (http://www.cbioportal.org/study/clinicalData?id=ccle_broad_2019 (accessed on 6 August 2020)). To identify biomarkers associated with sensitivity to NTX-301, we classified 199 CCLs (profiled for sensitivity to NTX-301) into responders and nonresponders based on their AUC values (bottom third vs. top third). Then, we statistically examined differentially expressed genes, differential mutational events, and differentially methylated regions between the responders and nonresponders. We also integrated IC50 values with RPPA data from CCLE to identify IC50-correlated proteins in 29 blood CCLs. ## 2.6. Statistical Analysis GraphPad Prism v9.0 (GraphPad Software) was used for statistical analyses and graphical presentation. The two-tailed unpaired t test was used to compare the mean differences between the two groups. To analyze overall survival, we generated Kaplan–Meier curves and performed the log-rank (Mantel–Cox) test. The synergistic effect of drug combinations was determined by the combination index (CI) using CompuSyn. The experiments were performed in at least triplicate, and statistical significance was defined as $p \leq 0.05.$ ## 2.7. Data Availability All data are available from the NCBI GEO Datasets (https://www.ncbi.nlm.nih.gov/geo/ (accessed on 10 October 2019)) via the accession numbers GSE188392, GSE187285, and GSE187293. ## 3.1. Sensitivity Profiling of NTX-301 in 199 CCLs In a recent companion paper, we clearly demonstrated that NTX-301 exhibited better therapeutic potential than conventional HMAs in preclinical models of AML [12]. However, the applicability of NTX-301 as a broad-spectrum anticancer agent has not yet been evaluated. To comprehensively assess the anticancer activity of NTX-301, we evaluated the viability of 199 CCLs after NTX-301 treatment based on two sensitivity metrics: the IC50 value and the area under the dose–response curve (AUC) (Supplementary Table S1). Consistent with the current use of HMAs as therapeutics for treating hematologic malignancies, NTX-301 displayed skewed sensitivity toward CCLs of hematopoietic origin (Figure 1A; odds ratio (OR)NTX-301 = 3.97, $$p \leq 0.0003$$). When analyzed with public data that are available through CTRP [21], DAC and AZA also showed the highest anticancer activity in hematopoietic CCLs (Figure 1B,C). The absolute cytotoxic activity of DAC (AUCDAC: 2.1–18.1) was much stronger than that of AZA (AUCAZA: 10.9–18.7), while that of NTX-301 could not be directly compared due to the use of different experimental platforms. NTX-301 and DAC exhibited relatively low efficacy against CCLs that originated from the breast, skin, and CNS (OR < 1), but these drugs exerted opposite effects against kidney-derived CCLs (ORNTX-301 = 3.20 vs. ORDAC = 0.27) (Figure 1A,B). We then compared the relative efficacy of NTX-301 and DAC against solid CCLs. NTX-301 treatment caused stronger cytotoxicity in 34 of 160 solid CCLs ($21.3\%$) compared with the median AUCNTX-301 observed in hematopoietic CCLs. However, DAC achieved higher efficacy in only 6 of 668 solid CCLs ($0.89\%$) compared with the median AUCDAC observed in hematopoietic CCLs. Thus, this sensitivity profiling revealed that NTX-301 primarily sensitizes hematologic malignancies and that it exhibits superior and broader anticancer activity against solid cancers. ## 3.2. Cellular Phenotypes Associated with the Antileukemic Activity of NTX-301 To dissect the primary activity of NTX-301 in hematopoietic CCLs, we assessed cellular phenotypic changes that occurred after NTX-301 treatment in detail. Cell viability assays revealed that NTX-301 exerted stronger effects than DAC on all five AML cell lines that were examined (Figure 2A). Even at early treatment stages (2 days), NTX-301 promoted stronger cytotoxicity than DAC (Figure 2A). Given the slow kinetics of demethylation, this rapid effect of NTX-301 may suggest that its effect is mediated by nonepigenetic mechanisms. Both agents tended to be more effective in p53-wild-type (WT) AML cells (MV4-11 and MOLM-13) than in p53-null AML cells (HL-60, THP-1, and KG-1) (Figure 2A). However, when applied for a long period of time (6 days), the potency of both agents was dramatically improved, regardless of p53 status (Figure 2A), suggesting the anticancer effects of HMAs through passive demethylation after multiple rounds of replication. Consistent with the early-onset effect upon NTX-301 treatment, cell cycle analysis revealed that NTX-301 increased S-phase arrest ($30\%$ with NTX-301 vs. $7\%$ with DAC) and decreased mitotic progression more significantly than DAC after 2 days of treatment (Figure 2B). Moreover, apoptosis analyses with two different methods (Annexin V staining and cleaved caspase-3) consistently showed that NTX-301 increased the apoptotic cell population more than DAC after 2 days of treatment (Figure 2C,D). Previously, long-term treatment with HMAs at low doses resulted in p53-independent anticancer effects dominated by AML differentiation rather than by cytotoxicity [22]. Hence, we finally performed an AML differentiation assay and found that a low dose of NTX-301 (60 nM) increased the population of CD11b+/CD14+ differentiated cells more markedly than DAC (Figure 2E). The change was accompanied by differentiation-like morphological changes (increased cell size, cytoplasm vacuolization, and decreased nuclear–cytoplasmic ratio; Figure 2F). Collectively, these results suggested that the superior antileukemic activity of NTX-301 was attributed to its more effective induction of cell cycle arrest, apoptosis, and differentiation. ## 3.3. Antitumor Efficacy of NTX-301 in a Preclinical Model of AML Given its superior efficacy in vitro, low doses of NTX-301 could achieve a profound therapeutic effect. Recently, even a lower dose of NTX-301 alone (2.0 mg/kg (p.o.)) resulted in greater efficacy and better survival outcomes than treatment with AZA alone (5.0 mg/kg (i.p.)) or the combination of AZA (2.5 mg/kg (i.p.)) + venetoclax (50–100 mg/kg (p.o.)) [ 12]. As a follow-up study, we sought to test the efficacy of low doses of NTX-301 in vivo using the same model as that in the previous experiment [12], namely, a systemic NOD/SCID mouse model that bears luciferase-labeled MV4-11 cell-derived tumors. NTX-301 is a good substrate for cytidine deaminase, which catalyzes the conversion of deoxycytidines into deoxyuridines [10]. Therefore, to prevent the clearance of NTX-301, we administered the cytidine deaminase inhibitor tetrahydrouridine (42.0 mg/kg) 1 h before NTX-301 treatment. After NTX-301 treatment at doses of 0.15, 0.30, and 0.45 mg/kg (p.o.), which were 13.3, 6.7, and 4.4 times lower than the previously used dose (2.0 mg/kg) [12], respectively, bioluminescence imaging showed substantial in vivo tumor regression (Figure 3A,B). These doses showed weaker activity than 2.0 mg/kg (p.o.) NTX-301, but the effects were comparable to those of 5.0 mg/kg (i.p.) AZA (Supplementary Figure S1), and no notable weight loss was observed (Figure 3C). ## 3.4. Transcriptome Analyses Cataloged the MoAs of NTX-301 Clinical responses to HMAs are frequently not apparent for up to six treatment cycles [23]. Given that HMAs are currently being used to treat elderly patients with AML who may not be able to tolerate long-term treatment, predicting the response to HMAs is important for avoiding unnecessary treatment. Hence, we attempted to identify markers of sensitivity to NTX-301, which also participate in the MoAs of NTX-301, through integrative multiomics analyses. First, to elucidate the MoAs of NTX-301, we examined the biological features of NTX-301-induced transcriptome alterations in the three AML CCLs (MV4-11, MOLM-13, and HL-60) using RNA sequencing. In support of the superior efficacy of NTX-301, heatmap analysis showed that NTX-301 elicited more intense and extensive transcriptional changes than DAC (Figure 4A). Gene set enrichment analysis (GSEA) revealed that the gene sets that were more strongly upregulated by NTX-301 than by DAC were associated with biological processes such as the DNA damage response (DDR), the p53 pathway, the immune response, and apoptosis (Figure 4B–D). In our previous study, the DDR and the p53 pathway were already described as the most significantly activated MoAs upon NTX-301 treatment [12]. Gene Ontology analysis also highlighted the enrichment of the DDR, the p53 pathway, and immune responses as activated hallmarks (Supplementary Figure S2), indicating their importance as MoAs. In addition, NTX-301 reversed immunotherapy resistance signatures [24] more significantly than DAC (Supplementary Figure S3), indicating its potential to be used in combination with immunotherapy [25]. Twenty genes that were commonly upregulated by NTX-301 in the three different CCLs were involved in immune activation, and these genes included known tumor suppressors such as BTG2 [26], ALOX5 [27], SOCS1 [28], and TP53INP1 [29] (Supplementary Figure S4). GSEA querying genes that were more strongly downregulated by NTX-301 implied the suppression of biological hallmarks, including cholesterol biosynthesis, the cell cycle, DNA replication, and pyrimidine metabolism (Figure 4B–D). Ingenuity Pathway Analysis (IPA) identified SREBF$\frac{1}{2}$ and mTORC$\frac{1}{2}$, which are master regulators of cholesterol synthesis and lipid metabolism, as upstream regulators of these downregulated genes (Supplementary Figure S5A). Notably, the downregulation of cholesterol biosynthesis-related genes was associated with enhanced overall survival in four AML cohorts (Supplementary Figure S5B), suggesting a clinical benefit of this MoA. NTX-301 also suppressed the oncogenic Myc and E2F signatures more strongly than DAC, which partially explained its higher efficacy (Supplementary Figure S6A,B). Moreover, NTX-301 repressed a transcriptional program associated with oxidative phosphorylation, which is a process that is essential for the maintenance and survival of leukemic stem cells [30], more significantly than DAC (Supplementary Figure S6C). NTX-301 also suppressed the transcription of pyrimidine metabolism-related genes (Figure 4B,C), and it is the known MoA of conventional HMAs [31]. ## 3.5. Integrative Data Analyses Identified Transcriptional Events Associated with Sensitivity to NTX-301 Next, to identify molecular events associated with sensitivity to NTX-301, we performed integrative analyses of the sensitivity profiles of 199 CCLs and multiomics data from the Cancer Cell Line Encyclopedia (CCLE). The 199 CCLs were classified into responders and nonresponders based on their AUC values (bottom third vs. top third), and CCLs with intermediate AUC values were excluded. To account for a confounding effect of tissue type, we included a binary covariate denoting whether a cell line was of hematopoietic origin. Then, we investigated transcriptional events associated with sensitivity to NTX-301 by examining differentially expressed genes between the responders and nonresponders. As a proof of concept, our analysis reproduced a previous finding related to conventional HMAs, namely, that the efficacy of HMA is correlated with the expression levels of the HMA metabolism-related genes DCK, CDA, and SAMHD1 (Supplementary Figure S7) [31,32]. In addition, our analysis comprehensively identified 362 sensitivity-associated and 325 resistance-associated genes (Padj < 0.05; Figure 5A, Supplementary Table S2). Subsequent GSEA revealed that the p53 pathway and apoptosis are sensitivity-associated processes and that the mTOR pathway is a resistance-associated process (Figure 5B). Similarly, an integrative analysis of IC50 values from 29 blood CCLs and reverse-phase protein array (RPPA) data from CCLE revealed that sensitive CCLs (low IC50) tended to exhibit higher expression levels of apoptotic proteins (e.g., cleaved PARP (r = −0.30) and Bak (r = −0.38)), whereas resistant CCLs (high IC50) tended to exhibit higher expression levels of the antiapoptotic protein Bcl-xL ($r = 0.35$) and the mTOR pathway-associated proteins phosphorylated-mTOR ($r = 0.37$) and phosphorylated-S6 ($r = 0.42$) (Figure 5C). Our analyses repeatedly identified the p53 pathway as a molecular event that is associated with sensitivity to NTX-301 as well as the MoA of NTX-301. To validate the impact of the pathway on sensitivity to NTX-301, we established TP53-knockout (KO) MV4-11 cells using CRISPR-Cas9. Indeed, TP53 KO dramatically decreased the efficacy of NTX-301, resulting in a >14-fold increase in IC50 (Figure 5D). We also noted that the resistance-associated mTOR pathway was essential for the tumorigenesis and propagation of AML as well as various solid tumors [33]. To confirm the impact of the mTOR pathway on resistance, we examined the efficacy of NTX-301 when used in combination with the mTOR inhibitors Torin1 and AZD-8055. As expected, the combination of the inhibitors synergistically improved the efficacy of NTX-301 (Figure 5E). Accordingly, the integrative data analyses revealed intrinsic mechanisms related to sensitivity/resistance to NTX-301. ## 3.6. Integrative Data Analyses Identified Mutation Events That Are Associated with Sensitivity to NTX-301 We next examined the somatic mutations associated with sensitivity to NTX-301. To exclude clinically irrelevant mutations, we filtered out mutations that are predicted to be passenger mutations and are not present in patient-derived tumors (TCGA data). The examination of differential mutational shifts between the responders and nonresponders showed that the responders tended to have a higher mutational burden than the nonresponders (Figure 6A). Indeed, AUC-based sensitivity values were significantly correlated with the number of mutations identified in CCLs (Figure 6B). A high mutational burden can be caused by the hyperproliferative features of cancer cells that bring about excessive replication stress and spontaneous DNA damage [34]. In fact, the responders tended to have shorter doubling times (Figure 6C) and tended to exhibit higher expression levels of pChk1, pRb, Myc, and FOXM1, which are required for hyperproliferation (Figure 5C). These results may support the hypothesis that NTX-301 more effectively sensitizes hyperproliferating cells because hyperproliferating cells rapidly incorporate NTX-301 during replication. Due to this mutational bias in the responders, we focused only on mutations that frequently occurred in the nonresponders and identified mutations in three genes: TP53, RB1, and NSD1 (Figure 6A). Consistent with the MoAs of NTX-301, the nonresponders showed a higher frequency of mutations in TP53 (OR = 0.41, $$p \leq 0.035$$) and RB1 (OR = 0.24, $$p \leq 0.044$$) than the responders. The NSD1 mutations that frequently occurred in the nonresponders (OR = 0.098, $$p \leq 0.017$$) were also noteworthy because NSD1 mutations have been demonstrated to trigger global hypomethylation [35,36], which may affect the efficacy of HMAs [37]. ## 3.7. Integrative Data Analyses Identified Epigenomic Events That Are Associated with Sensitivity to NTX-301 We then examined differential methylation patterns in the promoter regions of the responders versus nonresponders and found a substantial shift toward global hypermethylation in the responders (Figure 6D). By examining the 352 most differentially methylated regions (Padj < 0.05; hereafter abbreviated as DMR352; Supplementary Table S3), we found that the average level of DMR352 methylation showed a significant correlation with the log2-transformed IC50 value (Figure 6E). The significance of DMR352 far exceeded that of 100,000 iterations of correlations calculated from 352 regions that were randomly selected from 21,337 promoter regions (Figure 6F); these results indicated the importance of DMR352 as a sensitivity marker. DMR352 may have a functional relationship with HMA, as it significantly overlapped with the upstream promoter regions of genes whose expression is regulated by DAC (Supplementary Figure S8). The association of global hypermethylation with sensitivity led us to determine whether NTX-301 functions as an HMA. To decipher the influence of NTX-301 on the methylome, we investigated genome-wide changes in methylation in two CCLs (MV4-11 and HL-60) using an 850 K methylation array. Consistent with a prior finding that NTX-301 and DAC completely deplete DNMT1 [12], both agents promoted significant global demethylation (Figure 6G). However, despite their similar DNMT1-depleting activities, the demethylation activity of NTX-301 was unexpectedly weaker than that of DAC (Figure 6G). The examination of differentially methylated CpGs (DMCs) also showed that cells treated with these two agents shared many DMCs, but NTX-301-induced DMCs were subordinate to DAC-induced DMCs (Figure 6H). We hypothesized that the distinct demethylation activities of these two drugs might be attributed to their differential activity in inducing S-phase arrest (Figure 2B), since the DNMT1-mediated maintenance of methylation depends on DNA replication [38]. Therefore, we explored the degree of demethylation during the progression of replication using Repli-seq data. Surprisingly, the CpGs that were demethylated by NTX-301 mapped predominantly to genomic regions that were replicated during the early S-phase (early replication timing) in MV4-11 cells, which were arrested in the S-phase (Figure 7A). Moreover, the changes in demethylation (∆β) induced by NTX-301, but not those induced by DAC, displayed a marked inverse correlation with replication timing (Figure 7B). The effect was greatly diminished in HL-60 cells, which were not arrested in the S-phase, although NTX-301 still led to a slight skewing of demethylation toward early replication (Figure 7A). These results suggest that the intra-S-phase arrest induced by NTX-301 allows for preferential demethylation during early replication and prevents replication-coupled passive demethylation. To further compare the demethylation activities of NTX-301 and DAC in various genomic contexts, we analyzed methylome data from HL-60 cells, which do not exhibit notable S-phase arrest and are thus sensitive to NTX-301-induced demethylation. The normalization of the number of DMCs that are demethylated by NTX-301 (vs. DAC) showed that the demethylation activity of NTX-301 was weaker than that of DAC in CpG islands but stronger than that of DAC in promoter regions (TSS1500) (Figure 7C). The demethylation activities of both agents were comparable in gene bodies and noncoding regions (Figure 7C). Since DNA replication is spatially and temporally coordinated with chromatin organization [39], we also evaluated the demethylation of chromatin features annotated using ChIP-seq data. The demethylation activity of NTX-301 was considerably lower than that of DAC in H3K9me3-marked regions, which are associated with heterochromatin and late replication timing, but the activity of NTX-301 was higher than that of DAC in H3K27ac-marked regions that are associated with open chromatin and active gene regulation (Figure 7D). Overall, our findings suggested that basal methylation levels had value for predicting NTX-301 sensitivity, while on-treatment demethylation levels may not properly reflect the efficacy of NTX-301 due to its unique demethylation patterns. ## 4. Discussion This study demonstrated previously unrecognized characteristics of NTX-301 through comparative analyses with conventional HMAs. Experimental investigations revealed that NTX-301 exerts primary anticancer effects on blood cancers, while it exhibits improved sensitivity profiles in solid cancers; additionally, integrative multiomics analyses revealed the MoAs and demethylation activity of NTX-301 as well as markers of sensitivity to NTX-301. Despite minor differences in chemical structures, NTX-301 exhibited substantially improved therapeutic potential at the molecular and cellular levels compared with conventional HMAs. In particular, low doses of NTX-301 (0.15, 0.30, and 0.45 mg/kg (p.o.)) achieved profound antitumor efficacy in vivo, showing comparable effects to a high dose of AZA (5.0 mg/kg (i.p.)). Improved anticancer effects when applied at low concentrations for a long period of time (Figure 2A) may indicate that NTX-301 has an anticancer effect through the involvement of passive demethylation after multiple rounds of replication under the depletion of DNMT1. These findings also suggest a therapeutic advantage of NTX-301 in overcoming toxicity when applied with conventional HMAs at high concentrations. According to our study and other studies [10,12], NTX-301 may satisfy many of the requirements for next-generation HMAs: [1] improved efficacy, [2] better survival benefits, [3] lower toxicity, [4] improved chemical stability and oral bioavailability, and [5] easy activation into triphosphate and incorporation into DNA without replication termination. Nonetheless, in-depth studies are still needed to address how NTX-301 is metabolized in cells and what off-target activity the metabolites possess. Cellular and molecular analyses revealed MoAs underlying the superior efficacy of NTX-301. In addition to its reported role in promoting stronger transcriptional reprogramming toward normal myeloid-like signatures [12], NTX-301 promoted more robust and extensive changes in the transcription of antileukemic hallmark genes than DAC. In particular, the DDR-p53 pathway may be an essential MoA, as transcriptome analyses repeatedly pinpointed this pathway, and p53 KO dramatically decreased the efficacy of NTX-301. The pathway acts as an antileukemic barrier by enforcing DDR-induced AML cell differentiation [40]. While the exact mechanisms by which NTX-301 predominantly activates the DDR are unknown, two possible mechanisms can be suggested. The first possible mechanism is that the bulky DNMT1 adducts that are trapped in NTX-301-substituted DNA can obstruct oncoming replication forks, causing replication fork collapse and DNA double-strand breaks [41]. The second possible mechanism involves the disruption of pyrimidine metabolism. Unbalanced dNTP pools cause the misincorporation of nucleotides and induce DDR [42]. In fact, our transcriptome analysis suggested that pyrimidine metabolism was inhibited after NTX-301 treatment (Figure 4B,C). The induction of immune responses, which is another important MoA that was elucidated by transcriptome analyses, not only reflects NTX-301-induced AML cell differentiation but may also represent epigenetic immunosensitization [43]. Recently, HMAs were found to trigger type I interferon signaling via double-stranded RNA and upregulate surface antigens, viral defense pathway components, and PD-L1 [44]. Therefore, the effect of NTX-301 when used in combination with immunotherapy is an interesting area to investigate. The inhibition of cholesterol biosynthesis-related genes is also another notable MoA because cholesterol starvation was previously shown to induce AML cell differentiation [45], and cholesterol synthesis inhibitors sensitized AML to standard antileukemic regimens by blocking adaptive cholesterol responses [46,47]. Our findings suggested that NTX-301 may be applicable to the treatment of solid tumors as well as hematologic malignancies. Among the 199 CCLs, the potency of NTX-301 against solid CCLs greatly surpassed that of conventional HMAs. Notably, NTX-301 showed promising preliminary efficacy against advanced solid tumors (NCT03366116), achieving stable disease in 11 of 14 patients (disease control rate = $78.6\%$). Although none of the conventional HMAs have been approved for the treatment of solid tumors, this observation provides a rationale for a current phase I trial that is evaluating the safety and tolerability of NTX-301 when used in combination with platinum-based chemotherapies for the treatment of ovarian and bladder cancer (NCT04851834). Our biomarker analyses revealed various molecular events that are associated with sensitivity to NTX-301. Although these results are limited due to a lack of clinical validation, they provide insights into the important aspects of NTX-301 treatment: [1] the role of global hypermethylation or DMR352 as a predictor of sensitivity, [2] the identity of NTX-301 as an HMA that more effectively sensitizes CCLs with aberrant hypermethylation, [3] the high concordance between markers of sensitivity and MoAs, and [4] the role of the mTOR pathway as an intrinsic mechanism underlying resistance to NTX-301. We therefore anticipate that these biomarkers will be validated in clinical trials. ## 5. Conclusions Our study provides a rationale for the further development of NTX-301 for use in the clinic. We believe that the MoAs and biomarkers identified in this work will improve our understanding of the potential of NTX-301 to function as a next-generation HMA and facilitate patient stratification during the development of NTX-301, thus providing a novel therapeutic option for cancer patients. ## References 1. Yu X., Li M., Guo C., Wu Y., Zhao L., Shi Q., Song J., Song B.. Therapeutic Targeting of Cancer: Epigenetic Homeostasis. Front. Oncol. (2021) 11 747022. DOI: 10.3389/fonc.2021.747022 2. Baylin S.B., Jones P.A.. Epigenetic Determinants of Cancer. Cold Spring Harb. Perspect. Biol. (2016) 8 a019505. DOI: 10.1101/cshperspect.a019505 3. Hanahan D., Weinberg R.A.. Hallmarks of cancer: The next generation. Cell (2011) 144 646-674. DOI: 10.1016/j.cell.2011.02.013 4. Moore L.D., Le T., Fan G.. DNA methylation and its basic function. NeuropsychoPharmacology (2013) 38 23-38. DOI: 10.1038/npp.2012.112 5. Williams B.P., Gehring M.. Principles of Epigenetic Homeostasis Shared Between Flowering Plants and Mammals. Trends Genet. (2020) 36 751-763. DOI: 10.1016/j.tig.2020.06.019 6. Schoofs T., Berdel W.E., Muller-Tidow C.. Origins of aberrant DNA methylation in acute myeloid leukemia. Leukemia (2014) 28 1-14. DOI: 10.1038/leu.2013.242 7. Stresemann C., Lyko F.. Modes of action of the DNA methyltransferase inhibitors azacytidine and decitabine. Int. J. Cancer (2008) 123 8-13. DOI: 10.1002/ijc.23607 8. Sato T., Issa J.J., Kropf P.. DNA Hypomethylating Drugs in Cancer Therapy. Cold Spring Harb. Perspect. Med. (2017) 7 a026948. DOI: 10.1101/cshperspect.a026948 9. Pollyea D.A., Pratz K., Letai A., Jonas B.A., Wei A.H., Pullarkat V., Konopleva M., Thirman M.J., Arellano M., Becker P.S.. Venetoclax with azacitidine or decitabine in patients with newly diagnosed acute myeloid leukemia: Long term follow-up from a phase 1b study. Am. J. Hematol. (2021) 96 208-217. DOI: 10.1002/ajh.26039 10. Parker W.B., Thottassery J.V.. 5-Aza-4′-thio-2′-deoxycytidine, a New Orally Bioavailable Nontoxic “Best-in-Class”: DNA Methyltransferase 1-Depleting Agent in Clinical Development. J. Pharmacol. Exp. Ther. (2021) 379 211-222. DOI: 10.1124/jpet.121.000758 11. Pappalardi M.B., Keenan K., Cockerill M., Kellner W.A., Stowell A., Sherk C., Wong K., Pathuri S., Briand J., Steidel M.. Discovery of a first-in-class reversible DNMT1-selective inhibitor with improved tolerability and efficacy in acute myeloid leukemia. Nat. Cancer (2021) 2 1002-1017. DOI: 10.1038/s43018-021-00249-x 12. Lim B., Yoo D., Chun Y., Go A., Cho K.J., Choi D., Jung M.E., Lee H.Y., Boohaker R.J., Lee J.S.. The preclinical efficacy of the novel hypomethylating agent NTX-301 as a monotherapy and in combination with venetoclax in acute myeloid leukemia. Blood Cancer J. (2022) 12 57. DOI: 10.1038/s41408-022-00664-y 13. Thottassery J.V., Sambandam V., Allan P.W., Maddry J.A., Maxuitenko Y.Y., Tiwari K., Hollingshead M., Parker W.B.. Novel DNA methyltransferase-1 (DNMT1) depleting anticancer nucleosides, 4’-thio-2’-deoxycytidine and 5-aza-4′-thio-2′-deoxycytidine. Cancer ChemoTher. Pharmacol. (2014) 74 291-302. DOI: 10.1007/s00280-014-2503-z 14. Bibikova M., Lin Z., Zhou L., Chudin E., Garcia E.W., Wu B., Doucet D., Thomas N.J., Wang Y., Vollmer E.. High-throughput DNA methylation profiling using universal bead arrays. Genome Res. (2006) 16 383-393. DOI: 10.1101/gr.4410706 15. Bolger A.M., Lohse M., Usadel B.. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics (2014) 30 2114-2120. DOI: 10.1093/bioinformatics/btu170 16. Kim D., Langmead B., Salzberg S.L.. HISAT: A fast spliced aligner with low memory requirements. Nat. Methods (2015) 12 357-360. DOI: 10.1038/nmeth.3317 17. Pertea M., Pertea G.M., Antonescu C.M., Chang T.C., Mendell J.T., Salzberg S.L.. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads. Nat. Biotechnol. (2015) 33 290-295. DOI: 10.1038/nbt.3122 18. Chen E.Y., Tan C.M., Kou Y., Duan Q., Wang Z., Meirelles G.V., Clark N.R., Ma’ayan A.. Enrichr: Interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinform. (2013) 14. DOI: 10.1186/1471-2105-14-128 19. Mootha V.K., Lindgren C.M., Eriksson K.F., Subramanian A., Sihag S., Lehar J., Puigserver P., Carlsson E., Ridderstrale M., Laurila E.. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. (2003) 34 267-273. DOI: 10.1038/ng1180 20. Subramanian A., Tamayo P., Mootha V.K., Mukherjee S., Ebert B.L., Gillette M.A., Paulovich A., Pomeroy S.L., Golub T.R., Lander E.S.. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA (2005) 102 15545-15550. DOI: 10.1073/pnas.0506580102 21. Rees M.G., Seashore-Ludlow B., Cheah J.H., Adams D.J., Price E.V., Gill S., Javaid S., Coletti M.E., Jones V.L., Bodycombe N.E.. Correlating chemical sensitivity and basal gene expression reveals mechanism of action. Nat. Chem. Biol. (2016) 12 109-116. DOI: 10.1038/nchembio.1986 22. Ng K.P., Ebrahem Q., Negrotto S., Mahfouz R.Z., Link K.A., Hu Z., Gu X., Advani A., Kalaycio M., Sobecks R.. p53 independent epigenetic-differentiation treatment in xenotransplant models of acute myeloid leukemia. Leukemia (2011) 25 1739-1750. DOI: 10.1038/leu.2011.159 23. Santini V., Ossenkoppele G.J.. Hypomethylating agents in the treatment of acute myeloid leukemia: A guide to optimal use. Crit. Rev. Oncol. Hematol. (2019) 140 1-7. DOI: 10.1016/j.critrevonc.2019.05.013 24. Jerby-Arnon L., Shah P., Cuoco M.S., Rodman C., Su M.J., Melms J.C., Leeson R., Kanodia A., Mei S., Lin J.R.. A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade. Cell (2018) 175 984-997.e924. DOI: 10.1016/j.cell.2018.09.006 25. Villanueva L., Alvarez-Errico D., Esteller M.. The Contribution of Epigenetics to Cancer Immunotherapy. Trends Immunol. (2020) 41 676-691. DOI: 10.1016/j.it.2020.06.002 26. Yuniati L., Scheijen B., van der Meer L.T., van Leeuwen F.N.. Tumor suppressors BTG1 and BTG2: Beyond growth control. J. Cell. Physiol. (2019) 234 5379-5389. DOI: 10.1002/jcp.27407 27. Wang Y., Skibbe J.R., Hu C., Dong L., Ferchen K., Su R., Li C., Huang H., Weng H., Huang H.. ALOX5 exhibits anti-tumor and drug-sensitizing effects in MLL-rearranged leukemia. Sci. Rep. (2017) 7 1853. DOI: 10.1038/s41598-017-01913-y 28. Zhang X.H., Yang L., Liu X.J., Zhan Y., Pan Y.X., Wang X.Z., Luo J.M.. Association between methylation of tumor suppressor gene SOCS1 and acute myeloid leukemia. Oncol. Rep. (2018) 40 1008-1016. DOI: 10.3892/or.2018.6508 29. Seillier M., Peuget S., Gayet O., Gauthier C., N’Guessan P., Monte M., Carrier A., Iovanna J.L., Dusetti N.J.. TP53INP1, a tumor suppressor, interacts with LC3 and ATG8-family proteins through the LC3-interacting region (LIR) and promotes autophagy-dependent cell death. Cell Death Differ. (2012) 19 1525-1535. DOI: 10.1038/cdd.2012.30 30. Kuntz E.M., Baquero P., Michie A.M., Dunn K., Tardito S., Holyoake T.L., Helgason G.V., Gottlieb E.. Targeting mitochondrial oxidative phosphorylation eradicates therapy-resistant chronic myeloid leukemia stem cells. Nat. Med. (2017) 23 1234-1240. DOI: 10.1038/nm.4399 31. Gu X., Tohme R., Tomlinson B., Sakre N., Hasipek M., Durkin L., Schuerger C., Grabowski D., Zidan A.M., Radivoyevitch T.. Decitabine- and 5-azacytidine resistance emerges from adaptive responses of the pyrimidine metabolism network. Leukemia (2021) 35 1023-1036. DOI: 10.1038/s41375-020-1003-x 32. Oellerich T., Schneider C., Thomas D., Knecht K.M., Buzovetsky O., Kaderali L., Schliemann C., Bohnenberger H., Angenendt L., Hartmann W.. Selective inactivation of hypomethylating agents by SAMHD1 provides a rationale for therapeutic stratification in AML. Nat. Commun. (2019) 10 3475. DOI: 10.1038/s41467-019-11413-4 33. Hoshii T., Tadokoro Y., Naka K., Ooshio T., Muraguchi T., Sugiyama N., Soga T., Araki K., Yamamura K., Hirao A.. mTORC1 is essential for leukemia propagation but not stem cell self-renewal. J. Clin. Investig. (2012) 122 2114-2129. DOI: 10.1172/JCI62279 34. Tubbs A., Nussenzweig A.. Endogenous DNA Damage as a Source of Genomic Instability in Cancer. Cell (2017) 168 644-656. DOI: 10.1016/j.cell.2017.01.002 35. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature (2015) 517 576-582. DOI: 10.1038/nature14129 36. Choufani S., Cytrynbaum C., Chung B.H., Turinsky A.L., Grafodatskaya D., Chen Y.A., Cohen A.S., Dupuis L., Butcher D.T., Siu M.T.. NSD1 mutations generate a genome-wide DNA methylation signature. Nat. Commun. (2015) 6 10207. DOI: 10.1038/ncomms10207 37. Yan P., Frankhouser D., Murphy M., Tam H.H., Rodriguez B., Curfman J., Trimarchi M., Geyer S., Wu Y.Z., Whitman S.P.. Genome-wide methylation profiling in decitabine-treated patients with acute myeloid leukemia. Blood (2012) 120 2466-2474. DOI: 10.1182/blood-2012-05-429175 38. Chilakala S., Feng Y., Li L., Mahfouz R., Quteba E., Saunthararajah Y., Xu Y.. Tracking Decitabine Incorporation into Malignant Myeloid Cell DNA in vitro and in vivo by LC-MS/MS with Enzymatic Digestion. Sci. Rep. (2019) 9 4558. DOI: 10.1038/s41598-019-41070-y 39. Fu H., Baris A., Aladjem M.I.. Replication timing and nuclear structure. Curr. Opin. Cell Biol. (2018) 52 43-50. DOI: 10.1016/j.ceb.2018.01.004 40. Santos M.A., Faryabi R.B., Ergen A.V., Day A.M., Malhowski A., Canela A., Onozawa M., Lee J.E., Callen E., Gutierrez-Martinez P.. DNA-damage-induced differentiation of leukaemic cells as an anti-cancer barrier. Nature (2014) 514 107-111. DOI: 10.1038/nature13483 41. Orta M.L., Calderon-Montano J.M., Dominguez I., Pastor N., Burgos-Moron E., Lopez-Lazaro M., Cortes F., Mateos S., Helleday T.. 5-Aza-2’-deoxycytidine causes replication lesions that require Fanconi anemia-dependent homologous recombination for repair. Nucleic Acids Res. (2013) 41 5827-5836. DOI: 10.1093/nar/gkt270 42. Hastak K., Paul R.K., Agarwal M.K., Thakur V.S., Amin A.R., Agrawal S., Sramkoski R.M., Jacobberger J.W., Jackson M.W., Stark G.R.. DNA synthesis from unbalanced nucleotide pools causes limited DNA damage that triggers ATR-CHK1-dependent p53 activation. Proc. Natl. Acad. Sci. USA (2008) 105 6314-6319. DOI: 10.1073/pnas.0802080105 43. Zhang H., Pandey S., Travers M., Sun H., Morton G., Madzo J., Chung W., Khowsathit J., Perez-Leal O., Barrero C.A.. Targeting CDK9 Reactivates Epigenetically Silenced Genes in Cancer. Cell (2018) 175 1244-1258.e26. DOI: 10.1016/j.cell.2018.09.051 44. Chiappinelli K.B., Strissel P.L., Desrichard A., Li H., Henke C., Akman B., Hein A., Rote N.S., Cope L.M., Snyder A.. Inhibiting DNA Methylation Causes an Interferon Response in Cancer via dsRNA Including Endogenous Retroviruses. Cell (2015) 162 974-986. DOI: 10.1016/j.cell.2015.07.011 45. Sanchez-Martin C.C., Davalos A., Martin-Sanchez C., de la Pena G., Fernandez-Hernando C., Lasuncion M.A.. Cholesterol starvation induces differentiation of human leukemia HL-60 cells. Cancer Res. (2007) 67 3379-3386. DOI: 10.1158/0008-5472.CAN-06-4093 46. Li H.Y., Appelbaum F.R., Willman C.L., Zager R.A., Banker D.E.. Cholesterol-modulating agents kill acute myeloid leukemia cells and sensitize them to therapeutics by blocking adaptive cholesterol responses. Blood (2003) 101 3628-3634. DOI: 10.1182/blood-2002-07-2283 47. Chen F., Wu X., Niculite C., Gilca M., Petrusca D., Rogozea A., Rice S., Guo B., Griffin S., Calin G.A.. Classic and targeted anti-leukaemic agents interfere with the cholesterol biogenesis metagene in acute myeloid leukaemia: Therapeutic implications. J. Cell. Mol. Med. (2020) 24 7378-7392. DOI: 10.1111/jcmm.15339
--- title: Vitamin A Deficiency and Its Association with Visceral Adiposity in Women authors: - Érica Góes - Adryana Cordeiro - Claudia Bento - Andrea Ramalho journal: Biomedicines year: 2023 pmcid: PMC10046475 doi: 10.3390/biomedicines11030991 license: CC BY 4.0 --- # Vitamin A Deficiency and Its Association with Visceral Adiposity in Women ## Abstract Body adiposity is associated with increased metabolic risk, and evidence indicates that vitamin A is important in regulating body fat. The aim of this study was to evaluate serum concentrations of vitamin A and its association with body adiposity in women with the recommended intake of vitamin A. A cross-sectional study was designed with 200 women divided into four groups according to Body Mass Index (BMI): normal weight (NW), overweight (OW), class I obesity (OI), and class 2 obesity (OII). The cut-off points to assess inadequate participants were retinol < 1.05 µmol/L and β-carotene < 40 µg/dL. Body adiposity was assessed through different parameters and indexes, including waist circumference (WC), waist-to-height ratio (WHtR), hypertriglyceridemic waist (HW), lipid accumulation product (LAP), Visceral Adiposity Index (VAI), and Body Adiposity Index (BAI). It was observed that $55.5\%$ of women had low serum concentrations of β-carotene (34.9 ± 13.8 µmol/L, $p \leq 0.001$) and $43.5\%$ had low concentrations of retinol (0.71 ± 0.3 µmol/L, $p \leq 0.001$). Women classified as OI and OII had lower mean values of β-carotene (OI—35.9 ± 4.3 µg/dL: OII—32.0 ± 0.9 µg/dL [$p \leq 0.001$]). IAV showed significant negative correlation with retinol (r = −0.73, $p \leq 0.001$). Vitamin A deficiency is associated with excess body adiposity in women with the recommended intake of vitamin. Greater body adiposity, especially visceral, was correlated with reduced serum concentrations of vitamin A. ## 1. Introduction Obesity rates have grown progressively over the years. The most recent data indicate that in, the period from 1975 to 2016, the prevalence of obesity tripled, meaning that 650 million individuals now have obesity [1]. If current trends continue, by 2030 more than half of the adult population will be overweight [2]. It has been noted that women are gaining more prominence for having the highest rates of overweight and obesity across all age groups, when compared to men [3]. As such, obesity affects $15\%$ of women and $11\%$ of men worldwide [1]. This profile can be found in many countries. In the United States, women had a higher percentage of obesity ($40\%$) compared to men ($38\%$) [4]. An analysis of 105 different countries showed that women continued to stand out with the higher prevalence of being overweight and suffering from obesity when compared to men [5]. Among men, the rate is lower, reaching $21.8\%$ and $57.5\%$, respectively [6]. It is estimated that if there is a reduction of at least $5\%$ in the Body Mass Index (BMI) at a population level, in 2030 there would be a reduction in medical costs related to obesity of at least EUR 495 million over the next 20 years [7]. Following global trends, in Brazil, for 17 years (2002–2019), women have had the highest rates of obesity and being overweight. The most current national statistics shows that $29.5\%$ of women suffer from obesity (approximately one in every three women) and $62.6\%$ are overweight [6]. Excess body adiposity is an important risk factor for the development of Chronic Non-Communicable Diseases (NCDs), including Type 2 Diabetes Mellitus (DM2), some types of cancer, and cardiovascular disease (CVD) [1]. Body adiposity is commonly classified using the BMI [1]. However, it is known that BMI has limitations in measuring body adiposity and identifying its distribution and that such distribution is strongly correlated with the development of comorbidities and metabolic risk [8]. Besides BMI, other parameters for assessing body adiposity support the prognosis of metabolic risks resulting from excess fat mass, especially visceral fat mass, and its associated diseases. Evidence points to vitamin A (VA) as an important regulator of body fat reserve through actions on nuclear receptors in both liver and adipose tissue [9]. Retinol metabolites regulate body adiposity since most of their effects are anti-adipogenic; they inhibit adipocyte differentiation and intracellular lipid accumulation [10]. Findings in the literature show that higher BMI is related to lower serum concentrations of VA, suggesting that vitamin A deficiency (VAD) may be related to excess body weight [11,12]. Some other studies have shown that lower VA intake was positively associated with excess adiposity [13,14]. There are also studies showing lower serum concentrations of VA in individuals with obesity compared to normal weight individuals. They also suggest that inadequate dietary intake of this vitamin is the primary cause of VAD, but no assessment of the dietary intake of this nutrient has been performed [15]. Due to the scarcity of studies evaluating the relationship between VA and body adiposity, especially visceral adiposity, the current study aims to assess serum concentrations of VA and its relationship with body adiposity by using Waist Circumference (WC), Waist-to-Height Ratio (WHtR), Hypertriglyceridemic Waist (HW), Body Adiposity Index (BAI), Visceral Adiposity Index (VAI), and Lipid Accumulation Product (LAP) as complementary parameters to assess body adiposity in women with the recommended intake of this vitamin. ## 2. Materials and Methods This is a descriptive cross-sectional study conducted from January 2019 to October 2021, comprising adult women in different BMI classes who attended the Nutrition Service of a Municipal Health Unit, located in Rio de Janeiro. To participate in the study, women were required to be aged <20; 59>. Exclusion criteria were unmet dietary recommendation for VA, pregnant and lactating women, those with liver diseases (except non-alcoholic fatty liver disease), disabsorptive syndromes and surgeries, acute infection, excessive alcohol intake (≥45 g), chemical dependence, nephropathies, acquired immunodeficiency syndrome, cancer, and/or the use of supplements containing VA in the six months prior to collection. This study was approved by the Research Ethics Committee of Hospital Universitário Clementino Fraga Filho, Federal University of Rio de Janeiro, Brazil (Research Protocol number $\frac{011}{15}$-CEP). ## 2.1. Sample Size The sample size was calculated based on a national study that assessed the Brazilian prevalence of micronutrient inadequacy (POF 2017–2018) [16]. Based on these findings, the prevalence of inadequacy for vitamin A in the female population is $80.1\%$. To obtain the sample size with a $95\%$ confidence interval, considering the prevalence of adequacy of $20\%$ and with a sampling error of $5\%$, 148 women with the recommended daily food intake according to the Institute of Medicine [17] would be needed to conduct the present study. ## 2.2. Selection of Study Participants The first phase of the research consisted of the application of the exclusion criteria; it was carried out to assess the level of vitamin A consumption. For those who met the established profile, there was clarification about the objectives and procedures of the study; then, the Informed Consent Form was provided to the women who agreed to participate in the study, who then signed and returned the form. After applying the exclusion criteria (Figure 1), 849 women who did not meet the required criteria exited the study, and those who met the recommendation for daily VA intake continued ($$n = 231$$). ## Assessment of Vitamin A Dietary The assessment of vitamin A dietary intake was carried out using the three-day food intake record (two during the week and one on the weekend); this record quantified the average of vitamin A dietary. Additionally, the study used the 24 h recall and the food consumption frequency questionnaires. The 24 h recall was used to obtain information about the consumption of vitamin A on the day before the interview. It considered the preparation of food and information on the weight and size of portions in grams, milliliters, and in household measurements. During data collection, each woman was instructed to ensure that no meals or snacks were missed. The primary purpose of the 24 h recall was to enable the population under study to understand the correct filling of the instrument that quantified the average intake of vitamin A during the three-day food intake record. This was performed with a view to minimizing filling errors and improving the quality of information. The three-day food intake record, including two days during the week and one on the weekend, was the method used to quantify the average intake of vitamin A in the study, as it was carried out at the time the food was being consumed. Thus, it was not based on the individual’s memory; instead, it measured current consumption and identified the types of food consumed and meal times. The responses obtained using the three-day food intake record were inserted and plotted in Dietbox software, which calculated the average daily VA intake according to the VA content in foods as published in the Brazilian Table of Food Composition [18], which is integrated into the Dietbox software. The portion size was evaluated using the Photo Atlas of Food Portion Sizes [19]. VA intake was compared with the daily intake values recommended by the Institute of Medicine [17]. The cutoff point adopted for the recommended dietary intake of VA was 700 μg/day of retinol equivalent. The food consumption frequency questionnaire, based on a list of food sources of vitamin A recommended by the International Vitamin A Consultative Group (IVACG) [20], was used to determine how often these foods were consumed (daily, weekly, every two weeks, monthly, or never). The information was used during the nutritional guidance stage, and was provided to all the women who participated in the study. This guidance aimed to promote the necessary dietary changes, indicating the inclusion, greater frequency of consumption, or even the exclusion of foods in the nutritional guidance stage of our study. One month after the first procedure, all women participating in the first stage of the study were scheduled to receive information about nutritional diagnosis, including VA intake, in addition to relevant dietary guidelines. Only women who had a recommended dietary intake of VA according to the results of the dietary intake surveys used in the study were included ($$n = 231$$). There was a loss of 31 women due to questionnaire completion errors and failure to submit the records. Two hundred women continued in the study and moved on to the third phase, where their body composition was assessed. Subsequently, the sample was divided into four groups according to BMI ranges: normal weight (NW), overweight (OW), obesity class I (OI), and obesity class II (OII). Participants were instructed to fast before blood collection. After this procedure, the women received the results of the biochemical tests, the diagnosis of the nutritional status of VA accompanied by appropriate nutritional guidance, and a prescription of VA supplementation (5000 IU/day of retinol palmitate for 12 weeks) for all those who had VAD. ## 2.3. Assessment of Body Variables Weight and height were evaluated, and BMI was calculated by dividing body weight (in kilograms—kg) by height (in square meters—m2). The classification of this variable was performed considering the cut-off points proposed by the World Health Organization (WHO): normal weight, between 18.5 and 24.9 kg/m2; overweight, between 25.0 and 29.9 kg/m2; class I obesity, between 30.0 and 34.9 kg/m2; class II obesity, between 35.0 and 39.9 kg/m2; and class III obesity, ≥40 kg/m2 [1]. Abdominal adiposity was assessed using WC, WHtR HW, and BAI, while visceral adiposity was estimated by the VAI and LAP. The formulas and cut-off points used are presented below:WC is a widely used anthropometric parameter to assess abdominal fat. It was considered high if >88 cm [1].WHtR is applied to diagnose abdominal obesity and plays an important role in assessing the risk of cardiovascular events. It was calculated using WC (in cm) divided by Height (in cm), with a cut-off point >0.5 [21].HW is a marker for the simultaneous presence of WC and elevated serum triglyceride levels. It is a simple and practical indicator that can be used as a predictor of metabolic disease. It is characterized by the simultaneous presence of increased WC (≥80 cm) and elevated serum triglyceride (TG) levels (≥1.7 mmol/L) [22].BAI evaluates the percentage of body fat in adults. It is a method used to estimate body adiposity and is considered an alternative predictor of body fat in the absence of more complex techniques or more expensive methods. According to the formula: (hip circumference (cm) ÷ height (m) 1.5) − 18; the cut-off point is >33 [23].VAI can estimate the distribution of fat and the dysfunction of the visceral adipose tissue, resulting from a specific mathematical formula for each gender. According to the formula: (WC (cm) ÷ (36.58 + (BMI * 1.89) * (TG ÷ 0.81) * (1.52 ÷ HDL-c) for women, where TG and high-density lipoprotein cholesterol (HDL-c) are expressed in mmol/L, with a cut-off point >1 [24].LAP can represent lipotoxicity and may be a marker of abdominal adiposity that correlates with central fat accumulation. This index was calculated: (WC (cm) − 58) × (TG (mmol/L)). The cut-off point used was 37.9 [25]. ## 2.4. Biochemical Measurements of Vitamin A For the biochemical assessment of VA, a blood sample was obtained using venipuncture after 12 h fasting. Serum concentrations of retinol and β-carotene were quantified using High-Performance Liquid Chromatography with an ultraviolet detector (HPLC-UV); the following cut-off points were used for inadequate retinol (<1.05 µmol/L) and low β-carotene concentrations (<40 µg/dL) [26]. ## 2.5. Other Biochemical Measurements For other biochemical assessments, a blood sample was obtained using venipuncture after 12 h fasting. Laboratory tests were performed to assess the lipid profile (total cholesterol, TG, HDL-c, and low-density lipoprotein cholesterol [LDL-c]). The following concentrations were considered normal values: total cholesterol < 200 mg/dL; TG < 150 mg/dL; HDL-c > 50 mg/dL; and LDL-c < 150 mg/dL. The determinations of TG, total cholesterol, and HDL-c were performed using the colorimetric enzymatic method. Reagents for these biochemical evaluations were purchased from Labtest Diagnóstica S.A., Minas Gerais, Brazil. The LDL-c fraction was determined according to the Friedewald formula. ## 2.6. Statistical Analysis The Kolmogorov–Smirnov test was used to test the normality of data and expressed as means and standard deviations for clinical, dietary, and biochemical variables. Analysis of variance (ANOVA) and Bonferroni’s test for multiple comparisons were used. Pearson’s correlation coefficient was applied for serum concentrations of retinol and β-carotene with body adiposity variables. The significance level adopted was $5\%$ ($p \leq 0.05$). Statistical analysis was performed using the Statistical Package for the Social Sciences for Windows version 26. ## 3. Results The sample comprised 200 adult women who met the recommended dietary intake of VA (772.3 ± 59.9 μg/day—retinol equivalent). The mean age was 50.0 ± 5.6 years. Related to body composition of the total sample, the mean BMI, WC, WHtR, VAI, BAI and LAP were 28.1 ± 5.1 Kg/m2, 102.5 ± 21.0 cm, 0.6 ± 0.1, 4.4 ± 1.1, 31.2 ± 8.7 and 90.3 ± 55.1, respectively. The general characteristics of the sample, according to BMI ranges, are described in Table 1. According to the results, $59.5\%$ of women belonged to the HW phenotype ($$n = 119$$). However, there was no relationship between the HW phenotype and serum concentrations of VA. There was greater inadequacy of β-carotene when compared to retinol, where $55.5\%$ of women ($$n = 101$$) showed deficiency in β-carotene (34.9 ± 13.8 µmol/L, $p \leq 0.001$), while $43.5\%$ ($$n = 87$$) showed deficiency in retinol (0.71 ± 0.3 µmol/L, $p \leq 0.001$). The mean serum concentrations of retinol and β-carotene of the groups are shown in Table 2. There was a significant decrease in serum concentrations of retinol and β-carotene as BMI increased. In addition, women classified as OI and OII had a deficiency in both retinol and β-carotene; women classified as OW had a deficiency in retinol. Only women classified as NW had adequate retinol and β-carotene levels. Body adiposity parameters according to the nutritional status of retinol, β-carotene, and BMI groups are shown in Table 3. Greater inadequacy of Vitamin A, both retinol and β-carotene, can be observed as body adiposity increased, especially visceral adiposity. A significant negative correlation was found between serum concentrations of retinol and β-carotene with body adiposity parameters (Table 4). ## 4. Discussion The increase in the prevalence of excess body weight among women suggests an important epidemiological scenario related to parity, sexual hormones, and metabolic health, favoring the redistribution of body fat, with emphasis on visceral fat, in addition to negatively influencing serum concentrations of micronutrients, such as vitamin A [2]. Obesity is not only an excessive accumulation of adipose tissue but may also be accompanied by low-grade inflammation with macrophage infiltration that forms a vicious cycle through a paracrine circle [27]. Hypertrophic adipocytes release excess saturated fatty acids and activate macrophages via the Toll-like receptor 4 signaling pathway. These macrophages secrete proinflammatory cytokines and react to hypertrophic adipocytes, inducing an inflammatory response by activating the nuclear factor kB pathway and promoting further release of free fatty acids, which aggravates obesity [28]. Some explanations can be proposed to account for the beneficial effect of carotenoids; carotenoids and their conversion products may affect the inflammatory and secretory profiles of adipose tissue by actively interacting with these pathways in adipocytes and adipose tissue macrophages [12]. In this context, the findings of the present study deserve attention as they demonstrate that excess body adiposity, especially visceral adiposity, may represent an important cause of VA depletion, even in the face of recommended intake. Furthermore, it reinforces that, in women with excess body adiposity, the nutritional needs of VA may be much higher than the current recommendations. This scenario is even worse if we consider the results of the most recent national survey conducted by the Family Budget Survey [16], carried out in urban and rural regions, that evaluated the adult food consumption of the Brazilian population. The survey demonstrated a very high prevalence of inadequate consumption of VA, affecting more than $80\%$ of the adult population. Women have an even higher percentage (around $81.2\%$) [16]. Low concentrations of β-carotene were observed when compared to retinol. Since carotenoids, especially β-carotene, are considered precursors of retinoids, this indicates that, when their concentrations are reduced, retinoid synthesis is also reduced, thereby affecting the body’s ability to prevent weight gain and obesity [29]. β-carotene inhibits adipogenesis through the production of β-apo-140-carotene and the suppression of peroxisome proliferator-activated receptor type alpha (PPARα), peroxisome proliferator-activated receptor type gamma (PPARγ), and retinoid X receptor (RXR), in addition to the production of all-trans retinoic acid [30]. This study showed that adipocyte differentiation followed by a change in all-trans retinoic acid signaling, in mature adipocytes, enabled all-trans retinoic acid to activate both retin-oic acid receptors and PPARβ/δ, increasing lipolysis and depleting lipid stores. Therefore, these results indicate that improvement of obesity and insulin resistance by retinoic acid is widely mediated by PPARβ/δ and is further enhanced by activation of all-trans retinoic acid receptors [31]. Studies suggest that pro-VA carotenoids, including β-carotene, assist in the control of energy homeostasis by modulating the production of leptin and inflammatory cytokines [32,33]. Furthermore, in an experimental study [30], β-carotene was associated, in a dose-dependent manner, with increased expression of brown adipocyte uncoupling protein 1 (UCP-1), that dissipates energy from ATP synthesis for heat generation. The use of different indices and parameters to evaluate body adiposity deserves to be highlighted, especially for those related to visceral fat, which, in excess, leads to an inflammatory process, the presence of insulin resistance (IR), and adiponectin dysregulation. This study also showed that serum concentrations of retinol and β-carotene were reduced as body adiposity, especially visceral adiposity, increased when evaluated by WC, WHtR, VAI, and LAP. This finding can be explained, in part, by the greater demand for nutrients with antioxidant action. Once the visceral adipose tissue (VAT) expands, it becomes inflamed due to the infiltration of macrophages into the hypertrophied adipocytes, leading to increased production of inflammatory cytokines, including interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), and interleukin-1β (IL-1β); the reduced production of anti-inflammatory cytokines, such as adiponectin [34], consequently increases the release of reactive oxygen species, giving rise to oxidative stress (OS) [35,36]. These indices have shown promise because they have a good ability to predict adverse health outcomes. The way fat deposits are distributed, particularly in VAT, is considered the major contributor to metabolic risk, and results in different metabolic impacts more strongly associated with OS, inflammatory profile, IR, and higher mortality in general; this is more often seen in women [37,38,39,40,41]. Our findings are in line with the studies developed by Bento et al. [ 2018] [14] and Kabat et al. [ 2016] [42] on the inverse relationship between body adiposity and serum concentrations of retinol and β-carotene. However, the use of different indices of body adiposity measurement demonstrated that body fat distribution is a key issue regarding the impact of adiposity on serum concentrations of VA. Given the results, it is important to emphasize the role of VA metabolites in the activity of differentiation and functionality in adipose tissue, such as inhibition of cell differentiation in adipocyte culture. The action of retinoic acid in inducing UCP-1 expression and the consequent activation of thermogenesis in brown adipose tissue and the browning of white adipose tissue were well established [12]. In addition, retinoic acid is also associated with increased lipolysis in white adipose tissue via PPAR-y, decreased RXR expression, improved oxidative metabolism [32], increased lipolysis in adipocytes, decreased leptin and resistin expression, activation of thermogenesis by UCPs, and reduced cell differentiation of pre-adipocytes into adipocytes, thus reducing lipid storage capacity in adipocytes due to the action of the enzyme monooxygenase-1 β, β-carotene [43]. In this scenario, increased body adiposity can trigger VA depletion via the increased OS, inflammation, and, in some cases, in adipocyte sequestration [30]. Additionally, a deficiency of this vitamin can activate additional metabolic pathways associated with increased body adiposity. The way body fat was distributed, especially high visceral adiposity, showed an association with compromised nutritional status regardless of the recommended dietary intake [44]. One limitation of this study is related to the assessment of food intake, the same limitations found in most studies based on self-reported recall, in particular the underreporting of intake. However, we see no reason to believe that underreporting could have been different among the groups we evaluated. The findings of the present study reinforce the inclusion of WC, WHtR, VAI, and LAP as useful and complementary tools for the assessment of body adiposity since they provide additional information about the distribution of body fat and the relationship of this distribution with the nutritional status of VA. We suggest the development of further studies with the inclusion of inflammatory markers, aiming at greater detailing about the relationship between VA and body fatness. Such information may support the development of control and prevention strategies for two major public health problems: obesity and VAD. ## 5. Conclusions This study demonstrated an association between inadequate serum concentrations of retinol and β-carotene with excess body adiposity in women, even with the recommended intake of this vitamin. It was possible to observe the impaired nutritional status of VA by assessing body adiposity through the different indices and parameters used; these showed that the site of fat distribution is an impact factor on the serum concentrations of this vitamin. Considering the importance of VA in the metabolic regulation of adipose tissue, it is important to monitor its serum concentrations as well as to assess body fat distribution by including body adiposity assessment parameters such as WC, WHtR, VAI, and LAP. ## References 1. Obesity and Overweight. Revised 18 February 2018 2. Chooi Y.C., Ding C., Magkos F.. The epidemiology of obesity. Metabolism (2019.0) 92 6-10. DOI: 10.1016/j.metabol.2018.09.005 3. González-Muniesa P., Mártinez-González M.-A., Hu F.B., Després J.-P., Matsuzawa Y., Loos R.J.F., Moreno L.A., Bray G.A., Martinez J.A.. Obesity. Nat. Rev. Dis. Prim. (2017.0) 3 17034. DOI: 10.1038/nrdp.2017.34 4. Singh G.K., Di Bari J.N.. Marked Disparities in Pre-Pregnancy Obesity and Overweight Prevalence among US Women by Race/Ethnicity, Nativity/Immigrant Status, and Sociodemographic Characteristics, 2012–2014. J. Obes. (2019.0) 2019 2419263. DOI: 10.1155/2019/2419263 5. Kanter R., Caballero B.. Global Gender Disparities in Obesity: A Review. Adv Nutr. (2012.0) 3 491-498. DOI: 10.3945/an.112.002063 6. 6. Instituto Brasileiro de Geografia e Estatística-IBGE Pesquisa Nacional de Saúde 2019 Atenção Primária à Saúde e Informações AntropométricasMinistério da Saúde do BrazilBrazilia, Brazil2020. Pesquisa Nacional de Saúde 2019 Atenção Primária à Saúde e Informações Antropométricas (2020.0) 7. Yao N., Yan S., Guo Y., Wang H., Li X., Wang L., Hu W., Li B., Cui W.. The association between carotenoids and subjects with overweight or obesity: A systematic review and meta-analysis. Food Funct. (2021.0) 12 4768-4782. DOI: 10.1039/D1FO00004G 8. Gurunathan U., Myles P.S.. Limitations of body mass index as an obesity measure of perioperative risk. BJA Br. J. Anaesth. (2016.0) 116 319-321. DOI: 10.1093/bja/aev541 9. Herz C.T., Kiefer F.W.. The transcriptional role of vitamin A and the retinoid axis in brown fat function. Front. Endocrinol. (2020.0) 11 608. DOI: 10.3389/fendo.2020.00608 10. Blaner W.S.. Vitamin A signaling and homeostasis in obesity, diabetes, and metabolic disorders. Pharmacol. Ther. (2019.0) 197 153-178. DOI: 10.1016/j.pharmthera.2019.01.006 11. Godala M.M., Materek-Kuśmierkiewicz I., Moczulski D., Rutkowski M., Szatko F., Gaszyńska E., Tokarski S., Kowalski J.. The risk of plasma vitamin A, C, E and D deficiency in patients with metabolic syndrome: A case-control study. Adv. Clin. Exp. Med. (2017.0) 26 581-586. DOI: 10.17219/acem/62453 12. Bonet M.L., Canas J.A., Ribot J., Palou A., Stange C.. Carotenoids in Adipose Tissue Biology and Obesity. Carotenoids in Nature. Subcellular Biochemistry (2016.0) Volume 79. DOI: 10.1007/978-3-319-39126-7_15 13. Bento C., Matos A., Cordeiro A., Ramalho A.. Serum concentration of vitamin A and its relationship with body adiposity, oxidative stress, and cardiovascular risk in women with recommended dietary intake of vitamin A. Nutr. Hosp. (2020.0) 37 1135-1142. DOI: 10.20960/nh.03129 14. Bento C., Matos A.C., Cordeiro A., Ramalho A.. Vitamin A deficiency is associated with body mass index and body adiposity in women with the recommended intake of vitamin A. Nutr. Hosp. (2018.0) 35 1072-1078. DOI: 10.20960/nh.1630 15. Araujo M.C., Verly Junior E., Junger W.L., Sichieri R.. Independent associations of income and education with nutrient intakes in Brazilian adults: 2008–2009 National Dietary Survey. Public Health Nutr. (2014.0) 17 2740-2752. DOI: 10.1017/S1368980013003005 16. 16. Instituto Brasileiro de Geografia e Estatística (IBGE) Pesquisa de Orçamentos Familiares 2017–2018: Análise do Consumo Alimentar Pessoal no Brasil/IBGE, Coordenação deTrabalho e RendimentoIBGERio de Janeiro, Brazil2020. Pesquisa de Orçamentos Familiares 2017–2018: Análise do Consumo Alimentar Pessoal no Brasil/IBGE, Coordenação deTrabalho e Rendimento (2020.0) 17. Vitamin A. Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc (2001.0) 82-161 18. Tabela Brasileira de Composição de Alimentos (TBCA). Universidade de São Paulo (USP). Food Research Center (FoRC). Versão 7.2. São Paulo. (2022.0) 19. Pinheiro A.B.V., Lacerda E.M.D.A., Benzecry E.H., Gomes M.C.D.S., Costa V.M.D.. Tabela Para Avaliação de Consumo Alimentar em Medidas Caseiras (2001.0) Volume 5 352 20. 20. International Vitamin A Consultative Group (IVACG) The Annecy Accords to Assess and Control Vitamin A Deficiency: Summary of Recommendations and ClarificationsIVACGWashington, DC, USA2003. The Annecy Accords to Assess and Control Vitamin A Deficiency: Summary of Recommendations and Clarifications (2003.0) 21. Zeng Q., He Y., Dong S., Zhao X., Chen Z., Song Z., Chang G., Yang F., Wang Y.. Optimal cut-off values of BMI, waist circumference and waist: Height ratio for defining obesity in Chinese adults. Br. J. Nutr. (2014.0) 112 1735-1744. DOI: 10.1017/S0007114514002657 22. Freitas R.S., Fonseca M.D.J.M.D., Schmidt M.I., Molina M.D.C.B., Almeida M.D.C.C.D.. Fenótipo cintura hipertrigliceridêmica: Fatores associados e comparação com outros indicadores de risco cardiovascular e metabólico no ELSA-Brasil. Cad. De Saúde Pública (2018.0) 34 e00067617. DOI: 10.1590/0102-311x00067617 23. Bergman R.N., Stefanovski D., Buchanan T.A., Sumner A.E., Reynolds J.C., Sebring N.G., Xiang A., Watanabe R.M.. A Better Index of Body Adiposity. Obesity (2011.0) 19 1083-1089. DOI: 10.1038/oby.2011.38 24. Amato M.C., Giordano C., Galia M., Criscimanna A., Vitabile S., Midiri M., Galluzzo A.. Visceral Adiposity Index: A reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes Care (2010.0) 33 920-922. DOI: 10.2337/dc09-1825 25. Vieira J.N., Braz M.A.D., Gomes F.O., Silva P.R.D., Santos O.T.M., Rocha I.M.G.D., Sousa I.M., Fayh A.P.T.. Cardiovascular risk assessment using the lipid accumulation product index among primary healthcare users: A cross-sectional study. Sao Paulo Med. J. (2019.0) 137 126-131. DOI: 10.1590/1516-3180.2018.0293240119 26. Mason J.. Vitamins, Trace Minerals, and other Micronutrients in Goldman’s Cecil Medicine (2012.0). DOI: 10.1016/C2009-0-42832-0 27. Wang C., Ling C.-W., Ding D., Li Y.-H., Cao W.-T., Tang X.-Y., Chen Y.-M.. Associations of Serum Carotenoids with DXA-Derived Body Fat and Fat Distribution in Chinese Adults: A Prospective Study. J. Acad. Nutr. (2020.0) 120 985-1001. DOI: 10.1016/j.jand.2019.12.010 28. Ide J., Gagnon A., Molgat A.S., Landry A., Foster C., Sorisky A.. Macrophage-conditioned medium inhibits the activation of cyclindependent kinase 2 by adipogenic inducers in 3T3-L1 preadipocytes. J. Cell Physiol. (2011.0) 226 2297-2306. DOI: 10.1002/jcp.22566 29. Saini R.K., Prasad P., Lokesh V., Shang X., Shin J., Keum Y.-S., Lee J.-H.. Carotenoids: Dietary Sources, Extraction, Encapsulation, Bioavailability, and Health Benefits—A Review of Recent Advancements. Antioxidants (2022.0) 11. DOI: 10.3390/antiox11040795 30. Mounien L., Tourniaire F., Landrier J.-F.. Anti-Obesity Effect of Carotenoids: Direct Impact on Adipose Tissue and Adipose Tissue-Driven Indirect Effects. Nutrients (2019.0) 11. DOI: 10.3390/nu11071562 31. Da Cruz B., Cardozo L.F.M.D.F., Magliano D.C., Stockler-Pinto M.B.. Nutritional strategies to modulate inflammation pathways via regulation of peroxisome proliferator-activated receptor β/δ. Nutr. Rev. (2020.0) 78 207-214. DOI: 10.1093/nutrit/nuz058 32. Bonet M.L., Ribot J., Galmés S., Serra F., Palou A.. Carotenoids and carotenoid conversion products in adipose tissue biology and obesity: Pre-clinical and human studies. Biochim. Biophys Acta Mol. Cell Biol. Lipids (2020.0) 1865 158676. DOI: 10.1016/j.bbalip.2020.158676 33. Checa J., Aran J.M.. Reactive Oxygen Species: Drivers of Physiological and Pathological Processes. J. Inflamm. (2020.0) 13 1057-1073. DOI: 10.2147/JIR.S275595 34. Neeland I.J., Ross R., Després J.-P., Matsuzawa Y., Yamashita S., Shai I., Seidell J., Magni P., Santos R.D., Arsenault B.. Visceral and ectopic fat, atherosclerosis, and cardiometabolic disease: A position statement. Lancet Diabetes Endocrinol. (2019.0) 7 715-725. DOI: 10.1016/S2213-8587(19)30084-1 35. Ellulu M.S., Patimah I., Khaza’ai H., Rahmat A., Abed Y.. Obesity and inflammation: The linking mechanism and the complications. Arch. Med. Sci. (2017.0) 13 851-863. DOI: 10.5114/aoms.2016.58928 36. Haaker M.W., Vaandrager A.B., Helms J.B.. Retinoids in health and disease: A role for hepatic stellate cells in affecting retinoid levels. Biochim. Biophys Acta Mol. Cell Biol. Lipids (2020.0) 1865 158674. DOI: 10.1016/j.bbalip.2020.158674 37. Elffers T.W., De Mutsert R., Lamb H.J., De Roos A., Van Dijk K.W., Rosendaal F.R., Jukema J.W., Trompet S.. Body fat distribution, in particular visceral fat, is associated with cardiometabolic risk factors in obese women. PLoS ONE (2017.0) 12. DOI: 10.1371/journal.pone.0185403 38. Stefan N.. Causes, consequences, and treatment of metabolically unhealthy fat distribution. Lancet Diabetes Endocrinol. (2020.0) 8 616-627. DOI: 10.1016/S2213-8587(20)30110-8 39. Schorr M., Dichtel L.E., Gerweck A.V., Valera R.D., Torriani M., Miller K.K., Bredella M.A.. Sex differences in body composition and association with cardiometabolic risk. Biol. Sex Differ. (2018.0) 9 28. DOI: 10.1186/s13293-018-0189-3 40. De Mutsert R., Gast K., Widya R., de Koning E., Jazet I., Lamb H., le Cessie S., de Roos A., Smit J., Rosendaal F.. Associations of abdominal subcutaneous and visceral fat with insulin resistance and secretion differ between men and women: The Netherlands epidemiology of obesity study. Metab Syndr. Relat. Disord. (2018.0) 16 54-63. DOI: 10.1089/met.2017.0128 41. Ahmed B., Sultana R., Greene M.W.. Adipose tissue and insulin resistance in obese. Biomed Pharm. (2021.0) 137 111315. DOI: 10.1016/j.biopha.2021.111315 42. Kabat G.C., Heo M., Ochs-Balcom H.M., LeBoff M.S., Mossavar-Rahmani Y., Adams-Campbell L.L., Nassir R., Ard J., Zaslavsky O., Rohan T.E.. Longitudinal association of measures of adiposity with serum antioxidant concentrations in postmenopausal women. Eur. J. Clin. Nutr. (2016.0) 70 47-53. DOI: 10.1038/ejcn.2015.74 43. Thomas-Valdés S., das Graças V., Tostes M., Anunciação P.C., da Silva B.P., Sant’Ana H.M.P.. Association between vitamin deficiency and metabolic disorders related to obesity. Crit. Rev. Food Sci. Nutr. (2017.0) 57 3332-3343. DOI: 10.1080/10408398.2015.1117413 44. Ribamar A., Cruz S., Bento C., Ramalho A.. Visceral and body adiposity are negatively associated with vitamin A nutritional status independently of Body Mass Index and recommended intake of vitamin A in Brazilian Women. J. Nutr. Biochem. (2022.0) 109 109-120. DOI: 10.1016/j.jnutbio.2022.109120
--- title: 'Association of Clinical Aspects and Genetic Variants with the Severity of Cisplatin-Induced Ototoxicity in Head and Neck Squamous Cell Carcinoma: A Prospective Cohort Study' authors: - Ligia Traldi Macedo - Ericka Francislaine Dias Costa - Bruna Fernandes Carvalho - Gustavo Jacob Lourenço - Luciane Calonga - Arthur Menino Castilho - Carlos Takahiro Chone - Carmen Silvia Passos Lima journal: Cancers year: 2023 pmcid: PMC10046479 doi: 10.3390/cancers15061759 license: CC BY 4.0 --- # Association of Clinical Aspects and Genetic Variants with the Severity of Cisplatin-Induced Ototoxicity in Head and Neck Squamous Cell Carcinoma: A Prospective Cohort Study ## Abstract ### Simple Summary Cisplatin is recognized as the standard agent for head and neck squamous cell carcinoma therapy, despite the relevant risk of permanent hearing damage. The aim of this study was to evaluate the possible associations of the clinicopathological features and inherited genotypes encoding cisplatin metabolism in eighty-nine patients undergoing chemoradiation with the risk of hearing loss. We were able to confirm race, body mass index, and cumulative cisplatin dose as independent clinical risk factors. Patients with specific isolated and combined genotypes encoding cisplatin efflux (GSTM1, GSTP1 c.313A>G), DNA repair (XPC c.2815A>C, XPD c.934G>A, EXO1 c.1762G>A, MSH3 c.3133A>G), and apoptosis-related proteins (FASL c.-844A>T, P53 c.215G>C) presented up to 32.22 higher odds of moderate or severe ototoxicity. These findings reinforce the importance of inherited nucleotide variants involved in cisplatin metabolism as candidate variables for predictive models of adverse events. ### Abstract Background: Cisplatin (CDDP) is a major ototoxic chemotherapy agent for head and neck squamous cell carcinoma (HNSCC) treatment. Clinicopathological features and genotypes encode different stages of CDDP metabolism, as their coexistence may influence the prevalence and severity of hearing loss. Methods: HNSCC patients under CDDP chemoradiation were prospectively provided with baseline and post-treatment audiometry. Clinicopathological features and genetic variants encoding glutathione S-transferases (GSTT1, GSTM1, GSTP1), nucleotide excision repair (XPC, XPD, XPF, ERCC1), mismatch repair (MLH1, MSH2, MSH3, EXO1), and apoptosis (P53, CASP8, CASP9, CASP3, FAS, FASL)-related proteins were analyzed regarding ototoxicity. Results: Eighty-nine patients were included, with a cumulative CDDP dose of 260 mg/m2. Moderate/severe ototoxicity occurred in 26 ($29\%$) patients, particularly related to hearing loss at frequencies over 3000 Hertz. Race, body-mass index, and cumulative CDDP were independent risk factors. Patients with specific isolated and combined genotypes of GSTM1, GSTP1 c.313A>G, XPC c.2815A>C, XPD c.934G>A, EXO1 c.1762G>A, MSH3 c.3133A>G, FASL c.-844A>T, and P53 c.215G>C SNVs had up to 32.22 higher odds of presenting moderate/severe ototoxicity. Conclusions: Our data present, for the first time, the association of combined inherited nucleotide variants involved in CDDP efflux, DNA repair, and apoptosis with ototoxicity, which could be potential predictors in future clinical and genomic models. ## 1. Introduction Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide, with 878,348 new cases and 444,347 deaths estimated in 2020 [1,2]. Approximately $75\%$ of patients with HNSCC present locally advanced disease at diagnosis, and the standard therapy for most cases involves chemoradiation or the induction of multi-agent chemotherapy, in which cisplatin (CDDP) is usually included [3,4]. Alternative treatments, such as carboplatin or cetuximab, were studied in the context of chemoradiation, though their equivalence regarding efficacy has yet to be validated by randomized trials [5]. Additionally, in patients with treatment-naïve metastatic disease or platinum-sensitive relapse, CDDP-based regimens are commonly used in clinical practice, with benefits in progression-free survival (PFS) and overall survival (OS) [6,7]. Nonetheless, CDDP is related to significant adverse events, such as nausea, vomiting, nephrotoxicity, hypersensitivity reactions, and ototoxicity [3,8,9]. Among these, hearing impairment is a current concern since there are, to date, no effective otoprotective measures, resulting in potentially permanent and quality-of-life-limiting damage [10,11]. Every year, one in five patients submitted to CDDP-based chemotherapy will suffer severe to profound hearing loss [10,12,13]. Regarding chemoradiation for HNSCC, major losses are described in higher frequencies, with reported pure-tone median threshold increases ranging from 9.52 to 25 decibels (dB) at 4 kilohertz (kHz) and 18.57 to 27.14 dB at 8 kHz [14,15]. This event has a major negative impact on the quality of life [16] and requires essential care regarding dosage management and the duration of therapy [17]. Despite the association of cumulative CDDP dose, history of noise exposure, and smoking as independent risk factors, the prevalence and intensity of hearing impairment are remarkably heterogeneous among patients with similar characteristics and regimens [18]. This finding indicates the involvement of unknown risk factors, with single-nucleotide variants (SNVs), on genes encoding proteins related to CDDP metabolism, being potential candidates for this risk [19,20]. Numerous proteins act in the mechanisms of CDDP cellular detoxification, as well as in the pathways of damage repair and apoptosis [21,22] (Figure 1). The detoxification of CDDP occurs mainly through its conjugation with glutathione, encoded by the Mu1 (GSTM1), Theta1 (GSTT1), and Pi1 (GSTP1) genes [23], in which the lack of functional proteins involved in this cascade may contribute to intracellular CDDP accumulation and cytotoxic effects [24]. The cytotoxic activity of CDDP is also attributed to its DNA binding, leading to the activation of repair mechanisms. The DNA lesion induced by CDDP can be removed through the nucleotide excision repair (NER) pathway [25], mediated by the xeroderma pigmentosum (XPC, XPD, and XPF) [26,27] and excision repair cross-complementation group 1 (ERCC1) genes [28], as well as by the mismatch repair (MMR) pathway, mediated through proteins encoded by MutL homolog 1 (MLH1) [29], MutS homolog 2 (MSH2) [30], MutS homolog 3 (MSH3), and exonuclease 1 (EXO1) genes [29]. If the repair is ineffective, apoptosis is mediated by proteins encoded by P53, Caspase 8 (CASP8), CASP9, CASP3, Fas cell surface death receptor (FAS), and Fas ligand (FASL) tumor necrosis factors [21,31]. Defects in these pathways may promote increased DNA damage and/or apoptosis, with greater potential for toxicity [32]. Genome-wide studies have described SNVs in acylphosphatase 2 (ACYP2), involved in calcium homeostasis [33,34,35] and Mendelian deafness WFS1 genes [20,33,36,37], as predictors of CDDP-induced ototoxicity. Genes encoding thiopurine S- (TPMT) and cathecol-O methyltransferases (COMT) have also been described as potential risk factors [35]. In CDDP-treated patients, GSTM1, GSTT1 [18,38,39,40,41,42], and GSTP1 c.313A>G [38,39,41,43] were seen in pediatric solid or adult testicular tumors with controversial results in ototoxicity, while XPC c.2815A>C SNV influenced ototoxicity in osteosarcoma patients [44]. To our knowledge, the only cohort that evaluated SNVs in genes of distinct pathways of CDDP metabolism, damage repair, and apoptosis (GSTM1, GSTT1, GSTP1 c.313A>G, XPC c.2815A>C, XPD c.934G>A, XPD c.2251A>C, XPF c.2505T>C, ERCC1 c.354C>T, MLH1 c.-93G>A, MSH2 c.211 +9G>C, MSH3 c.3133A>G, EXO1 c.1762G>A, P53 c.215G>C, FAS c.-671A>G, FAS c.-1378G>A, FASL c.-844C>T, CASP3 c.-1191A>G, and CASP3 c.-182-247G>T) in the ototoxicity of HNSCC treated with CDDP chemoradiation was previously conducted by our group, and the functional roles of each SNV described in the literature are presented in Table A1. We found that GSTT1 [45], EXO1 [19], XPC [46], and FASL [47] SNVs altered the occurrence of all-grade ototoxicity. Since there is scarce information regarding pure tone and audiometric speech changes in patients under CDDP chemoradiation, considering that moderate/severe ototoxicity influences quality of life and the fact that patients may inherit defects in more than one pathway, we conducted a descriptive and pharmacogenetic study focusing isolated factors related to CDDP metabolism, aiming to contribute to the prompt recognition of patients at high risk of ototoxicity before treatment initiation and thus enabling treatment modifications. ## 2.1. Study Population This cohort prospectively enrolled HNSCC patients who were eligible for treatment with definitive chemoradiation at the Clinical Oncology Service of the University of Campinas, Brazil, between June 2011 and February 2014. Eastern Cooperative Oncology Group (ECOG) performance status of equal to or less than 1 [48], creatinine clearance greater than 45 mL/min, and the absence of baseline moderate or severe hearing impairment were required. Patients who were not candidates for treatment with CDDP or who were under induction, adjuvant, or palliative therapy were excluded. Patients received high-dose CDDP (starting dose of 80–100 mg/m2 on days 1, 22, and 43) [49] associated with RT (35 sessions; planned total radiation dose of 70 Gray—Gy). All patients received anti-emetic prophylaxis with intravenous ondansetron and dexamethasone pre-infusion, in addition to oral dexamethasone and metoclopramide, for the following three days. Mannitol and hydration with saline solution, potassium chloride, and magnesium sulfate were administered, as reported [46]. Dose delays and reductions were applied in toxicity events with grades equal to or greater than 3, according to the National Cancer Institute criteria for adverse events (NCI CTCAE) [50]. Patients were followed from recruitment to 30 days after treatment completion. The study was approved by the local institutional review board (Protocol $\frac{274}{2011}$ and 62870722.1.0000.5404), and all patients enrolled in the study agreed to participate and declared consent in accordance with the Declaration of Helsinki. The results of this study were reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [51]. ## 2.2. Clinical Data Data related to age, gender, race, history of tobacco and alcohol use [52], ECOG status [48], body mass index (BMI) [53], and presence of diabetes [54] or systemic hypertension [55] as comorbidities of interest were collected. Regarding disease characteristics, primary tumor location, tumor side, and histological grade were also computed. Diagnosis and tumor staging followed the American Joint Committee on Cancer criteria [56]. Data related to cumulative CDDP dose, radiotherapy (RT) technique (2D or 3D), and final total dose in Gy, including total doses from supraclavicular fossa, cervico-facial, cervico-posterior, and boost, were also registered for analysis. ## 2.3. Hearing Assessment Patients were submitted to otoscopic examination before any audiometric measurements. If there were identifiable diseases of the external acoustic meatus, tympanic membranes, middle ears, or other conditions that could interfere with the audiological evaluation, patients received treatment and were followed up until resolution. Audiometric evaluations were performed on two occasions, before treatment initiation and up to 30 days after therapy completion in an acoustic booth previously calibrated to meet the specifications of internal noise levels allowed according to the International Organization for Standardization ((ISO) 8253-1:2010 criteria, using the Interacoustics audiometer model AC 30 (Interacoustics A/S, Middlefart, Denmark). ## 2.3.1. Pure Tone Audiometry Pure tone audiometry was conducted in air and bone conduction for both the left and right sides. For the air conduction assessment, the tonal auditory thresholds were measured at sound frequencies 0.25, 0.50, 1, 2, 3, 4, 6, and 8 kHz, with earphones model TDH 39, applying the descending–ascending technique. In each test, the smallest sound stimulus perceived by the patient in at least $50\%$ of the presentations was considered. Bone conduction evaluation was performed in a similar descending–ascending manner, registering minimum dB thresholds at frequencies 0.25, 0.50, 1, 2, 3, and 4 kHz through a conduction receiver bow fixed on the mastoid (Interacoustics A/S, Middlefart, Denmark). The corresponding hearing thresholds for each frequency in both ears were collected, considering that the normal expected range was lower than 15 dB [57]. ## 2.3.2. Speech Audiometry Speech audiometry was performed when applicable, assessing the speech recognition threshold (SRT) in dB for the repetition of $50\%$ disyllabic words for left and right ears. The speech discrimination score (SDS), calculating the percentage of syllables repeated correctly, was also registered for each side when possible [58]. SRT is normally within 10 dB of pure tone average thresholds, while the normal range of SDS is $92\%$ to $100\%$ [59]. Patients unable to speak owing to disease-related limitations or other causes had their exam halted and reasons noted. ## 2.4.1. Global Burden of Disease (GBD) Hearing Loss Classification The GBD Hearing Loss Classification was performed, assessing the ISO threshold average for 0.5, 1, 2, and 4 kHz in dB, as recommended by the World Health Organization [60], pre- and post-treatment in air and bone conduction assessments [61]. Patients were categorized according to the criteria of unilateral (<20 dB in the better and >35 dB in the worst ears, respectively), mild (20 to 34 dB in the better ear), moderate (35–49 dB in the better ear), moderately severe (50–64 dB in the better ear), severe (65–79 dB in the better ear), and profound (80–94 dB in the better ear) hearing loss. ## 2.4.2. National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) Hearing loss in the right and left ears were also classified based on grades (G) 1 to 4, according to the NCI CTCAE v4.0 criteria [50] following the monitoring of at least 1, 2, 3, 4, 6, and 8 kHz audiogram, as follows: G 1, “threshold shift of 15 to 25 dB averaged at two contiguous test frequencies in at least one ear”; G 2, “threshold shift of >25 dB averaged at two contiguous test frequencies in at least one ear”; G 3, “threshold shift of >25 dB averaged at three contiguous test frequencies in at least one ear”; G 4, “decrease in hearing to profound bilateral loss (absolute threshold >80 dB hearing loss at 2 kHz and above)”. ## 2.5. Genetic Variants Analysis Genetic variants were selected for study based on the National Center for Biotechnology Information database, minor allele frequency greater than $10\%$, previous association with risk/outcome of solid tumors and/or CDDP metabolism, and the availability of financial resources (Figure A1). For genotyping, DNA samples from peripheral blood were collected, where the genotypes of the GSTM1 and GSTT1 variants [62] were obtained through the multiplex polymerase chain reaction (PCR) followed by digestion assays with enzymes of restriction. The additional genetic variants were evaluated by real-time PCR using TaqMan® SNP Genotyping Assays (Applied Biosystems®, Thermo Fisher Scientific Inc., Waltham, MA, USA), as follows: GSTP1 c.313A>G (rs1695) [63], XPC c.2815A>C (rs2228001) [64], XPD c.934G>A (rs1799793) [65], XPD c.2251A>C (rs13181) [65], XPF c.2505T>C (rs1799801) [66], ERCC1 c.354C>T (rs11615) [67], MLH1 c.-93G>A (rs1800734) [68], MSH2 c.211 +9G>C (rs2303426) [69], MSH3 c.3133A>G (rs26279) [70], EXO1 c.1762G>A (rs1047840) [71], P53 c.215G>C (rs1042522) [72], FAS c.-1378G>A (rs2234767) [73], FAS c.-671A>G (rs1800682) [74], FASL c.-844C>T (rs763110) [74], CASP3 c.-1191A>G (rs12108497) [75], and CASP3 c.-182-247G>T (rs4647601) [76]. Positive and negative controls were used in all genotyping reactions, and replications of $10\%$ randomly selected samples were also performed in independent experiments, with $100\%$ agreement. ## 2.6. Statistical Analysis Descriptive statistics were performed according to the variables under study, with mean values and standard deviation (SD) in normal distribution or median and interquartile ranges (IQR) when applicable. Pre- and post-treatment pure tone thresholds were described individually, as well as the averages of high-frequency minimum thresholds (considering 3, 4, 6, and 8 kHz) and ISO averages (0.5, 1, 2, and 4 kHz) in each ear. Wilcoxon’s signed-rank test for paired data was applied in the comparison of speech audiometry, pure tone averages, and frequency thresholds before and after chemoradiation, with the latter controlling for false discovery rates with the Benjamini–Hochberg correction in multiple testing [77]. Cochran’s Q test was used for the GBD classification of hearing loss before and after therapy. To assess the influence of clinicopathological aspects and genotypes related to high-frequency minimum threshold average changes from baseline, we performed multiple linear regression. Data were transformed into ranks. The significance level adopted for the study was $5\%$. The main endpoint of this study was the proportion of patients with NCI CTCAE hearing loss G equal to or greater than 3 during follow-up based on audiometry monitoring. Multiple logistic regression was used to obtain the odds ratio (ORs) adjusted for any specific discrepancies for each independent variable, considering a $95\%$ confidence interval (CI). Variables were selected using a conditional stepwise approach, permitting a p-value of under 0.10 in univariate regression. Post hoc power analyses (PA) were also conducted for associations, taking into consideration p-value and CI as the measures of statistical significance [78,79]. After multivariate analysis, results with p ≤ 0.05 were validated using bootstrap [80] to verify the stability of risk estimates and account for missing data (1000 replications). Isolated SNVs associated with the increase in the hearing thresholds or grade 3 ototoxicity and combined SNVs associated with an increase in hearing thresholds or grade 3 ototoxicity with PA >$70\%$ were selected for this study. All tests were performed using the Statistical Analysis System (SAS) for Windows, version 9.4 (SAS Institute Inc., 2002–2008, Cary, NC, USA) and Stata Statistical Software: Release 15 (Stata Corp LP, College Station, TX, USA). ## 3.1. Study Population In a median follow-up of 142 days, 152 patients were enrolled, of whom 89 were included in the analysis with the completion of baseline and post-treatment audiometry (Figure 2). The median age was 56 years, and most patients were male and white, with a high rate of tobacco and alcohol consumption. Median BMI was within the normally acceptable range, most presented an ECOG status of 0, and the proportion of diabetes and hypertension was 10 and $26.9\%$, respectively. Most primary tumors were in the oral cavity or oropharynx, evenly distributed between the right and left sides of the head and neck, well or moderately differentiated, and at advanced stages (III or IV). The median cumulative CDDP dose among patients was 260 mg/m2. Eighty-eight patients received 2D RT, with a total dose of 70 Gy. The clinicopathological aspects of patients enrolled in the study are further detailed in Table 1. ## 3.2.1. Pure Tone Audiometry Analyzing the median thresholds for each frequency upon baseline, we were able to observe a normal range below 2 kHz and a trend toward higher thresholds, starting from 3 kHz in both conduction modalities (Figure 3). After treatment, there was significant damage regarding higher frequencies over 2 kHz, which was more evident in air conduction analysis. Hearing thresholds for each frequency in pure tone audiometry are further detailed in Table 2. Following the ISO average criteria, a median increase of 5 dB ($p \leq 0.001$) on the right side and 6.25 dB ($p \leq 0.001$) on the left side was observed when comparing baseline to post-treatment assessments. For bone conduction, there was a median increase of 6.25 ($p \leq 0.001$) on the right side and 6.25 dB ($p \leq 0.001$) on the left side, respectively (Table 3). Regarding high-frequency average thresholds for pure tone air conduction audiometry, there was a median increase of 18 dB ($p \leq 0.001$) on the right and 19 dB ($p \leq 0.001$) on the left sides observed after exposure to CDDP and RT. ## 3.2.2. Speech Audiometry Data from speech audiometry were retrievable in 62 patients since 27 had limited speech capability (nine were submitted to tracheostomy and eighteen presented tumors in the oral cavity). The median baseline SRT was 10 and 15 dB in the right and left ears, respectively. Pre- and post-treatment median differences were null on both sides (Table 3). Regarding SDS, median baseline and post-treatment scores were $96\%$ and $92\%$, respectively, for both ears, with a decrease of $4\%$ on the right side. ## 3.2.3. GBD Classification for Hearing Loss Before chemoradiation, mild hearing loss was seen in about one-third and one-quarter of patients analyzed by air and bone conduction, respectively, and only three patients presented moderate hearing impairment in both assessments. After treatment, there was a significant increase in the proportion of mild and moderate hearing loss identified in air and bone conduction (chi2 20.16, $p \leq 0.001$ for air; chi2 18.24, $p \leq 0.001$ for bone conduction), although a severe degree of hearing loss was not observed throughout the study (Table 3). The unilateral loss was more evident in air conduction after treatment, although the same proportion was not observed in bone conduction analyses. All patients with unilateral damage after treatment had pharyngeal carcinoma located on the side of hearing loss and with changes at baseline. ## 3.2.4. Hearing Impairment in Monitoring Audiometry According to NCI CTCAE Criteria The proportion of any-grade hearing impairment by air conduction after chemoradiation was $76.4\%$ (68 out of 89 patients). The ototoxicity of G1 and G2 was observed in 23 ($25.8\%$) and 19 ($21.3\%$) patients, respectively. G3 or moderate/severe ototoxicity occurred in 26 ($29.3\%$) participants, and G4 was not identified in this study. ## 3.3.1. Average of Minimum Threshold for Pure Tone Air Conduction Audiometry at High Frequencies (3, 4, 6, and 8 kHz) In univariate analysis, gender and cumulative CDDP dose were associated with hearing loss in the right ear, while BMI was associated with hearing loss in both ears. Only cumulative CDDP dose was associated with hearing loss in the right ear in multivariate analysis (regression coefficient = 0.08, $$p \leq 0.02$$), where the higher the dose of CDDP, the greater the hearing impairment (Supplemental Table S1). When SNVs were analyzed individually, it was observed that patients with XPC c.2815AA genotype presented higher average threshold increases after CDDP chemoradiation than those with XPC c.2815AC or CC genotypes (23.8 versus 17.5 dB in the right ear; 27.5 versus 16.3 dB in left ear), as represented in Table 4. Higher average threshold increases were also seen after treatment in patients with combined genotypes GSTM1 null plus EXO1 c.1762GA or AA (21.3 versus 5.0 dB in the right ear; 22.5 versus 8.8 dB in the left ear) and with GSTP1 c.313AG or GG plus XPC c.2815AA (30.0 versus 17.5 dB in the right ear; 38.8 versus 16.3 dB in the left ear) in comparison to other related variants. The analyses of isolated and combined SNVs with biological significance with hearing loss at high frequencies are presented in Supplemental Table S2 and Table S3, respectively. ## 3.3.2. Hearing Impairment in Monitoring Audiometry According to NCI CTCAE Criteria Race and BMI were significantly associated with the risk of G3 ototoxicity in univariate and multivariate analyses, but potential clinical risk factors such as age, gender, diabetes, hypertension, smoking, alcohol consumption, tumor stage, tumor side, and CDDP cumulative dose did not alter the risk of ototoxicity in univariate analysis (Supplemental Table S4). The occurrence of moderate/severe ototoxicity was more common in non-white than in white patients ($66.7\%$ versus $25.0\%$, respectively), with OR = 5.43 ($95\%$ CI: 1.21–24.27, $$p \leq 0.02$$) in multivariate analysis. BMI was also a potential predictor, as participants with grade 3 hearing impairment presented lower median BMI (17.8 versus 19.7), with the OR = 0.82 higher for every decrease in BMI ($95\%$ CI: 0.72–0.98, $$p \leq 0.03$$) in multivariate analysis. When analyzed individually (Table 5), two SNVs were identified as independent factors for this outcome. Patients with GSTP1 c.313AG or GG genotypes had about 4.20 higher odds of having grade 3 or greater ototoxicity. Moreover, XPC c.2815AA genotype was associated with greater odds of severe hearing impairment, with a reported OR of 3.13 ($$p \leq 0.01$$) in the multivariate regression model. In associations of SNVs, it was observed that patients with GSTM1 null plus the XPC c.2815AA genotype had 8.19 greater odds of having moderate/severe hearing impairment ($$p \leq 0.02$$, PA = $99\%$). GSTP1 c.313AG or GG genotypes plus XPC c.2815AA, XPD c.934AA and EXO1 c.1762AA had ORs of 32.22 ($$p \leq 0.004$$, PA = $97\%$), 19.44 ($$p \leq 0.02$$, PA = $92\%$), and 12.08 ($$p \leq 0.01$$, PA = $81\%$), respectively. In addition, we observed relevant associations amongst DNA repair and apoptosis-related SNVs in patients with XPC c.2815AA genotype plus MSH3 c.3133A>G and FASL c.-844CC, where individuals with the respective profile had OR 17.09 ($$p \leq 0.009$$, PA = $88\%$) and 22.29 ($$p \leq 0.01$$, PA = $82\%$). Finally, patients with the combined genotypes EXO1 c.1792GA or AA and P53 c.215 CC had OR 20.97 ($$p \leq 0.02$$, PA = $85\%$). Further details of other SNVs and their combinations are summarized in Supplemental Table S5 and Table S6, respectively. ## 4. Discussion In this clinical and pharmacogenetic cohort, it was possible to reaffirm the clinical relevance of hearing loss induced by CDDP. CDDP induces ototoxicity through the promotion of oxidative stress and inflammation in the cochlea, with the increased generation of reactive oxygen species (ROS) [81]. The long-term accumulation of CDDP in the cochlear endolymph was also described through plasma mass spectrometry in preclinical models, justifying the potential for permanent damage [82]. Firstly, substantial hearing loss before treatment was observed in our cohort; high-frequency minimum thresholds were higher at baseline, ranging from 35 to 40 dB over 4 kHz. This may be attributable to the high proportion of smokers in our sample since smoking is a reported risk factor for loss at high frequencies [18,83,84]. A history of noise exposure, not assessed in this cohort, could also explain this finding as well as uneven losses in left and right ears not associated with the tumor side [85,86,87]. We were able to observe a meaningful change after CDDP exposure in regard to minimum hearing thresholds, particularly in higher frequencies in univariate analysis, as suggested by previous studies [15], with limitations involving higher pitch sounds. Caballero and colleagues described similar findings in a cohort of 103 patients, with significantly meaningful changes after CDDP exposure (median change of 9.5 dB in the right and 18.75 dB in the left ears for 4 kHz; 18.6 dB in the right and 28.7 dB in the left, for 8 kHz). The limitations to quality of life related to hearing loss from CDDP have already been reported in a recent systematic review from Pearson and colleagues [16]. Regarding additional clinical factors, cumulative CDDP dose was observed as a risk factor for greater change in high-frequency averages (3, 4, 6, and 8 kHz) for the right ear, which prompted the inclusion of this variable in multivariate analysis for both sides. When accounting for the 0.25 to 4 kHz interval, there was also a significant relative increase in mild and moderate hearing loss after CDDP in our analysis, following GBD classification. The percentage of $64.1\%$ with a threshold ≥ 20 dB after treatment is markedly superior to the overall prevalence reported in the literature for the general population ($19.3\%$) [88], pointing to the cytotoxic effects of CDDP on hearing impairment. To our knowledge, this is the first study to report this classification before and after CDDP in patients diagnosed with HNSCC [89]. Unilateral hearing damage was observed in four patients ($4.5\%$) after therapy under pure tone audiometry air conduction, from which two ($2.2\%$) had reported losses in both conduction modalities (air and bone). It is worth commenting that all patients with unilateral damage had pharyngeal carcinoma located on the side of hearing loss, and most had changes at baseline. Even though the RT technique and CDDP dose did not differ amongst patients, the location of the tumor in relation to the inner ear could have influenced this finding. On the other hand, median outcomes from speech audiometry (SRT and SDS) were practically unchanged after platinum exposure. One possible explanation for this may be related to the fact that human speech usually ranges from 0.25 to 4 kHz [90], while CDDP-related hearing loss involves more relevant changes beyond 3 kHz. In an isolated acoustic environment, frequencies related to speech may be unaltered, though it is possible to expect a greater extent of limitation in terms of communication with background noise, which was not assessed in this cohort. The largest study analyzing speech audiometry after CDDP exposure was performed by Shahbazi and colleagues [91], evaluating the prevalence of speech recognition disability, defined as SRT greater than 15 dB, in testicular cancer survivors. In 1347 patients, speech recognition disability was identified in $10.4\%$, and the association of the cumulative CDDP dose could also not be confirmed. Those findings are distinct from our analysis, where $51.6\%$ could be classified as speech-disabled before therapy and $60.7\%$ after therapy. The study populations are markedly different since the Platinum Study [91] included younger patients not bearing primary tumors in the head and neck and without risk factors such as tobacco and alcohol consumption. There are, to date, only scarce amounts of the literature data on speech audiometry for HNSCC, thus limiting further comparisons. When considering the NCI CTCAE criteria for the classification of hearing loss, the proportion of $29.5\%$ moderate/severe ototoxicity was marginally higher than previously reported literature data, ranging from 20 to $25\%$ in adults [10,37]. Except for race and BMI, other clinical variables such as age, sex, tumor location, and staging could not be identified as prognostic factors in this analysis, and although cumulative CDDP is recognized as a risk factor for hearing damage [37], this association could not be observed in the present data for this outcome specifically. Some recent studies have suggested the presence of emotional stress as a possible risk factor for enhanced tumorigenesis and neurotoxicity induced by chemotherapy in general [92]. A cross-sectional analysis of 623 cancer survivors described a higher association of tinnitus ($$p \leq 0.029$$) and hearing loss ($$p \leq 0.007$$) amongst patients with higher distress scores [93]. Due to the characteristics of the study design, it is not possible to differentiate stress as an independent risk factor, as opposed to a consequence of long-term toxicity. Even though a longitudinal study of the current analysis could potentially assess this variable, distress scores were not previously planned and included in this cohort. In this study, GSTP1 c313AG or GG and XPC c.2815 AA genotypes increased the odds of moderate/severe ototoxicity 4.20- and 3.13-fold, respectively. Preclinical studies have demonstrated that GSTP1 c313 A>G encodes a change from isoleucine to valine in codon 105, leading to reduced protein activity and detoxification [63], while the XPC c.2815 C allele promotes the change from lysine to glutamine in codon 939, also diminishing protein activity and, consequently, DNA repair (Table A1) [64]. There is, however, marked heterogeneity of clinical effects in terms of the currently available literature. For instance, GSTP1 c313AG or GG was associated with an increased risk of moderate/severe hearing impairment in 106 [41] and 64 children [43], respectively, treated with platinum agents, using the Brock hearing loss classification of equal or greater than 2 [20]. The association between cumulative CDDP and ototoxicity was found in the study conducted by Lui and colleagues [41] but not in the study by Sherief and colleagues [43]. Even though our findings in a previous analysis of the data [45] are similar and in agreement with the functional roles of GSTP1 c313A>G [94], this SNV was not related to CDDP-induced ototoxicity in an additional cohort that recruited 71 children and young patients with various solid tumors [38], while in 173 patients with testicular carcinoma, post-treatment audiometric evaluations prompted divergent results, even though baseline assessments were not retrievable [39]. Reported results were also conflicting for isolated XPC c.2815A>G [35]. The XPC c.2815AA genotype was associated with an increased risk of any grade of toxicity [46] in a previous analysis of the data conducted by our group, and the same effect was observed in a smaller subset of patients with osteosarcoma [44]. Nonetheless, Lui and colleagues [41] did not present a significant association among 106 pediatric patients treated with platin analogs. Functional analyses performed for this variant [64] suggest the presence of the C allele reduces DNA repair, which would theoretically increase the risk of toxicity in contrast to what is currently reported, though an additional assay from Khan and colleagues [26] did not demonstrate a clear difference for the rate of nucleotide excision repair. Differences in the results obtained from the studies are not easily explained and may have originated from limitations related to sample size, patient baseline characteristics, tumor types, and treatment administered. There are also markedly distinct hearing loss classifications applied in previous cohorts, hampering proper direct comparisons with NCI CTCAE v4.03. Larger cohorts, in addition to further functional assays, would be ideal to better confront these findings. The metabolism of CDDP is known to involve cellular efflux, NER, and MMR damage repair, as well as apoptosis [37]. We were able to observe meaningful interactions between variants encoding those distinct pathways, suggesting that toxicity may be enhanced by the coexistence of more than one mutation in the different stages of CDDP metabolism and cytotoxic effect. The combination of GSTM1 null plus XPC c.2815AA, GSTP1 c.313AG or GG plus XPC c.2815AA, XPD c.934AA or EXO1 c.1762AA, or XPC c.2815AA plus MSH3 c.3133AA or FASL c.-844CC (Table A1) intensified the odds of moderate/severe ototoxicity up to 32.22-fold. The variant alleles from SNVs XPD c.934 (A) [65], EXO1 c.1762 (A) [71], and MSH3 c.3133 (A) [70] have been shown to reduce DNA repair activity by encoding amino acid replacements with the consequent loss of protein function or expression (Table A1). Additionally, the SNV FASL c.-844 is located within the enhancer-biding region of FASL, and luciferase assays have described the variant genotype TT to promote protein affinity twice lower than wild CC donors, leading to less protein expression and, as such, reduced apoptosis [74]. Hence, the combination of genotypes enhancing CDDP accumulation and reducing repair or activating apoptosis could potentiate the risk of ototoxicity, as observed in this analysis. To our knowledge, no studies focusing on the effects of the combinations of SNVs on the genes of distinct pathways of CDDP metabolism have been conducted to date. An association with MMR and apoptosis mechanisms was also noted in this study, as the combination of EXO1 c.1762GA or AA and P53 c.215CC genotypes increased the risk for events with OR 20.97. The P53 c.215 wild allele encoding arginine (G) was described as more efficient in inducing apoptosis than the proline (C) variant [72]. The P53 protein signaling pathway promotes cell death triggered by the generation of ROS [35]. In addition to apoptosis, P53 is related to cell cycle arrest, cell senescence, and DNA repair [95]. Cellular senescence is a state of permanent cell cycle arrest that is able to promote local inflammation and tissue damage [96]. In vitro studies have suggested that early senescence in response to genotoxic stress was P53-dependent and EXO1-depleted [97,98]. Moreover, Benkafadar and colleagues [99] observed that the response to ROS-induced DNA damage leads to cochlear cell senescence by activating the P53 pathway and hence contributing to age-related hearing loss. To date, there is a lack of evidence for a direct association between ototoxicity by CDDP and cell senescence. However, the accumulation of senescent neuronal cells is associated with CDDP-induced peripheral neuropathy in mice [100]. Thus, we may infer that patients with EXO1 c.1792GA or AA and P53 c.215CC combined genotypes were more efficient in promoting cell cycle arrest and senescence of sensory cells after injury by CDDP and, consequently, were at greater risk of hearing loss when compared to patients carrying other genotypes. We are aware that this study is limited for its sample size; thus, similarly to previous studies, lacking power for further SNVs combinations or polygenic evaluations and correction for other possible confounders. Though statistical tools were used to stabilize risk, such as bootstrap and power post hoc calculations, there may still be unknown influential factors not identifiable in this sample. It must also be considered that not all SNVs in the genes related to CDDP detoxification, DNA repair, and the apoptosis of damaged cells were evaluated in this study; only those recognized with a greater potential to induce ototoxicity were evaluated here. Thus, it is possible that other SNVs with equal or even greater importance in CDDP ototoxicity will be identified in future studies. Furthermore, other known SNVs for CDDP-induced ototoxicity unrelated to stages regarding drug absorption, distribution, metabolism, and excretion were not assessed and could be additional confounders. There is evidence supporting additional SNVs as risk factors for ototoxicity induced by CDDP related to ACYP2 [33,34], TPMT [35], COMT [35], and WFS1 [36] genes. ACYP2 is known to influence ATP-dependent calcium signaling, which may play a role in sensorineural hearing loss [33]. TPMT and COMT are methyltransferases that may inactivate CDDP and purine compounds. The Mendelian deafness genes, amongst which WSF1 is included, encode proteins reported to control endothelial reticulum stress response, thus influencing inner ear cellular damage [36]. Apart from known and unknown genetic risk factors for toxicity and hearing loss, clinical variables, such as a history of noise exposure [18] and distress scores [93], were also not collected from this cohort. We believe, however, that the exclusion of patients with reported hearing loss and hearing impairment in audiometry before treatment could, to some extent, attenuate these limitations. It is also important to consider distinct patient characteristics when assessing the generalizability of this study for other tumors since the population was predominantly male, with a high frequency of smokers and alcohol users, as well as locally advanced stages of HNSCC. Treatment approaches in the field of RT may also be distinct and could influence the prevalence and severity of adverse events, mainly in institutions with more frequent use of intensity-modulated RT. Though prespecified treatment protocols were strictly followed, therefore preventing confounding to some extent, heterogeneity in therapy protocols could affect the generalizability of these results. ## 5. Conclusions The results of this cohort suggest, for the first time, the interactions of inherited genetic abnormalities involved in CDDP metabolism as potential candidate targets for future risk models in ototoxicity. The development of genetic and clinical risk prediction tools is essential, not only for optimizing treatment selection based on efficacy but also to assist in supportive care during therapy. We believe these results may be included in future polygenic and clinical predictive models. ## References 1. Siegel R.L., Miller K.D., Fuchs H.E., Jemal A.. Cancer Statistics, 2021. CA Cancer J. Clin. (2021.0) 71 7-33. DOI: 10.3322/caac.21654 2. Sung H., Ferlay J., Siegel R.L., Laversanne M., Soerjomataram I., Jemal A., Bray F.. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. (2021.0) 71 209-249. DOI: 10.3322/caac.21660 3. Vermorken J.B., Remenar E., van Herpen C., Gorlia T., Mesia R., Degardin M., Stewart J.S., Jelic S., Betka J., Preiss J.H.. Cisplatin, Fluorouracil, and Docetaxel in Unresectable Head and Neck Cancer. N. Engl. J. Med. (2007.0) 357 1695-1704. DOI: 10.1056/NEJMoa071028 4. Argiris A., Karamouzis M.V., Raben D., Ferris R.L.. Head and Neck Cancer. Lancet Lond. Engl. (2008.0) 371 1695-1709. DOI: 10.1016/S0140-6736(08)60728-X 5. Bonner J.A., Harari P.M., Giralt J., Azarnia N., Shin D.M., Cohen R.B., Jones C.U., Sur R., Raben D., Jassem J.. Radiotherapy plus Cetuximab for Squamous-Cell Carcinoma of the Head and Neck. N. Engl. J. Med. (2006.0) 354 567-578. DOI: 10.1056/NEJMoa053422 6. Vermorken J.B., Mesia R., Rivera F., Remenar E., Kawecki A., Rottey S., Erfan J., Zabolotnyy D., Kienzer H.-R., Cupissol D.. Platinum-Based Chemotherapy plus Cetuximab in Head and Neck Cancer. N. Engl. J. Med. (2008.0) 359 1116-1127. DOI: 10.1056/NEJMoa0802656 7. Burtness B., Harrington K.J., Greil R., Soulières D., Tahara M., de Castro G., Psyrri A., Basté N., Neupane P., Bratland Å.. Pembrolizumab Alone or with Chemotherapy versus Cetuximab with Chemotherapy for Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (KEYNOTE-048): A Randomised, Open-Label, Phase 3 Study. Lancet (2019.0) 394 1915-1928. DOI: 10.1016/S0140-6736(19)32591-7 8. Albers J.W., Chaudhry V., Cavaletti G., Donehower R.C.. Interventions for Preventing Neuropathy Caused by Cisplatin and Related Compounds. Cochrane Database Syst. Rev. (2014.0) CD005228. DOI: 10.1002/14651858.CD005228.pub4 9. Pincinato E.C., Visacri M.B., de Souza C.M., Tuan B.T., Ferrari G.B., de Oliveira D.N., Barbosa C.R., Rodrigues R.F., Granja S., Ambrósio R.F.L.. Impact of Drug Formulation and Free Platinum/Cisplatin Ratio on Hypersensitivity Reactions to Cisplatin: Formulation Matters. J. Clin. Pharm. Ther. (2015.0) 40 41-47. DOI: 10.1111/jcpt.12220 10. Ganesan P., Schmiedge J., Manchaiah V., Swapna S., Dhandayutham S., Kothandaraman P.P.. Ototoxicity: A Challenge in Diagnosis and Treatment. J. Audiol. Otol. (2018.0) 22 59-68. DOI: 10.7874/jao.2017.00360 11. Laurell G.. Pharmacological Intervention in the Field of Ototoxicity. HNO (2019.0) 67 434-439. DOI: 10.1007/s00106-019-0663-1 12. Bokemeyer C., Berger C., Hartmann J., Kollmannsberger C., Schmoll H., Kuczyk M., Kanz L.. Analysis of Risk Factors for Cisplatin-Induced Ototoxicity in Patients with Testicular Cancer. Br. J. Cancer (1998.0) 77 1355-1362. DOI: 10.1038/bjc.1998.226 13. Frisina R.D., Wheeler H.E., Fossa S.D., Kerns S.L., Fung C., Sesso H.D., Monahan P.O., Feldman D.R., Hamilton R., Vaughn D.J.. Comprehensive Audiometric Analysis of Hearing Impairment and Tinnitus After Cisplatin-Based Chemotherapy in Survivors of Adult-Onset Cancer. J. Clin. Oncol. (2016.0) 34 2712-2720. DOI: 10.1200/JCO.2016.66.8822 14. Chen W.C., Jackson A., Budnick A.S., Pfister D.G., Kraus D.H., Hunt M.A., Stambuk H., Levegrun S., Wolden S.L.. Sensorineural Hearing Loss in Combined Modality Treatment of Nasopharyngeal Carcinoma. Cancer (2006.0) 106 820-829. DOI: 10.1002/cncr.21683 15. Caballero M., Mackers P., Reig O., Buxo E., Navarrete P., Blanch J.L., Grau J.J.. The Role of Audiometry Prior to High-Dose Cisplatin in Patients with Head and Neck Cancer. Oncology (2017.0) 93 75-82. DOI: 10.1159/000468522 16. Pearson S.E., Taylor J., Patel P., Baguley D.M.. Cancer Survivors Treated with Platinum-Based Chemotherapy Affected by Ototoxicity and the Impact on Quality of Life: A Narrative Synthesis Systematic Review. Int. J. Audiol. (2019.0) 58 685-695. DOI: 10.1080/14992027.2019.1660918 17. Rademaker-Lakhai J.M., Crul M., Zuur L., Baas P., Beijnen J.H., Simis Y.J.W., van Zandwijk N., Schellens J.H.M.. Relationship Between Cisplatin Administration and the Development of Ototoxicity. J. Clin. Oncol. (2006.0) 24 918-924. DOI: 10.1200/JCO.2006.10.077 18. Talach T., Rottenberg J., Gal B., Kostrica R., Jurajda M., Kocak I., Lakomy R., Vogazianos E.. Genetic Risk Factors of Cisplatin Induced Ototoxicity in Adult Patients. Neoplasma (2016.0) 63 263-268. DOI: 10.4149/212_140820N391 19. Silva Nogueira G.A., Dias Costa E.F., Lopes-Aguiar L., Penna Lima T.R., Visacri M.B., Pincinato E.C., Lourenço G.J., Calonga L., Mariano F.V., Messias de Almeida Milani Altemani A.. Polymorphisms in DNA Mismatch Repair Pathway Genes Predict Toxicity and Response to Cisplatin Chemoradiation in Head and Neck Squamous Cell Carcinoma Patients. Oncotarget (2018.0) 9 29538-29547. DOI: 10.18632/oncotarget.25268 20. Clemens E., Broer L., Langer T., Uitterlinden A.G., de Vries A.C.H., van Grotel M., Pluijm S.F.M., Binder H., Byrne J., Broeder E.. van D.; et al. Genetic Variation of Cisplatin-Induced Ototoxicity in Non-Cranial-Irradiated Pediatric Patients Using a Candidate Gene Approach: The International PanCareLIFE Study. Pharm. J. (2020.0) 20 294-305. DOI: 10.1038/s41397-019-0113-1 21. Siddik Z.H.. Cisplatin: Mode of Cytotoxic Action and Molecular Basis of Resistance. Oncogene (2003.0) 22 7265-7279. DOI: 10.1038/sj.onc.1206933 22. Kuo M.T., Chen H.H.W., Song I.-S., Savaraj N., Ishikawa T.. The Roles of Copper Transporters in Cisplatin Resistance. Cancer Metastasis Rev. (2007.0) 26 71-83. DOI: 10.1007/s10555-007-9045-3 23. Hayes J.D., Flanagan J.U., Jowsey I.R.. Glutathione Transferases. Annu. Rev. Pharmacol. Toxicol. (2005.0) 45 51-88. DOI: 10.1146/annurev.pharmtox.45.120403.095857 24. Ishikawa T., Wright C., Ishizuka H.. GS-X Pump Is Functionally Overexpressed in Cis-Diamminedichloroplatinum (II)-Resistant Human Leukemia HL-60 Cells and down-Regulated by Cell Differentiation. J. Biol. Chem. (1994.0) 269 29085-29093. DOI: 10.1016/S0021-9258(19)62016-8 25. Shuck S.C., Short E.A., Turchi J.J.. Eukaryotic Nucleotide Excision Repair: From Understanding Mechanisms to Influencing Biology. Cell Res. (2008.0) 18 64-72. DOI: 10.1038/cr.2008.2 26. Khan S.G.. A New Xeroderma Pigmentosum Group C Poly(AT) Insertion/Deletion Polymorphism. Carcinogenesis (2000.0) 21 1821-1825. DOI: 10.1093/carcin/21.10.1821 27. Lainé J.P., Mocquet V., Bonfanti M., Braun C., Egly J.M., Brousset P.. Common XPD (ERCC2) Polymorphisms Have No Measurable Effect on Nucleotide Excision Repair and Basal Transcription. DNA Repair (2007.0) 6 1264-1270. DOI: 10.1016/j.dnarep.2007.02.010 28. Goode E.L., Ulrich C.M., Potter J.D.. Polymorphisms in DNA Repair Genes and Associations with Cancer Risk. Cancer Epidemiol. Biomark. Prev. (2002.0) 11 1513 29. Martin L.P., Hamilton T.C., Schilder R.J.. Platinum Resistance: The Role of DNA Repair Pathways. Clin. Cancer Res. (2008.0) 14 1291-1295. DOI: 10.1158/1078-0432.CCR-07-2238 30. Mello J.A., Acharya S., Fishe R., Essigmann’ J.M.. The Mismatch-Repair Protein HMSH2 Binds Selectively to DNA Adducts of the Anticancer Drug Cisplatin. Chem. Biol. (1996.0) 3 579-589. DOI: 10.1016/S1074-5521(96)90149-0 31. Singh S.. Cytoprotective and Regulatory Functions of Glutathione S-Transferases in Cancer Cell Proliferation and Cell Death. Cancer Chemother. Pharmacol. (2015.0) 75 1-15. DOI: 10.1007/s00280-014-2566-x 32. De Zio D., Cianfanelli V., Cecconi F.. New Insights into the Link Between DNA Damage and Apoptosis. Antioxid. Redox Signal. (2013.0) 19 559-571. DOI: 10.1089/ars.2012.4938 33. Drögemöller B.I., Brooks B., Critchley C., Monzon J.G., Wright G.E.B., Liu G., Renouf D.J., Kollmannsberger C.K., Bedard P.L., Hayden M.R.. Further Investigation of the Role of. Clin. Cancer Res. (2018.0) 24 1866. DOI: 10.1158/1078-0432.CCR-17-2810 34. Driessen C.M., Ham J.C., te Loo M., van Meerten E., van Lamoen M., Hakobjan M.H., Takes R.P., van der Graaf W.T., Kaanders J.H., Coenen M.J.H.. Genetic Variants as Predictive Markers for Ototoxicity and Nephrotoxicity in Patients with Locally Advanced Head and Neck Cancer Treated with Cisplatin-Containing Chemoradiotherapy (The PRONE Study). Cancers (2019.0) 11. DOI: 10.3390/cancers11040551 35. Tserga E., Nandwani T., Edvall N.K., Bulla J., Patel P., Canlon B., Cederroth C.R., Baguley D.M.. The Genetic Vulnerability to Cisplatin Ototoxicity: A Systematic Review. Sci. Rep. (2019.0) 9 3455. DOI: 10.1038/s41598-019-40138-z 36. Wheeler H.E., Gamazon E.R., Frisina R.D., Perez-Cervantes C., El Charif O., Mapes B., Fossa S.D., Feldman D.R., Hamilton R.J., Vaughn D.J.. Variants in. Clin. Cancer Res. (2017.0) 23 3325. DOI: 10.1158/1078-0432.CCR-16-2809 37. Trendowski M.R., El Charif O., Dinh P.C., Travis L.B., Dolan M.E.. Genetic and Modifiable Risk Factors Contributing to Cisplatin-Induced Toxicities. Clin. Cancer Res. (2019.0) 25 1147-1155. DOI: 10.1158/1078-0432.CCR-18-2244 38. Peters U., Preisler-Adams S., Hebeisen A., Hahn M., Seifert E., Lanvers C., Heinecke A., Horst J., Jürgens H., Lamprecht-Dinnesen A.. Glutathione S-Transferase Genetic Polymorphisms and Individual Sensitivity to the Ototoxic Effect of Cisplatin. Anticancer Drugs (2000.0) 11 639-643. DOI: 10.1097/00001813-200009000-00007 39. Oldenburg J., Kraggerud S.M., Cvancarova M., Lothe R.A., Fossa S.D.. Cisplatin-Induced Long-Term Hearing Impairment Is Associated With Specific Glutathione S-Transferase Genotypes in Testicular Cancer Survivors. J. Clin. Oncol. (2007.0) 25 708-714. DOI: 10.1200/JCO.2006.08.9599 40. Choeyprasert W., Sawangpanich R., Lertsukprasert K., Udomsubpayakul U., Songdej D., Unurathapan U., Pakakasama S., Hongeng S.. Cisplatin-Induced Ototoxicity in Pediatric Solid Tumors: The Role of Glutathione S-Transferases and Megalin Genetic Polymorphisms. J. Pediatr. Hematol. Oncol. (2013.0) 35 e138-e143. DOI: 10.1097/MPH.0b013e3182707fc5 41. Lui G., Bouazza N., Denoyelle F., Moine M., Chastagner P., Corradini N., Entz-Werle N., Vérité C., Landmanparker J., Sudour-Bonnange H.. Association between Genetic Polymorphisms and Platinum-Induced Ototoxicity in Children. Oncotarget (2018.0) 9 30883-30893. DOI: 10.18632/oncotarget.25767 42. Budai B., Prekopp P., Noszek L., Kovács E., Szőnyi M., Erdélyi D., Biró K., Geczi L.. GSTM1 Null and GSTT1 Null: Predictors of Cisplatin-Caused Acute Ototoxicity Measured by DPOAEs. J. Mol. Med. (2020.0) 98 963-971. DOI: 10.1007/s00109-020-01921-y 43. Sherief L.M., Rifky E., Attia M., Ahmed R., Kamal N.M., Oshi M.A.M., Hanna D.. Platinum-Induced Ototoxicity in Pediatric Cancer Survivors: GSTP1 c.313A>G Variant Association. Medicine (2022.0) 101 e31627. DOI: 10.1097/MD.0000000000031627 44. Caronia D., Patiño-García A., Milne R.L., Zalacain-Díez M., Pita G., Alonso M.R., Moreno L.T., Sierrasesumaga-Ariznabarreta L., Benítez J., González-Neira A.. Common Variations in ERCC2 Are Associated with Response to Cisplatin Chemotherapy and Clinical Outcome in Osteosarcoma Patients. Pharm. J. (2009.0) 9 347-353. DOI: 10.1038/tpj.2009.19 45. Pincinato E.C., Costa E.F.D., Lopes-Aguiar L., Nogueira G.A.S., Lima T.R.P., Visacri M.B., Costa A.P.L., Lourenço G.J., Calonga L., Mariano F.V.. GSTM1, GSTT1 and GSTP1 Ile105Val Polymorphisms in Outcomes of Head and Neck Squamous Cell Carcinoma Patients Treated with Cisplatin Chemoradiation. Sci. Rep. (2019.0) 9 9312. DOI: 10.1038/s41598-019-45808-6 46. Lopes-Aguiar L., Costa E.F.D., Nogueira G.A.S., Lima T.R.P., Visacri M.B., Pincinato E.C., Calonga L., Mariano F.V., de Almeida Milani Altemani A.M., Altemani J.M.C.. XPD c.934G>A Polymorphism of Nucleotide Excision Repair Pathway in Outcome of Head and Neck Squamous Cell Carcinoma Patients Treated with Cisplatin Chemoradiation. Oncotarget (2017.0) 8 16190-16201. DOI: 10.18632/oncotarget.7668 47. Costa E.F.D., Lima T.R.P., Lopes-Aguiar L., Nogueira G.A.S., Visacri M.B., Quintanilha J.C.F., Pincinato E.C., Calonga L., Mariano F.V., de Almeida Milani Altemani A.M.. FAS and FASL Variations in Outcomes of Tobacco- and Alcohol-Related Head and Neck Squamous Cell Carcinoma Patients. Tumor Biol. (2020.0) 42 1010428320938494. DOI: 10.1177/1010428320938494 48. Oken M., Creech R., Tormey D., Horton J., Davis T., McFadden E., Carbone P.. Toxicity and Response Criteria of the Eastern Cooperative Oncology Group. Am. J. Clin. Oncol. (1982.0) 6 649-655. DOI: 10.1097/00000421-198212000-00014 49. Forastiere A.A., Zhang Q., Weber R.S., Maor M.H., Goepfert H., Pajak T.F., Morrison W., Glisson B., Trotti A., Ridge J.A.. Long-Term Results of RTOG 91-11: A Comparison of Three Nonsurgical Treatment Strategies to Preserve the Larynx in Patients with Locally Advanced Larynx Cancer. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. (2013.0) 31 845-852. DOI: 10.1200/JCO.2012.43.6097 50. Common Terminology Criteria for Adverse Events (CTCAE) v5.0. Published November, 27, 2017 51. von Elm E., Altman D.G., Egger M., Pocock S.J., Gøtzsche P.C., Vandenbroucke J.P.. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. PLoS Med. (2007.0) 4. DOI: 10.1371/journal.pmed.0040296 52. Hayes R.B., Bravo-Otero E., Kleinman D.V., Brown L.M., Fraumeni J.F., Harty L.C., Winn D.M.. Tobacco and Alcohol Use and Oral Cancer in Puerto Rico. Cancer Causes Control (1999.0) 10 27-33. DOI: 10.1023/A:1008876115797 53. Bray G.A., Heisel W.E., Afshin A., Jensen M.D., Dietz W.H., Long M., Kushner R.F., Daniels S.R., Wadden T.A., Tsai A.G.. The Science of Obesity Management: An Endocrine Society Scientific Statement. Endocr. Rev. (2018.0) 39 79-132. DOI: 10.1210/er.2017-00253 54. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes—2022. Diabetes Care (2021.0) 45 S17-S38. DOI: 10.2337/dc22-S002 55. Toto R.D.. Defining Hypertension: Role of New Trials and Guidelines. Clin. J. Am. Soc. Nephrol. CJASN (2018.0) 13 1578-1580. DOI: 10.2215/CJN.05350418 56. Lydiatt W.M., Patel S.G., O’Sullivan B., Brandwein M.S., Ridge J.A., Migliacci J.C., Loomis A.M., Shah J.P.. Head and Neck Cancers—Major Changes in the American Joint Committee on Cancer Eighth Edition Cancer Staging Manual. CA Cancer J. Clin. (2017.0) 67 122-137. DOI: 10.3322/caac.21389 57. Clark J.. Uses and Abuses of Hearing Loss Classification. ASHA (1981.0) 23 493-500. PMID: 7052898 58. American Speech-Language-Hearing Association Determining Threshold Level for Speech [Guidelines]. (1988.0) 59. Jerger J., Speaks C., Trammell J.L.. A New Approach to Speech Audiometry. J. Speech Hear. Disord. (1968.0) 33 318-328. DOI: 10.1044/jshd.3304.318 60. Programme for the Prevention of Deafness and Hearing Impairment. Report of the Informal Working Group on Prevention of Deafness and Hearing Impairment Programme Planning, Geneva, 18–21 June 1991 (1991.0) 61. Stevens G., Flaxman S., Brunskill E., Mascarenhas M., Mathers C.D., Finucane M.. On behalf of the Global Burden of Disease Hearing Loss Expert Group Global and Regional Hearing Impairment Prevalence: An Analysis of 42 Studies in 29 Countries. Eur. J. Public Health (2013.0) 23 146-152. DOI: 10.1093/eurpub/ckr176 62. Moyer A.M., Salavaggione O.E., Hebbring S.J., Moon I., Hildebrandt M.A.T., Eckloff B.W., Schaid D.J., Wieben E.D., Weinshilboum R.M.. Glutathione S-Transferase T1 and M1: Gene Sequence Variation and Functional Genomics. Clin. Cancer Res. (2007.0) 13 7207-7216. DOI: 10.1158/1078-0432.CCR-07-0635 63. Terrier P., Townsend A.J., Coindre J.M., Triche T.J., Cowan K.H.. An Immunohistochemical Study of Pi Class Glutathione S-Transferase Expression in Normal Human Tissue. Am. J. Pathol. (1990.0) 137 845-853. PMID: 1977319 64. Zhu Y., Yang H., Chen Q., Lin J., Grossman H.B., Dinney C.P., Wu X., Gu J.. Modulation of DNA Damage/DNA Repair Capacity by XPC Polymorphisms. DNA Repair (2008.0) 7 141-148. DOI: 10.1016/j.dnarep.2007.08.006 65. Spitz M.R., Wu X., Wang Y., Wang L.-E., Shete S., Amos C.I., Guo Z., Lei L., Mohrenweiser H., Wei Q.. Modulation of Nucleotide Excision Repair Capacity by XPD Polymorphisms in Lung Cancer Patients. Cancer Res. (2001.0) 61 1354. PMID: 11245433 66. Vaezi A., Wang X., Buch S., Gooding W., Wang L., Seethala R.R., Weaver D.T., D’Andrea A.D., Argiris A., Romkes M.. XPF Expression Correlates with Clinical Outcome in Squamous Cell Carcinoma of the Head and Neck. Clin. Cancer Res. (2011.0) 17 5513-5522. DOI: 10.1158/1078-0432.CCR-11-0086 67. Yu J.J., Lee K.B., Mu C., Li Q., Abernathy T.V., Bostick-Bruton F., Reed E.. Comparison of Two Human Ovarian Carcinoma Cell Lines (A2780/CP70 and MCAS) That Are Equally Resistant to Platinum, but Differ at Codon 118 of the ERCC1 Gene. Int. J. Oncol. (2000.0) 16 555-615. DOI: 10.3892/ijo.16.3.555 68. Perera S., Mrkonjic M., Rawson J.B., Bapat B.. Functional Effects of the MLH1-93G>A Polymorphism on MLH1/EPM2AIP1 Promoter Activity. Oncol. Rep. (2011.0) 25 809-815. DOI: 10.3892/or.2010.1129 69. Marra G., D’Atri S., Corti C., Bonmassar L., Cattaruzza M.S., Schweizer P., Heinimann K., Bartosova Z., Nyström-Lahti M., Jiricny J.. Tolerance of Human MSH2+/− Lymphoblastoid Cells to the Methylating Agent Temozolomide. Proc. Natl. Acad. Sci. USA (2001.0) 98 7164-7169. DOI: 10.1073/pnas.121136498 70. Nogueira G.A.S., Lourenço G.J., Oliveira C.B.M., Marson F.A.L., Lopes-Aguiar L., Costa E.F.D., Lima T.R.P., Liutti V.T., Leal F., Santos V.C.A.. Association between Genetic Polymorphisms in DNA Mismatch Repair-Related Genes with Risk and Prognosis of Head and Neck Squamous Cell Carcinoma. Int. J. Cancer (2015.0) 137 810-818. DOI: 10.1002/ijc.29435 71. Wei K., Clark A.B., Wong E., Kane M.F., Mazur D.J., Parris T., Kolas N.K., Russell R., Hou H., Kneitz B.. Inactivation of Exonuclease 1 in Mice Results in DNA Mismatch Repair Defects, Increased Cancer Susceptibility, and Male and Female Sterility. Genes Dev. (2003.0) 17 603-614. DOI: 10.1101/gad.1060603 72. Dumont P., Leu J.I.-J., Della Pietra A.C., George D.L., Murphy M.. The Codon 72 Polymorphic Variants of P53 Have Markedly Different Apoptotic Potential. Nat. Genet. (2003.0) 33 357-365. DOI: 10.1038/ng1093 73. Sibley K., Rollinson S., Allan J.M., Smith A.G., Law G.R., Roddam P.L., Skibola C.F., Smith M.T., Morgan G.J.. Functional FAS Promoter Polymorphisms Are Associated with Increased Risk of Acute Myeloid Leukemia1. Cancer Res. (2003.0) 63 4327-4330. PMID: 12907599 74. Wu J., Metz C., Xu X., Abe R., Gibson A.W., Edberg J.C., Cooke J., Xie F., Cooper G.S., Kimberly R.P.. A Novel Polymorphic CAAT/Enhancer-Binding Protein Beta Element in the FasL Gene Promoter Alters Fas Ligand Expression: A Candidate Background Gene in African American Systemic Lupus Erythematosus Patients. J. Immunol. (2003.0) 170 132-138. DOI: 10.4049/jimmunol.170.1.132 75. Jang J.S., Kim K.M., Choi J.E., Cha S.I., Kim C.H., Lee W.K., Kam S., Jung T.H., Park J.Y.. Identification of Polymorphisms in the Caspase-3 Gene and Their Association with Lung Cancer Risk. Mol. Carcinog. (2008.0) 47 383-390. DOI: 10.1002/mc.20397 76. Chen K., Zhao H., Hu Z., Wang L.-E., Zhang W., Sturgis E.M., Wei Q.. Clin. Cancer Res. (2008.0) 14 6343. DOI: 10.1158/1078-0432.CCR-08-1198 77. Benjamini Y., Hochberg Y.. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J. R. Stat. Soc. Ser. B Methodol. (1995.0) 57 289-300. DOI: 10.1111/j.2517-6161.1995.tb02031.x 78. Levine M., Ensom M.H.H.. Post Hoc Power Analysis: An Idea Whose Time Has Passed?. Pharm. J. Hum. Pharmacol. Drug Ther. (2001.0) 21 405-409. DOI: 10.1592/phco.21.5.405.34503 79. Wang M., Xu S.. Statistical Power in Genome-Wide Association Studies and Quantitative Trait Locus Mapping. Heredity (2019.0) 123 287-306. DOI: 10.1038/s41437-019-0205-3 80. Efron B.. Bootstrap Methods: Another Look at the Jackknife. Ann. Stat. (1979.0) 7 1-26. DOI: 10.1214/aos/1176344552 81. Kopke R., Liu W., Gabaizadeh R., Jacono A., Feghali J., Spray D., Garcia P., Steinman H., Malgrange B., Ruben R.. Use of Organotypic Cultures of Corti’s Organ to Study the Protective Effects of Antioxidant Molecules on Cisplatin-Induced Damage of Auditory Hair Cells. Am. J. Otol. (1997.0) 18 559-571. PMID: 9303151 82. Breglio A.M., Rusheen A.E., Shide E.D., Fernandez K.A., Spielbauer K.K., McLachlin K.M., Hall M.D., Amable L., Cunningham L.L.. Cisplatin Is Retained in the Cochlea Indefinitely Following Chemotherapy. Nat. Commun. (2017.0) 8 1654. DOI: 10.1038/s41467-017-01837-1 83. Nomura K., Nakao M., Morimoto T.. Effect of Smoking on Hearing Loss: Quality Assessment and Meta-Analysis. Prev. Med. (2005.0) 40 138-144. DOI: 10.1016/j.ypmed.2004.05.011 84. Capra Marques de Oliveira D.C., de Melo Tavares de Lima M.A.. Low and High Frequency Tonal Threshold Audiometry: Comparing Hearing Thresholds between Smokers and Non-Smokers. Braz. J. Otorhinolaryngol. (2009.0) 75 738-744. DOI: 10.1016/S1808-8694(15)30527-9 85. Le T.N., Straatman L.V., Lea J., Westerberg B.. Current Insights in Noise-Induced Hearing Loss: A Literature Review of the Underlying Mechanism, Pathophysiology, Asymmetry, and Management Options. J. Otolaryngol.—Head Neck Surg. (2017.0) 46 41. DOI: 10.1186/s40463-017-0219-x 86. Berg R., Pickett W., Linneman J., Wood D., Marlenga B.. Asymmetry in Noise-Induced Hearing Loss: Evaluation of Two Competing Theories. Noise Health (2014.0) 16 102-107. PMID: 24804714 87. Schmidt C.-M., Knief A., Lagosch A.K., Deuster D., am Zehnhoff-Dinnesen A.. Left-Right Asymmetry in Hearing Loss Following Cisplatin Therapy in Children—The Left Ear Is Slightly but Significantly More Affected. Ear Hear. (2008.0) 29 830-837. DOI: 10.1097/AUD.0b013e31818005a4 88. Hearing Loss Prevalence and Years Lived with Disability, 1990–2019: Findings from the Global Burden of Disease Study 2019. Lancet (2021.0) 397 996-1009. DOI: 10.1016/S0140-6736(21)00516-X 89. Dillard L.K., Lopez-Perez L., Martinez R.X., Fullerton A.M., Chadha S., McMahon C.M.. Global Burden of Ototoxic Hearing Loss Associated with Platinum-Based Cancer Treatment: A Systematic Review and Meta-Analysis. Cancer Epidemiol. (2022.0) 79 102203. DOI: 10.1016/j.canep.2022.102203 90. Hornsby B.W.Y., Ricketts T.A.. The Effects of Hearing Loss on the Contribution of High- and Low-Frequency Speech Information to Speech Understanding. II. Sloping Hearing Loss. J. Acoust. Soc. Am. (2006.0) 119 1752-1763. DOI: 10.1121/1.2161432 91. Shahbazi M., Zhang X., Dinh P.C., Sanchez V.A., Trendowski M.R., Shuey M.M., Nguyen T., Feldman D.R., Vaughn D.J.. Comprehensive Association Analysis of Speech Recognition Thresholds after Cisplatin-Based Chemotherapy in Survivors of Adult-Onset Cancer. Cancer Med. (2022.0) 12 2999-3012. DOI: 10.1002/cam4.5218 92. Eckerling A., Ricon-Becker I., Sorski L., Sandbank E., Ben-Eliyahu S.. Stress and Cancer: Mechanisms, Significance and Future Directions. Nat. Rev. Cancer (2021.0) 21 767-785. DOI: 10.1038/s41568-021-00395-5 93. Miaskowski C., Paul S.M., Mastick J., Abrams G., Topp K., Smoot B., Kober K.M., Chesney M., Mazor M., Mausisa G.. Associations Between Perceived Stress and Chemotherapy-Induced Peripheral Neuropathy and Otoxicity in Adult Cancer Survivors. J. Pain Symptom Manag. (2018.0) 56 88-97. DOI: 10.1016/j.jpainsymman.2018.02.021 94. Watson M.A., Stewart R.K., Smith G.B., Massey T.E., Bell D.A.. Human Glutathione S-Transferase P1 Polymorphisms: Relationship to Lung Tissue Enzyme Activity and Population Frequency Distribution. Carcinogenesis (1998.0) 19 275-280. DOI: 10.1093/carcin/19.2.275 95. Aubrey B.J., Kelly G.L., Janic A., Herold M.J., Strasser A.. How Does P53 Induce Apoptosis and How Does This Relate to P53-Mediated Tumour Suppression?. Cell Death Differ. (2018.0) 25 104-113. DOI: 10.1038/cdd.2017.169 96. Hernandez-Segura A., Nehme J., Demaria M.. Hallmarks of Cellular Senescence. Trends Cell Biol. (2018.0) 28 436-453. DOI: 10.1016/j.tcb.2018.02.001 97. Aasland D., Götzinger L., Hauck L., Berte N., Meyer J., Effenberger M., Schneider S., Reuber E.E., Roos W.P., Tomicic M.T.. Temozolomide Induces Senescence and Repression of DNA Repair Pathways in Glioblastoma Cells via Activation of ATR–CHK1, P21, and NF-ΚB. Cancer Res. (2019.0) 79 99-113. DOI: 10.1158/0008-5472.CAN-18-1733 98. Allmann S., Mayer L., Olma J., Kaina B., Hofmann T.G., Tomicic M.T., Christmann M.. Benzo[a]Pyrene Represses DNA Repair through Altered E2F1/E2F4 Function Marking an Early Event in DNA Damage-Induced Cellular Senescence. Nucleic Acids Res. (2020.0) 48 12085-12101. DOI: 10.1093/nar/gkaa965 99. Benkafadar N., François F., Affortit C., Casas F., Ceccato J.-C., Menardo J., Venail F., Malfroy-Camine B., Puel J.-L., Wang J.. ROS-Induced Activation of DNA Damage Responses Drives Senescence-Like State in Postmitotic Cochlear Cells: Implication for Hearing Preservation. Mol. Neurobiol. (2019.0) 56 5950-5969. DOI: 10.1007/s12035-019-1493-6 100. Acklin S., Zhang M., Du W., Zhao X., Plotkin M., Chang J., Campisi J., Zhou D., Xia F.. Depletion of Senescent-like Neuronal Cells Alleviates Cisplatin-Induced Peripheral Neuropathy in Mice. Sci. Rep. (2020.0) 10 14170. DOI: 10.1038/s41598-020-71042-6
--- title: Low-Intensity Physical Exercise Decreases Inflammation and Joint Damage in the Preclinical Phase of a Rheumatoid Arthritis Murine Model authors: - Susana Aideé González-Chávez - Salma Marcela López-Loeza - Samara Acosta-Jiménez - Rubén Cuevas-Martínez - César Pacheco-Silva - Eduardo Chaparro-Barrera - César Pacheco-Tena journal: Biomolecules year: 2023 pmcid: PMC10046494 doi: 10.3390/biom13030488 license: CC BY 4.0 --- # Low-Intensity Physical Exercise Decreases Inflammation and Joint Damage in the Preclinical Phase of a Rheumatoid Arthritis Murine Model ## Abstract Lifestyle modifications in preclinical Rheumatoid Arthritis (RA) could delay the ongoing pathogenic immune processes and potentially prevent its onset. Physical exercise (PE) benefits RA patients; however, its impact in reducing the risk of developing RA has scarcely been studied. The objective was to describe the effects of low-intensity PE applied at the disease’s preclinical phase on the joints of DBA/1 mice with collagen-induced arthritis (CIA). Twelve mice with CIA were randomly distributed into two groups: the CIA-Ex group, which undertook treadmill PE, and the CIA-NoEx, which was not exercised. The effects of PE were evaluated through clinical, histological, transcriptomics, and immunodetection analyses in the mice’s hind paws. The CIA-Ex group showed lower joint inflammation and damage and a decreased expression of RA-related genes (Tnf Il2, Il10, Il12a, IL23a, and Tgfb1) and signaling pathways (Cytokines, Chemokines, JAK-STAT, MAPK, NF-kappa B, TNF, and TGF-beta). TNF-α expression was decreased by PE in the inflamed joints. Low-intensity PE in pre-arthritic CIA reduced the severity through joint down-expression of proinflammatory genes and proteins. Knowledge on the underlying mechanisms of PE in preclinical arthritis and its impact on reducing the risk of developing RA is still needed. ## 1. Introduction Rheumatoid arthritis (RA) is an inflammatory autoimmune disease whose pathogenesis is complex and yet incompletely understood [1]. RA has a preclinical, asymptomatic phase in which environmental stimuli interact with a genetically predisposed host. Eventually, these interactions result in breach of tolerance in sites such as the mucosal surfaces, leading to the initiation, maturation, and amplification of autoimmunity outside the joint. Subsequently, active elements of the innate and adaptative immune response access the synovium, resulting in overt synovitis that is clinically detectable. The period before clinical inflammatory arthritis is designated preclinical RA in those individuals who have progressed to a clinical diagnosis of RA, and an at-risk status in those who exhibit predictive biomarkers of RA but have not developed inflammatory arthritis [2,3,4,5]. A better comprehension of the preclinical pathophysiological process contributes to identifying critical mediators of the disease and may allow the establishment of lifestyle modifications and specific therapies in the early preclinical stages that can delay the ongoing pathogenic immune processes and potentially prevent RA onset [6]. For example, a recent systematic review demonstrated that the early treatment of at-risk individuals might effectively delay RA onset, thereby decreasing disease-related limitations in individuals in the preclinical phase of RA [7]. On the other hand, non-pharmacological preventive strategies, such as the modification of behavioral RA-risk factors (smoking, obesity, low physical activity (PA), low-quality diet, and poor dental hygiene) among at-risk individuals, may decrease RA risk [5,8]. The beneficial effects of PA and physical exercise (PE) on the clinical, metabolic, and cardiorespiratory features in patients with established RA have been widely documented [9]; indeed, PA is considered part of the comprehensive management of patients [10,11]. Nevertheless, studies evaluating the effects on the risk of developing RA are scarce, and evidence from observational studies is not entirely consistent [12,13]. Furthermore, the heterogeneity of the designs of the studies evaluating PE in RA, aside from the high risk of bias, precludes conclusive results; therefore, further research is still necessary [14]. The effects of PE on joint biology, including the inflammatory process and structural damage, have not been sufficiently explored, partly due to the inconvenience of repeated biopsies from human joints in PE trials. In this regard, animal models of RA provide a valuable tool for exploring this field. We have recently reviewed the topic [15], finding that, in animal models of RA, PE can potentially exacerbate the joint inflammatory process; however, the information is inconclusive as some studies have also shown a beneficial effect. In our review, we described how the differential metabolic and immune effects of PE are dependent on numerous extent variables, including the animal model (species, strain, and type of arthritis induction), the type of PE (aerobic or anaerobic), the intensity (low-, moderate- or high-intensity), the time of intervention (morning and night), the use of stimulus (voluntary and forced), and the duration (acute and chronic) [15]. The exploration of the impact of PE, in any of its modalities and variants, in the preclinical phase of joint inflammation in animal models of RA is practically nil. Therefore, we aimed to evaluate the effect of low-intensity PE in preclinical collagen-induced arthritis (CIA) in DBA/1 mice. The joint biology was evaluated through transcriptomic, clinical, and histopathological analyses. ## 2. Materials and Methods The effects of low-intensity treadmill PE on joint biology in the preclinical CIA in DBA/1 mice were evaluated by comparing exercised and non-exercised mice. In addition, clinical and histological severity, transcriptome modifications, and protein expression were analyzed through histopathology, DNA microarray, and reverse transcription (RT)-quantitative polymerase chain reaction (qPCR). ## 2.1. Animals and Study Groups The study included 16 male DBA/1 mice aged 11–12 weeks from the Faculty of Medicine and Biomedical Sciences of the Autonomous University of Chihuahua, in which CIA was induced. Mice were randomly distributed into two groups with eight mice each: [1] the CIA-Ex group, which received PE from day 14 (one day after the second collagen injection), and [2] the CIA-NoEx group, which did not exercise and was the control group. The sample size was calculated with the formula for comparison between two groups for quantitative data, with a type I error of $5\%$, statistic power of $80\%$, and standard deviation (SD) and effect size from our previous research evaluating the effect of exercise in arthritis rodent models [16,17]. Mice from two groups were kept in the same building under controlled luminosity (12 h light/12 h dark) and temperature (23 ± 2 °C) and received food and water ad libitum. Mice were monitored by the researchers and the veterinarian throughout the entire research, including the PE routines. We established a priori that the exercise session should be stopped if mice have an accidental injury, fatigue, unexpected adverse effects, behavioral issues, or poor performance caused by the animal’s unwillingness to exercise. Additionally, we established that an animal should be removed if it becomes permanently unable to perform the exercise and requires excessive motivation to exercise. However, no mouse removal was necessary, and all analyses included eight mice per group. This study complied with the Official Mexican Standard NOM-062-ZOO-1999 technical specifications for producing, caring for, and using laboratory animals. The research was approved by the Ethics Committee and Institutional Animal Care and Use Committee (IACUC), with the ID number: CI-036-20. ## 2.2. Arthritis Induction The CIA was performed as described by Brand et al. [ 18] in 11–12-week-old mice. Under isoflurane anesthesia, mice were intradermally injected in the tail base with a suspension containing 0.1 mg of type II bovine collagen emulsified with complete Freund’s adjuvant (day 0). A second injection was applied on day 14 using Freund’s incomplete adjuvant and the same quantity of collagen. Mice were sacrificed once they completed the PE intervention. The incidence and clinical score of arthritis were evaluated using the semiquantitative scale, i.e., 0: no evidence of erythema and swelling; 1: erythema and mild swelling confined to the tarsals or ankle joint; 2: erythema and mild swelling extending from the ankle to the tarsals; 3: erythema and moderate swelling extending from the ankle to metatarsal joints; and 4: erythema and severe swelling encompass the ankle, foot, and digits. The total score per mouse was obtained by adding the score of the four limbs. ## 2.3. Treadmill Physical Exercise Intervention The PE intervention was performed on a custom-designed and built treadmill compliant with the American Physiological Society recommendations. Specific times and speeds were programmed using the software MRLabEx (Figure 1). For mice in the experimental group, a one-week familiarization period was included to minimize the psychological stress of the mice by promoting visual/olfactory and sound/motion adaptation to the treadmill. Mice were placed on the treadmill daily for 15 min: 10 min on the switched-off and 5 min walking at 5 m/min. Familiarization occurred between the first and second collagen injections (Figure 1). After the familiarization period, the speed at which the mice would perform the low-intensity exercise ($30\%$ and $50\%$ of their maximum capacity) was calculated. Treadmill-running performance was evaluated through a treadmill exhaustion test. Mice were placed on the band and ran for five minutes at 5 m/min as a warm-up. After this time, the speed increased by 3 m/min every 2 min until the mice were exhausted. The maximum PE capacity ($100\%$) was defined as the maximum speed reached by each mouse; then, the speed’s mean and SD were obtained per group. DBA/1 mice’s maximum PE capacity resulted in 32.9 ± 2.9 m/min, so $30\%$ and $50\%$ of their capacity were defined as 10 m/min and 15 m/min, respectively. Mice from the CIA-Ex group were exercised for three weeks with a frequency of five 35 min sessions per week (Monday to Friday). The PE session included the phases of [1] acclimatization: mice were placed for 4 min on the switched-off treadmill; [2] warm-up: mice walked for 3 min at 5 m/min; [3] PE: mice ran for 10 min at 10 m/min, followed by 15 min at 15 m/min; and [4] cool-down: mice walked for 3 min at 3 m/min. Treadmill PE intervention started when the clinical arthritis score was ≤1, the day after the second collagen injection. Mice from both groups were euthanized with isoflurane on day 35 (Figure 1). ## 2.4. DNA Microarray and Bioinformatic Analysis RNA was obtained from tarsal bones and joints, including ligaments. Tissues were disrupted in liquid nitrogen using a biopulverizer. Total RNA was purified using the RNeasy® Mini Kit (Qiagen, Hilden, Germany), following the manufacturer’s protocol. The RNA quantity and quality were verified in the Qubit4 fluorometer (ThermoFisher Scientific, Waltham, MA USA). The RNA of each mouse in each study group was mixed in equimolar amounts to conform pools used in the DNA microarray. The microarray was carried out at the Institute of Cellular Physiology, Autonomous University of Mexico (UNAM), Mexico. DNA microarray was performed to compare the CIA-Ex group with respect to the CIA-NoEx group. Briefly, the RT-PCR was performed and the resulting complementary DNA (cDNA) from the CIA-Ex group was labeled with Cy5, while the cDNA from the CIA-NoEx group was labeled with Cy3. Hybridization was performed using the M22K_01 (UNAM, Mexico City, Mexico) chip containing 22,000 genes from the mouse genome. The scan and signal acquisition were developed using the ScanArray 4000 (Packard BioChips Technologies, Billerica, MA, USA). The analysis of the microarray scanning was carried out using GenArise Microarray Analysis Tool software (UNAM), and the lists of differentially expressed genes (DEGs) (Z-score ≥ 1.5 SD) were obtained [19]. The microarray dataset was registered in the Gene Expression Omnibus (GEO) of the National Center for Biotechnology Information (NCBI) database with the accession number GSE212262. The lists of DEGs by PE were further analyzed in DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/, accessed on 1 January 2023), an open-resource platform that classifies genes list into functional biological processes and KEGG (Kyoto Encyclopedia of Genes and Genomes) signaling pathways [20]. Furthermore, the STRING database 11.5 (https://string-db.org/, accessed on 1 January 2023) was used to obtain the analysis and integration of direct and indirect protein–protein interactions (PPI) centered on the functional association [21]. The DEGs identified in the microarray were loaded, and the interactions with minimal confidence (interaction score > 0.4) were selected. The PPI network was more thoroughly analyzed to obtain primary clusters of sub-networks using the Cytoscape software v3.9.1 with the Molecular Complex Detection (MCODE) complement (node score cutoff = 0.6) [22,23]. The genes of the two primary clusters were loaded in the STRING database to show the PPI-associated KEGG pathways; genes of relevance in arthritis were identified in different colors. ## 2.5. Histopathological Analysis The hind paws were dissected and fixed in $10\%$ phosphate-buffered formalin for 48 h, and were then demineralized using $5\%$ nitric acid for 48 h, dehydrated in graded ethanol, and embedded in paraffin. Sections of 5 μm in thickness were obtained and placed on adhesive-coated glass slides. Histological assessment was carried out using hematoxylin and eosin (H&E) staining. The images were acquired using a digital camera coupled to the optical microscope. The influence of PE on tarsal joint structures was evaluated in three slices of each sample, using the semiquantitative scale of 0, absent; 1, mild; 2, moderate; or 3, severe to describe inflammatory infiltrate, synovial hyperplasia, cartilage damage, and bone erosion in the tarsal joints. The mean score was calculated for each group. The histological parameters were the primary outcome measured. IHC analysis was performed with a specific antibody against tumor necrosis factor (TNF)-α (sc-52746, Santa Cruz Biotechnology, Dallas, TX, USA). Tissue sections were deparaffinized in xylene and dehydrated in descending concentrations of ethanol until water. Antigen retrieval was carried out using $0.05\%$ trypsin (T1426-250 mg, SIGMA Life Science, St. Louis, MO, USA) for 20 min at 37 °C, and then tissues were treated with $0.2\%$ Triton-X100 (Bio-Rad, Hercules, CA, USA). After blocking with $10\%$ bovine serum albumin (BSA) (Sigma Life Science, St. Louis, MO, USA) for 30 min at 37 °C in a humidified chamber, tissues were treated with hydrogen peroxide for 10 min at room temperature to remove endogenous peroxidase activity. Tissues were incubated with the primary antibody diluted 1:1000 in $1\%$ BSA at 4 °C overnight. The corresponding isotype’s biotin-streptavidin-conjugated secondary antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) was used in a 1:400 dilution. Immunodetection was carried out using Pierce® streptavidin horseradish peroxidase-conjugated (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and Diaminobenzidine (DAB) (D4293-50SET, SIGMA-ALDRICH Co., St. Louis, MO, USA) as the chromogen. The primary antibody was replaced with PBS buffer to establish a negative control. Images were acquired using a digital camera (AmScope MU1803, Irvine, CA, USA) coupled with an optical microscope (AxioStar Plus, Carl Zeiss, Berlin, Germany), taking at least 20 microscopic fields from each study subject. The expression of each marker was quantified with the ImageJ program and the IHC toolbox. The DAB color was extracted from each image, and the maximum and mean gray values were obtained. Each image’s optical density (OD) was obtained with log10 (maximum gray value/mean gray value). The OD means and SD were calculated and graphed per study group. ## 2.6. Inflammatory Cytokines RNA Quantification by RT-qPCR RNA quantification was performed through RT-qPCR for Tnf, interleukin (Il)2, Il6, Il10, Il12a, IL23a, transforming growth factor (Tgfb1), and Janus kinase (Jak)3, with the primers sets listed in Table 1. The ribosomal protein L (Rpl)13 was used as the reference gene [24]. The cDNA was synthesized from total RNA (1µg) by RT using the SupeScript® III First-Strand Synthesis System (Invitrogen by ThermoFisher Scientific) with random hexamer primers, and was then diluted to 100 µL. For each gene, 3 µL of individual cDNA was used for qPCR using the Radiant™ SYBER Green Hi-ROX qPCR kit (Radiant). Data were collected in real time during the elongation step of each cycle, using the Quant Studio 3 PCR System (ThermoFisher Scientific). Each cDNA sample was analyzed in triplicate. The relative quantification was performed using the ΔΔCt method. ## 2.7. Statistical Analysis The bioinformatics analysis of the microarray data included their statistical analysis. In DAVID, Fisher’s exact test is adopted to measure gene enrichment in annotation terms. Fisher’s Exact p-values are computed by summing probabilities p over defined sets of tables (Prob = ∑Ap) [20]. In the STRING database, the PPI enrichment p-value indicates that the nodes are not random and that the observed number of edges is significant; for the associated-KEGG pathways, the false discovery rate (FDR) is defined as FDR = E (V/R\|R > 0) P (R > 0) [25]. In Cytoscape-MCODE, the complex score is defined as the product of the complex subgraph, C = (V, E), density, and the number of vertices in the complex subgraph (DC × \|V\|) [23]. For clinical and histological variables and IHC and RT-qPCR measures, statistical analysis was made in SPSS statistics v22 software (SPSS Science Inc., Chicago, IL, USA). The Shapiro–Wilk and Kolmogorov–Smirnov tests were used to determine the data normality. Measures of central tendency and dispersion were estimated for each variable, and a Mann–Whitney U test was used to compare the effect of PE in the CIA-Ex group compared with the CIA-NoEx control group. Differences were considered significant when p ≤ 0.05. ## 3. Results Treadmill PE differentially expressed 2097 genes (1126 up/971 down) in the CIA-Ex group if compared to the CIA-NoEx. The bioinformatics analysis in the DAVID platform showed that the DEGs were significantly associated with 27 KEGG pathways (four up/23 down). The down-expressed KEGG pathways included those related to immune/inflammatory processes, such as Th1 and Th2 cell differentiation, Th17 cell differentiation, cytokine-cytokine receptor interaction and osteoclast differentiation (Figure 2a and Table S1). The up-expressed KEGG pathways included metabolic pathways and the hypoxia-inducible factor (HIF)-1 signaling pathway (Figure 3a and Table S1). The analyses in the STRING database and Cytoscape-MCODE software resulted in the PPI networks shown in Figure 2b and Figure 3b. The PPI from down-regulated genes resulted in 179 nodes and 572 edges, while from up-regulated genes in 207 nodes and 761 edges. PE down-regulated genes of several KEGG pathways related to immune/inflammatory processes, including cytokine–cytokine receptor interaction, Th1 and Th2 cell differentiation, phosphoinositide 3-kinase (PI3K)-Akt, Th17 cell differentiation, osteoclast differentiation, JAK-STAT, chemokines, mitogen-activated protein kinase (MAPK), nuclear factor (NF-)kappa B, TNF, TGF β, Toll-like receptor, and RA signaling pathways (Figure 2c). This PPI network highlighted the down-expression of the genes Tnf, Tgfb1, Il2, Il10, ll2a, and Il23 (Figure 2d). On the other hand, the up-expressed genes were only associated with the RA-related KEGG pathways of HIF-1, chemokine, and PI3K-Akt (Figure 3c,d). The effect of PE was also evaluated on clinical and histological arthritis (Figure 4). In the CIA-Ex group, the progression of swelling and erythema was lower than in the control group, reaching statistical differences at week 4 (Figure 4a,b). In addition, the histological analysis at the end of the experiment showed that PE significantly decreased inflammatory infiltrate and synovial hyperplasia (Figure 4c,d). The expression of inflammatory cytokines was analyzed by RT-qPCR and IHC (Figure 5). Compared to the CIA-NoEx control group, the relative expression of Tnf, Il2, Il10, Il12a, ILl23a, and Tgfb1 was significantly lower in the CIA-Ex group (Figure 5a). Additionally, the expression of TNF-α protein was significantly lower in the hind paws of the CIA-Ex group (Figure 5b,c). ## 4. Discussion Our current understanding of RA pathogenesis states that environmental factors induce specific post-translational modifications in genetically susceptible individuals, which trigger the pathological activation of the immune system and lead to disease onset. Consequently, research directed at the preclinical phases of RA is increasing since it is considered an opportune timeframe for preventive interventions, including pharmacological approaches and potential lifestyle modifications. PE is a proven and widely recommended strategy to preserve the health of the general population and patients with RA. Under physiological conditions, PE induces several intracellular signaling pathways that result in cellular endurance and adaptation in the musculoskeletal system [26,27]. Moreover, PE also influences the regulation of innate immunity and inflammation [28,29]. The precise role of PE in the joints of patients with RA is rarely reported in humans due to the unavailability of joint tissue. Most studies in animal models of RA dealing with PE focus on the differential effect of diverse types and doses of PE [30,31,32,33], while its impact on the preclinical phases has scarcely been explored. Animal models of RA demonstrate that PE influences joint pathological processes once arthritis is established; however, its effects are not conclusive [15]. Although positive outcomes of PE have been described, including the decrease in arthritis severity [32,34] and the down-expression of inflammatory genes and pathways [16], the exacerbation of joint damage by PE has also been reported [17,31,35]. This relation between PE and the influence in the arthritis course may differ in the pre-arthritic stage. In the present study, the low-intensity PE intervention began on the day of the second collagen injection, when joint inflammation was practically absent in the mice. The clinical evolution was significantly milder in the exercise group, and histologically less joint damage was also found at the end of the experiment. Microarray transcriptomic analysis revealed that PE deregulated genes and signaling pathways crucial to the pathogenesis of RA. Moreover, the down-expression of TNF-α was also confirmed by RT-qPCR and IHC. TNF-α is considered the major inflammatory cytokine in the pathogenesis of RA, and a successful target for biological therapy. TNF-α activates synovial fibroblasts, promotes epidermal hyperplasia, and recruits inflammatory cells [36]. The down-regulation of the Jak-STAT pathway demonstrated in our microarray was also a relevant finding because JAK/STAT plays a crucial role in the inflammatory process of the synovium and bone destruction in RA. Therefore, JAK inhibitors are currently used in patients with moderate to severe RA [37]. Interestingly, a recent study revealed that JAK-STAT signaling is up-regulated in high-risk individuals that developed joint inflammation since the pre-arthritic stage [38]. We also confirmed that low-intensity PE down-expressed genes (Il2, Il10, Il12a, Il23a, and Tgfb1) and signaling pathways (cytokine–cytokine receptor interaction, Th1 and Th2 cell differentiation, Th17 cell differentiation, and chemokine signaling pathway) that are identified in the phases of maturation and amplification of the systemic immune response that predates RA [2]. The reduction of several cytokines could be due to the overall decrease in the inflammatory process and, therefore, to a reduced involvement of active immune cells. In many cases, however, TNF-α and IL-10 are antagonists since the increased production of TNF-α increases the production of IL-10, creating a negative-feed loop that results in the subsequent reduction of TNF-α [39,40,41]; IL-10 is considered as an anti-inflammatory cytokine, yet that could be an oversimplification [42]. However, several interactions influence the production of IL-10 aside from TNF-α, and most relate to cytokines involved in the Th1/Th2 balance, such as IFN-γ. Under several pathologic scenarios, it has been shown that IL-10 and TNF-α concentrations fluctuate in the same direction [43,44,45] as we observed in our experiment. Interestingly, in most published articles, the effect of PE on IL-10 production shows an upregulation, and therefore PE is considered an anti-inflammatory intervention [46,47,48,49]. Nevertheless, some studies show no significant change [50,51], and in some other cases, as we found in our experiment, a decrease in IL-10 levels after PE has also been described [52,53,54,55,56]. The context in which this influence is explored and the timing after the PE could explain some of this variability. Regarding TGF-β, a key cytokine for tissue remodeling, given the context PE can either increase [57,58,59] or decrease [60,61,62] its production. Contrary to our findings, Fujii Y. et al. [ 34], who evaluated the impact of treadmill PE in pre-arthritic and established CIA in DA rats, found that only in the established arthritis did PE inhibit joint destruction, improve bone morphometry and reduce connexin (Cx)43 and TNF-α expression in the synovial membranes. Additionally, Cambré, I. et al. [ 35], who applied voluntary wheel running one day after the first collagen injection as well as one day after the booster (day 22) in C57BL/6 mice with CIA, found that exercised mice had a marked accelerated onset and significantly more inflammation in the forefoot of the hind paw. Moreover, their microarray analysis of the Achilles tendon showed that voluntary running increased the expression of proinflammatory genes such as Ccr2, Ccl2, Cxcl1, Mmp3, Lira6, Icam1, Ctsk, and Nfatc1. The above suggests that, as in the studies evaluating PE in arthritis-induced rodent models, when the PE is applied in the pre-arthritic phases the findings are heterogeneous; its effects possibly depends on the type and intensity of the PE. The intensity of the PE is a parameter that directly affects its physiological responses, which has been demonstrated in several studies in animal models. Toti et al. [ 63] investigated the effect of two different exercise protocols (at $60\%$ or $90\%$ of the maximal running velocity) on the fiber composition and metabolism of two specific muscles of healthy mice. They demonstrated that high-intensity exercise, in addition to metabolic changes consisting of a decrease in blood lactate and body weight, induces an increase in the mitochondrial enzymes and slow fibers. Moreover, this same group of researchers also found that the intensity of the exercise differentially affects the increase in the size of the adrenal gland; they reported a rise of $31.04\%$ for mice that underwent high-intensity PE and $10.08\%$ for mice that underwent low-intensity exercise, and this appeared to be the result of an increase in the area of both the adrenal cortex and adrenal medulla [64]. The application of low-intensity exercise in murine models of inflammatory arthritis, including RA and SpA, once the arthritis is established, has shown chiefly positive effects, whether applying aerobic exercise on a treadmill or in wheel running. Positive effects include the decreased expression of inflammatory cytokines (TNF-α, Cx43 [32,34], Il2rb, Il2ra, Il4, Il5, Il3, Cxcl9, and Cxcl12), the infra-expression of inflammatory signaling pathways (chemokine, cytokine-cytokine receptor interaction, complement, coagulation cascade, PI3K-Akt, Jak-STAT, and TLR [16]), the down-expression of markers of pain and disability such as the calcitonin gene-related peptide (CGRP) [65], and the decreased concentration of oxidative stress markers such as thiobarbituric acid reactive substances (TBARS) [66]. Nevertheless, this positive effect of low-intensity exercise cannot be generalized. In our experience, the application of the same low-intensity PE routine to two different rodent models of arthritis resulted in opposite effects. While in proteoglycan-induced arthritis (PGIA) in mice low-intensity PE resulted in a beneficial effect, including less joint destruction [16], in adjuvant-induced arthritis (AIA) in rats the same PE routine exacerbated the destruction process through the over-expression of hypoxia and oxidative stress processes [17]. Similarly, in the experiments carried out by Cambré et al. [ 31,35] evaluating the effect of voluntary wheel running PE in different models of RA and SpA, it was shown that PE accelerates the arthritis onset and increases its severity at the clinical, histological, and molecular level. The above highlights that, in addition to the intensity of the PE, the type of inflammatory arthritis is also a defining parameter in its effects. High-intensity exercise, on the other hand, is often associated with an increased risk of joint damage or exacerbation of inflammation that leads to osteoarthritis (OA) [67,68]. However, reports of high-intensity PE in models of inflammatory arthritis (RA and SpA) are practically nil, most likely because the level of joint inflammation (as shown in Figure 4b, below left) prevents the mice from doing this PE without compromising their well-being. In the present study, low-intensity PE resulted in a protective effect for the arthritis onset and a decrease in its severity; however, according to research carried out in the clinical phases of arthritis, as well as in models of other diseases or health conditions, this effect could not be generalized for moderate- or high-intensity PE. Ideally, these findings should be considered in prescribing PE for at-risk individuals since, considering the heterogeneity of the genetic and clinical presentation of the human disease, as well as the extent of types and intensities of PE, its effects on joint biology cannot be oversimplified. Finally, we recognize that the design of our study could have been improved by the inclusion of healthy mice that performed the same PE routine, as an additional control. However, we defined the scope of our study as evaluating the effect PE had when applied in arthritis’ preclinical phase, and decided to include only two groups for comparison, both susceptible to developing arthritis. The reason behind this decision was the complexity in the comparison of more than two groups because, in the type of microarrays that we use, only the comparison of two groups is allowed to obtain the list of DEGs; concluding more than one comparison would limit our possibility to derive conclusions. In addition, we considered that the effect of PE in healthy mice had been extensively explored, and we wanted to understand the differences between the pre-arthritic stage selectively and once the inflammation sets in. We recognize that setting our experiment in these two groups can be considered a limitation of our investigation. ## 5. Conclusions The present study explored the effect of low-intensity PE when applied in the pre-arthritic stage in CIA, demonstrating that, on the joints, PE reduced the severity of clinical and histological arthritis and down-expressed RA-related genes and signaling pathways, suggesting a beneficial effect. However, more research is needed on the types of exercise that can be implemented as a preventative strategy for RA in high-risk individuals. ## References 1. Smolen J.S., Aletaha D., Barton A., Burmester G.R., Emery P., Firestein G.S., Kavanaugh A., McInnes I.B., Solomon D.H., Strand V.. Rheumatoid Arthritis. Nat. Rev. Dis. Primer (2018) 4 18001. DOI: 10.1038/nrdp.2018.1 2. Romão V.C., Fonseca J.E.. Disease Mechanisms in Preclinical Rheumatoid Arthritis: A Narrative Review. Front. Med. (2022) 9 689711. DOI: 10.3389/fmed.2022.689711 3. Romão V.C., Fonseca J.E.. Etiology and Risk Factors for Rheumatoid Arthritis: A State-of-the-Art Review. Front. Med. (2021) 8 689698. DOI: 10.3389/fmed.2021.689698 4. Holers V.M., Kuhn K.A., Demoruelle M.K., Norris J.M., Firestein G.S., James E.A., Robinson W.H., Buckner J.H., Deane K.D.. Mechanism-driven Strategies for Prevention of Rheumatoid Arthritis. Rheumatol. Autoimmun. (2022) 2 109-119. DOI: 10.1002/rai2.12043 5. Alpizar-Rodriguez D., Finckh A.. Is the Prevention of Rheumatoid Arthritis Possible?. Clin. Rheumatol. (2020) 39 1383-1389. DOI: 10.1007/s10067-020-04927-6 6. Deane K.D., Holers V.M.. Rheumatoid Arthritis Pathogenesis, Prediction, and Prevention: An Emerging Paradigm Shift. Arthritis Rheumatol. (2021) 73 181-193. DOI: 10.1002/art.41417 7. Frazzei G., Musters A., de Vries N., Tas S.W., van Vollenhoven R.F.. Prevention of Rheumatoid Arthritis: A Systematic Literature Review of Preventive Strategies in at-Risk Individuals. Autoimmun. Rev. (2023) 22 103217. DOI: 10.1016/j.autrev.2022.103217 8. Sun L., Zhu J., Ling Y., Mi S., Li Y., Wang T., Li Y.. Physical Activity and the Risk of Rheumatoid Arthritis: Evidence from Meta-Analysis and Mendelian Randomization. Int. J. Epidemiol. (2021) 50 1593-1603. DOI: 10.1093/ije/dyab052 9. Gwinnutt J.M., Wieczorek M., Cavalli G., Balanescu A., Bischoff-Ferrari H.A., Boonen A., de Souza S., de Thurah A., Dorner T.E., Moe R.H.. Effects of Physical Exercise and Body Weight on Disease-Specific Outcomes of People with Rheumatic and Musculoskeletal Diseases (RMDs): Systematic Reviews and Meta-Analyses Informing the 2021 EULAR Recommendations for Lifestyle Improvements in People with RMDs. RMD Open (2022) 8 e002168. DOI: 10.1136/rmdopen-2021-002168 10. Chehade L., Jaafar Z.A., El Masri D., Zmerly H., Kreidieh D., Tannir H., Itani L., El Ghoch M.. Lifestyle Modification in Rheumatoid Arthritis: Dietary and Physical Activity Recommendations Based on Evidence. Curr. Rheumatol. Rev. (2019) 15 209-214. DOI: 10.2174/1573397115666190121135940 11. Rausch Osthoff A.-K., Niedermann K., Braun J., Adams J., Brodin N., Dagfinrud H., Duruoz T., Esbensen B.A., Günther K.-P., Hurkmans E.. 2018 EULAR Recommendations for Physical Activity in People with Inflammatory Arthritis and Osteoarthritis. Ann. Rheum. Dis. (2018) 77 1251-1260. DOI: 10.1136/annrheumdis-2018-213585 12. Zaccardelli A., Sparks J.A.. Challenges and Opportunities of Targeted Behavioral Interventions for Groups at Risk for Developing Rheumatoid Arthritis. Healthcare (2021) 9. DOI: 10.3390/healthcare9060641 13. Hughes-Austin J.M., Ix J.H., Ward S.R., Weisman M.H., ODell J.R., Mikuls T.R., Buckner J.H., Gregersen P.K., Keating R.M., Demoruelle M.K.. Evaluating Associations of Joint Swelling, Joint Stiffness and Joint Pain with Physical Activity in First-Degree Relatives of Patients with Rheumatoid Arthritis: Studies of the Aetiology of Rheumatoid Arthritis (SERA), a Prospective Cohort Study. BMJ Open (2021) 11 e050883. DOI: 10.1136/bmjopen-2021-050883 14. Metsios G.S., Brodin N., Vlieland T.P.M.V., Van den Ende C.H.M., Stavropoulos-Kalinoglou A., Fatouros I., van der Esch M., Fenton S.A.M., Tzika K., Moe R.H.. Position Statement on Exercise Dosage in Rheumatic and Musculoskeletal Diseases: The Role of the IMPACT-RMD Toolkit. Mediterr. J. Rheumatol. (2021) 32 378-385. DOI: 10.31138/mjr.32.4.378 15. González-Chávez S.A., Pacheco-Tena C.. Exercise-Driven Exacerbation of Inflammation: Contribution of Animal Models of Rheumatoid Arthritis and Spondyloarthritis. Connect. Tissue Res. (2022) 63 425-442. DOI: 10.1080/03008207.2022.2036734 16. González-Chávez S.A., Pacheco-Tena C., Quiñonez-Flores C.M., Espino-Solis G.P., Burrola-De Anda J.I., Muñoz-Morales P.M.. Positive Transcriptional Response on Inflammation and Joint Remodelling Influenced by Physical Exercise in Proteoglycan-Induced Arthritis: An Animal Study. Bone Jt. Res. (2020) 9 36-48. DOI: 10.1302/2046-3758.91.BJR-2019-0055.R2 17. González-Chávez S.A., Quiñonez-Flores C.M., Espino-Solís G.P., Vázquez-Contreras J.Á., Pacheco-Tena C.. Exercise Exacerbates the Transcriptional Profile of Hypoxia, Oxidative Stress and Inflammation in Rats with Adjuvant-Induced Arthritis. Cells (2019) 8. DOI: 10.3390/cells8121493 18. Brand D.D., Latham K.A., Rosloniec E.F.. Collagen-Induced Arthritis. Nat. Protoc. (2007) 2 1269-1275. DOI: 10.1038/nprot.2007.173 19. Ramírez Salcedo J., Chávez González L., Santillán Torres J.L., Guzmán León S.. Microarreglos de ADN: Fabricación, Proceso y Análisis. Herramientas Moleculares Aplicadas en Ecología: Aspectos Teóricos y Prácticos (2014) 203-229 20. Huang D.W., Sherman B.T., Tan Q., Kir J., Liu D., Bryant D., Guo Y., Stephens R., Baseler M.W., Lane H.C.. DAVID Bioinformatics Resources: Expanded Annotation Database and Novel Algorithms to Better Extract Biology from Large Gene Lists. Nucleic Acids Res. (2007) 35 W169-W175. DOI: 10.1093/nar/gkm415 21. Szklarczyk D., Morris J.H., Cook H., Kuhn M., Wyder S., Simonovic M., Santos A., Doncheva N.T., Roth A., Bork P.. The STRING Database in 2017: Quality-Controlled Protein–Protein Association Networks, Made Broadly Accessible. Nucleic Acids Res. (2017) 45 D362-D368. DOI: 10.1093/nar/gkw937 22. Isserlin R., Merico D., Voisin V., Bader G.D.. Enrichment Map—A Cytoscape App to Visualize and Explore OMICs Pathway Enrichment Results. F1000Research (2014) 3 141. DOI: 10.12688/f1000research.4536.1 23. Bader G.D., Hogue C.W.V.. An Automated Method for Finding Molecular Complexes in Large Protein Interaction Networks. BMC Bioinform. (2003) 4. DOI: 10.1186/1471-2105-4-2 24. Quiñonez-Flores C.M., López-Loeza S.M., Pacheco-Tena C., Muñoz-Morales P.M., Acosta-Jiménez S., González-Chávez S.A.. Stability of Housekeeping Genes in Inflamed Joints of Spontaneous and Collagen-Induced Arthritis in DBA/1 Mice. Inflamm. Res. (2021) 70 619-632. DOI: 10.1007/s00011-021-01453-2 25. Szklarczyk D., Gable A.L., Nastou K.C., Lyon D., Kirsch R., Pyysalo S., Doncheva N.T., Legeay M., Fang T., Bork P.. The STRING Database in 2021: Customizable Protein–Protein Networks, and Functional Characterization of User-Uploaded Gene/Measurement Sets. Nucleic Acids Res. (2021) 49 D605-D612. DOI: 10.1093/nar/gkaa1074 26. Li J., Li Y., Atakan M.M., Kuang J., Hu Y., Bishop D.J., Yan X.. The Molecular Adaptive Responses of Skeletal Muscle to High-Intensity Exercise/Training and Hypoxia. Antioxidants (2020) 9. DOI: 10.3390/antiox9080656 27. Solsona R., Pavlin L., Bernardi H., Sanchez A.M.. Molecular Regulation of Skeletal Muscle Growth and Organelle Biosynthesis: Practical Recommendations for Exercise Training. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22052741 28. Woods J.A., Vieira V.J., Keylock K.T.. Exercise, Inflammation, and Innate Immunity. Immunol. Allergy Clin. N. Am. (2009) 29 381-393. DOI: 10.1016/j.iac.2009.02.011 29. Domin R., Dadej D., Pytka M., Zybek-Kocik A., Ruchała M., Guzik P.. Effect of Various Exercise Regimens on Selected Exercise-Induced Cytokines in Healthy People. Int. J. Environ. Res. Public Health (2021) 18. DOI: 10.3390/ijerph18031261 30. Damasceno de Lima R., Pedersen M., Costa do Bomfim F.R., Chiarotto G.B., Canciglieri P.H., Pauli J.R., Felonato M.. Effects of Different Physical Training Protocols on Inflammatory Markers in Zymosan-induced Rheumatoid Arthritis in Wistar Rats. Cell Biochem. Funct. (2022) 40 321-332. DOI: 10.1002/cbf.3697 31. Cambré I., Gaublomme D., Schryvers N., Lambrecht S., Lories R., Venken K., Elewaut D.. Running Promotes Chronicity of Arthritis by Local Modulation of Complement Activators and Impairing T Regulatory Feedback Loops. Ann. Rheum. Dis. (2019) 78 787-795. DOI: 10.1136/annrheumdis-2018-214627 32. Shimomura S., Inoue H., Arai Y., Nakagawa S., Fujii Y., Kishida T., Ichimaru S., Tsuchida S., Shirai T., Ikoma K.. Treadmill Running Ameliorates Destruction of Articular Cartilage and Subchondral Bone, Not Only Synovitis, in a Rheumatoid Arthritis Rat Model. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19061653 33. Kito T., Teranishi T., Nishii K., Sakai K., Matsubara M., Yamada K.. Effectiveness of Exercise-Induced Cytokines in Alleviating Arthritis Symptoms in Arthritis Model Mice. Okajimas Folia Anat. Jpn. (2016) 93 81-88. DOI: 10.2535/ofaj.93.81 34. Fujii Y., Inoue H., Arai Y., Shimomura S., Nakagawa S., Kishida T., Tsuchida S., Kamada Y., Kaihara K., Shirai T.. Treadmill Running in Established Phase Arthritis Inhibits Joint Destruction in Rat Rheumatoid Arthritis Models. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20205100 35. Cambré I., Gaublomme D., Burssens A., Jacques P., Schryvers N., De Muynck A., Meuris L., Lambrecht S., Carter S., de Bleser P.. Mechanical Strain Determines the Site-Specific Localization of Inflammation and Tissue Damage in Arthritis. Nat. Commun. (2018) 9 4613. DOI: 10.1038/s41467-018-06933-4 36. Jang D., Lee A.-H., Shin H.-Y., Song H.-R., Park J.-H., Kang T.-B., Lee S.-R., Yang S.-H.. The Role of Tumor Necrosis Factor Alpha (TNF-α) in Autoimmune Disease and Current TNF-α Inhibitors in Therapeutics. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22052719 37. Hu L., Liu R., Zhang L.. Advance in Bone Destruction Participated by JAK/STAT in Rheumatoid Arthritis and Therapeutic Effect of JAK/STAT Inhibitors. Int. Immunopharmacol. (2022) 111 109095. DOI: 10.1016/j.intimp.2022.109095 38. O’Neil L.J., Meng X., Mcfadyen C., Fritzler M.J., El-Gabalawy H.S.. Serum Proteomic Networks Associate with Pre-Clinical Rheumatoid Arthritis Autoantibodies and Longitudinal Outcomes. Front. Immunol. (2022) 13 958145. DOI: 10.3389/fimmu.2022.958145 39. Asadullah K., Sterry W., Stephanek K., Jasulaitis D., Leupold M., Audring H., Volk H.D., Döcke W.D.. IL-10 Is a Key Cytokine in Psoriasis. Proof of Principle by IL-10 Therapy: A New Therapeutic Approach. J. Clin. Investig. (1998) 101 783-794. DOI: 10.1172/JCI1476 40. Pestka S., Krause C.D., Sarkar D., Walter M.R., Shi Y., Fisher P.B.. Interleukin-10 and Related Cytokines and Receptors. Annu. Rev. Immunol. (2004) 22 929-979. DOI: 10.1146/annurev.immunol.22.012703.104622 41. Saraiva M., O’Garra A.. The Regulation of IL-10 Production by Immune Cells. Nat. Rev. Immunol. (2010) 10 170-181. DOI: 10.1038/nri2711 42. Rasquinha M.T., Sur M., Lasrado N., Reddy J.. IL-10 as a Th2 Cytokine: Differences Between Mice and Humans. J. Immunol. (2021) 207 2205-2215. DOI: 10.4049/jimmunol.2100565 43. Lu Q., Zhu Z., Tan C., Zhou H., Hu Y., Shen G., Zhu P., Yang G., Xie X.. Changes of Serum IL-10, IL-1β, IL-6, MCP-1, TNF-α, IP-10 and IL-4 in COVID-19 Patients. Int. J. Clin. Pract. (2021) 75 e14462. DOI: 10.1111/ijcp.14462 44. Dissanayake K., Jayasinghe C., Wanigasekara P., Sominanda A.. Potential Applicability of Cytokines as Biomarkers of Disease Activity in Rheumatoid Arthritis: Enzyme-Linked Immunosorbent Spot Assay-Based Evaluation of TNF-α, IL-1β, IL-10 and IL-17A. PLoS ONE (2021) 16. DOI: 10.1371/journal.pone.0246111 45. Lau S.Y., Guild S.-J., Barrett C.J., Chen Q., McCowan L., Jordan V., Chamley L.W.. Tumor Necrosis Factor-Alpha, Interleukin-6, and Interleukin-10 Levels Are Altered in Preeclampsia: A Systematic Review and Meta-Analysis. Am. J. Reprod. Immunol. (2013) 70 412-427. DOI: 10.1111/aji.12138 46. Petersen A.M.W., Pedersen B.K.. The Anti-Inflammatory Effect of Exercise. J. Appl. Physiol. (2005) 98 1154-1162. DOI: 10.1152/japplphysiol.00164.2004 47. Abd El-Kader S.M., Al-Jiffri O.H.. Aerobic Exercise Modulates Cytokine Profile and Sleep Quality in Elderly. Afr. Health Sci. (2019) 19 2198-2207. DOI: 10.4314/ahs.v19i2.45 48. Shaw D.M., Merien F., Braakhuis A., Dulson D.. T-Cells and Their Cytokine Production: The Anti-Inflammatory and Immunosuppressive Effects of Strenuous Exercise. Cytokine (2018) 104 136-142. DOI: 10.1016/j.cyto.2017.10.001 49. Islam H., Neudorf H., Mui A.L., Little J.P.. Interpreting “anti-Inflammatory” Cytokine Responses to Exercise: Focus on Interleukin-10. J. Physiol. (2021) 599 5163-5177. DOI: 10.1113/JP281356 50. Bartlett D.B., Willis L.H., Slentz C.A., Hoselton A., Kelly L., Huebner J.L., Kraus V.B., Moss J., Muehlbauer M.J., Spielmann G.. Ten Weeks of High-Intensity Interval Walk Training Is Associated with Reduced Disease Activity and Improved Innate Immune Function in Older Adults with Rheumatoid Arthritis: A Pilot Study. Arthritis Res. Ther. (2018) 20 127. DOI: 10.1186/s13075-018-1624-x 51. Monteiro M.R.P., Aragão-Santos J.C., Vasconcelos A.B.S., Resende-Neto A.G.D., Chaves L.M.d.S., Cardoso A.P., Nogueira A.C., Carnero-Diaz A., Marcos-Pardo P.J., Corrêa C.B.. Bodyweight and Combined Training Reduce Chronic Low-Grade Inflammation and Improve Functional Fitness of Postmenopausal Women. Sports (2022) 10. DOI: 10.3390/sports10100143 52. Zhou Y., Jia N., Ding M., Yuan K.. Effects of Exercise on Inflammatory Factors and IGF System in Breast Cancer Survivors: A Meta-Analysis. BMC Women’s Health (2022) 22. DOI: 10.1186/s12905-022-02058-5 53. Balogh L., Szabó K., Pucsok J.M., Jámbor I., Gyetvai Á., Mile M., Barna L., Szodoray P., Tarr T., Csiki Z.. The Effect of Aerobic Exercise and Low-Impact Pilates Workout on the Adaptive Immune System. J. Clin. Med. (2022) 11. DOI: 10.3390/jcm11226814 54. Guo Y.-T., Peng Y.-C., Yen H.-Y., Wu J.-C., Hou W.-H.. Effects of Probiotic Supplementation on Immune and Inflammatory Markers in Athletes: A Meta-Analysis of Randomized Clinical Trials. Medicina (2022) 58. DOI: 10.3390/medicina58091188 55. Kjølhede T., Dalgas U., Gade A.B., Bjerre M., Stenager E., Petersen T., Vissing K.. Acute and Chronic Cytokine Responses to Resistance Exercise and Training in People with Multiple Sclerosis. Scand. J. Med. Sci. Sport. (2016) 26 824-834. DOI: 10.1111/sms.12504 56. Kim S.-D., Yeun Y.-R.. Effects of Resistance Training on C-Reactive Protein and Inflammatory Cytokines in Elderly Adults: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Int. J. Environ. Res. Public Health (2022) 19. DOI: 10.3390/ijerph19063434 57. Takahashi H., Alves C.R.R., Stanford K.I., Middelbeek R.J.W., Nigro P., Ryan R.E., Xue R., Sakaguchi M., Lynes M.D., So K.. TGF-Β2 Is an Exercise-Induced Adipokine That Regulates Glucose and Fatty Acid Metabolism. Nat. Metab. (2019) 1 291-303. DOI: 10.1038/s42255-018-0030-7 58. Sun X.J., Feng W.L., Hou N., Li N., Jiang W.L.. Effects of aerobic exercise and resveratrol on the expressions of JAK2 and TGF-β1 in renal tissue of type 2 diabetes rats. Chin. J. Appl. Physiol. (2020) 36 202-206. DOI: 10.12047/j.cjap.5918.2020.045 59. Heinemeier K., Langberg H., Olesen J.L., Kjaer M.. Role of TGF-Beta1 in Relation to Exercise-Induced Type I Collagen Synthesis in Human Tendinous Tissue. J. Appl. Physiol. (2003) 95 2390-2397. DOI: 10.1152/japplphysiol.00403.2003 60. Ma Y., Kuang Y., Bo W., Liang Q., Zhu W., Cai M., Tian Z.. Exercise Training Alleviates Cardiac Fibrosis through Increasing Fibroblast Growth Factor 21 and Regulating TGF-Β1-Smad2/3-MMP2/9 Signaling in Mice with Myocardial Infarction. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222212341 61. Eka Widiastuti I.A., Arsyad A., Idris I., Patellongi I., Kadriyan H., Buanayuda G.W., Sari D.P., Rosyidi R.M.. Exercise Adaptations and TGF-Β1 Levels in Recreational Cyclists. Ann. Med. Surg. (2021) 70 102872. DOI: 10.1016/j.amsu.2021.102872 62. Silva V.R.R., Lenhare L., Katashima C.K., Morari J., M-Assis A., Gaspar R.S., da Silva A.S.R., Moura L.P., Pauli J.R., Cintra D.E.. TGF-Β1 Downregulation in the Hypothalamus of Obese Mice through Acute Exercise. J. Cell. Biochem. (2019) 120 18186-18192. DOI: 10.1002/jcb.29124 63. Toti L., Bartalucci A., Ferrucci M., Fulceri F., Lazzeri G., Lenzi P., Soldani P., Gobbi P., La Torre A., Gesi M.. High-Intensity Exercise Training Induces Morphological and Biochemical Changes in Skeletal Muscles. Biol. Sport (2013) 30 301-309. DOI: 10.5604/20831862.1077557 64. Bartalucci A., Ferrucci M., Fulceri F., Lazzeri G., Lenzi P., Toti L., Serpiello F.R., La Torre A., Gesi M.. High-Intensity Exercise Training Produces Morphological and Biochemical Changes in Adrenal Gland of Mice. Histol. Histopathol. (2012) 27 753-769. DOI: 10.14670/HH-27.753 65. Tong Y., Ishikawa K., Sasaki R., Takeshita I., Sakamoto J., Okita M.. The Effects of Wheel-Running Using the Upper Limbs Following Immobilization after Inducing Arthritis in the Knees of Rats. Physiol. Res. (2021) 70 79-87. DOI: 10.33549/physiolres.934469 66. Dos Santos S.A., Dos Santos Vieira M.A., Simões M.C.B., Serra A.J., Leal-Junior E.C., de Carvalho P.. de T.C. Photobiomodulation Therapy Associated with Treadmill Training in the Oxidative Stress in a Collagen-Induced Arthritis Model. Lasers Med. Sci. (2017) 32 1071-1079. DOI: 10.1007/s10103-017-2209-7 67. Ni G.-X., Liu S.-Y., Lei L., Li Z., Zhou Y.-Z., Zhan L.-Q.. Intensity-Dependent Effect of Treadmill Running on Knee Articular Cartilage in a Rat Model. BioMed Res. Int. (2013) 2013. DOI: 10.1155/2013/172392 68. Ni G.-X., Lei L., Zhou Y.-Z.. Intensity-Dependent Effect of Treadmill Running on Lubricin Metabolism of Rat Articular Cartilage. Arthritis Res. Ther. (2012) 14 R256. DOI: 10.1186/ar4101
--- title: Hepatocyte Toll-Like Receptor 4 Mediates Alcohol-Induced Insulin Resistance in Mice authors: - Piumi B. Wickramasinghe - Shuwen Qian - Lyndsey E. Langley - Chen Liu - Lin Jia journal: Biomolecules year: 2023 pmcid: PMC10046504 doi: 10.3390/biom13030454 license: CC BY 4.0 --- # Hepatocyte Toll-Like Receptor 4 Mediates Alcohol-Induced Insulin Resistance in Mice ## Abstract Accumulating evidence has demonstrated the association between alcohol overconsumption and the development of insulin resistance. However, the underlying mechanisms are not completely understood. To investigate the requirement and sufficiency of hepatocyte toll-like receptor 4 (TLR4) in alcohol-induced insulin resistance, we used two mouse models (Tlr4fl/fl and Tlr4LoxTB) that allow ablation of TLR4 only in hepatocytes (Tlr4LKO) and restoration of endogenous TLR4 expression in hepatocytes on a TLR4-null background (Tlr4LoxTB × Alb-Cre), respectively. A Lieber-DeCarli feeding model was used to induce glucose intolerance and insulin resistance in mice. Glucose tolerance test, insulin tolerance test, and insulin signaling experiments were performed to examine systemic and tissue-specific insulin sensitivity. We found that alcohol-fed hepatocyte TLR4 deficient mice (Tlr4LKO) had lower blood glucose levels in response to intraperitoneal injection of insulin. Moreover, increased phosphorylation of glycogen synthase kinase-3β (GSK3β) was observed in the liver of Tlr4LKO mice after chronic alcohol intake. In contrast, when hepatic TLR4 was reactivated in mice (Tlr4LoxTB × Alb-Cre), alcohol feeding caused glucose intolerance in these mice compared with littermate controls (Tlr4LoxTB). In addition, AKT phosphorylation was dramatically reduced in the liver and epididymal white adipose tissue (eWAT) of alcohol-fed Tlr4LoxTB × Alb-Cre mice, which was similar to that of mice with whole-body TLR4 reactivation (Tlr4LoxTB × Zp3-Cre). Collectively, these findings suggest that hepatocyte TLR4 is both required and sufficient in the development of insulin resistance induced by alcohol overconsumption. ## 1. Introduction Excessive alcohol intake has become a growing public health concern worldwide, which causes detrimental effects on various organs. Increasing lines of evidence indicate an association between heavy drinking and the development of insulin resistance in humans [1,2,3] and rodents [4,5,6,7,8]. Specifically, reduced insulin signaling and decreased glucose transport have been reported in various tissues and cell types that were exposed to alcohol, including the liver [7,8,9,10,11], white adipose tissue [12], as well as isolated hepatocytes [13], and adipocytes [14,15]. However, the underlying mechanisms by which alcohol drinking causes insulin resistance and glucose dysregulation are not completely understood. Extensive studies have demonstrated that lipopolysaccharide (LPS) and its cell surface receptor, toll-like receptor 4 (TLR4) play critical roles in the development of alcohol-associated liver disease (ALD) [16,17,18]. In addition, the cell-type-specific role of TLR4 in alcohol-induced fatty liver disease has been reported. Unexpectedly, TLR4-expressing myeloid cells played a minimal role in regulating alcohol-induced fatty liver disease and adipose tissue inflammation [19,20]. Interestingly, alcohol-fed hepatocyte TLR4 deficient mice showed significantly reduced hepatic triglyceride content and decreased expression of inflammatory cytokines in the white adipose tissue [19]. However, it is largely unknown whether hepatocyte TLR4 contributes to the development of insulin resistance induced by chronic alcohol drinking. Increasing evidence suggests that hepatocyte TLR4 plays an important role in regulating obesity-associated glucose dysregulation and insulin resistance in mice. For example, Uchimura et al. reported that the knockdown of hepatic TLR4 by siRNA significantly improved insulin resistance in high-fat diet (HFD)-fed mice [21]. Moreover, mice lacking TLR4 selectively in hepatocytes exhibited enhanced systemic insulin sensitivity and reduced hepatic glucose production after HFD feeding [22]. These findings led us to investigate if hepatocyte TLR4 regulates alcohol-induced insulin resistance. In the current study, hepatocyte TLR4 ablated mice and their littermate controls were fed either a control or alcohol-containing liquid diet chronically for 8 weeks. During the diet feeding, systemic and tissue-specific insulin sensitivity was examined. We found that hepatocyte TLR4 is required in mediating insulin resistance following excessive alcohol intake. This is supported by the observation that mice lacking hepatocyte TLR4 showed improved systemic insulin sensitivity and significantly increased protein expression of phosphorylated glycogen synthase kinase-3β (GSK3β) in the liver after chronic alcohol feeding. In addition, we generated mice with endogenous TLR4 restored in hepatocytes on a TLR4-null background. This mouse model allowed us to investigate the sufficiency of hepatocyte TLR4 in alcohol-induced impairment in insulin signaling. Our data showed that reactivation of hepatocyte TLR4 exacerbated alcohol-induced glucose intolerance and decreased insulin signaling. Collectively, these findings indicate that hepatocyte TLR4 plays a role in regulating alcohol-induced insulin resistance in mice. ## 2.1. Animal Care and Chronic Liquid Diet Feeding The generation and validation of Tlr4fl/fl [19,22,23] and Tlr4LoxTB [24,25,26] mice have been previously reported. Albumin-Cre transgenic mice (Alb-Cre, JAX:003574) were crossed with Tlr4fl/fl animals to produce mice lacking TLR4 in hepatocytes (Tlr4LKO) [19,24]. In addition, Alb-Cre and zona pellucida 3 (Zp3)-Cre mice (JAX:003651) were bred with Tlr4LoxTB mice to restore endogenous TLR4 expression in hepatocytes (Tlr4LoxTB × Alb-Cre) and in all tissues (Tlr4LoxTB × Zp3-Cre) [24], respectively. All strains were backcrossed at least six generations to maintain C57BL/6J genetic background. Animals were housed ($$n = 4$$ per cage) in a temperature-controlled environment on a 12 h light/12 h dark cycle with ad libitum access to food and water unless specified otherwise. Here, 10–12-week-old male mice were subjected to a control liquid diet (Bio-Serv, F1259SP) for five days. After the acclimation, mice were fed either control or alcohol-containing liquid diets (Bio-Serv, F1258SP, $5\%$ alcohol (vol/vol)) for 8 weeks. Because mice consumed less of the alcohol-containing liquid diet than the control diet [27], mice on the control liquid diet were pair-fed. We recorded the food intake of the alcohol-fed mice based on the differences between the volume added the day before and the volume left. Then we calculated the average daily volume consumed by alcohol-fed mice. This calculated volume was the amount of control liquid diet given to pair-fed mice [27]. In addition, the dead volume on the bottom of the tubes was considered. An extra 5 mL of control liquid diet was added to the calculated volume to pair-fed mice. On the first day of liquid diet feeding, the volume of control liquid diets was estimated based on the pilot experiments and/or our previous experience. At the end of diet feeding, mice were fasted for 4 h. After deep anesthetization, blood was collected from the inferior vena cava, followed by an infusion of 10 mL PBS to the left ventricle of the heart to remove residual blood. Livers were removed and snap-frozen in liquid nitrogen and stored at −80 °C. Experiments were performed according to protocols reviewed and approved by the Institutional Animal Care and Use Committee of the University of Texas at Dallas (UTD). ## 2.2. Glucose and Insulin Tolerance Tests A glucose tolerance test (GTT) was performed after 6 weeks of liquid diet feeding. The mice were fasted for 5 h. After the measurement of basal blood glucose levels with a glucometer (Contour, Bayer, Parsippany, NJ, USA), a glucose solution (1.2 g/kg body weight (BW)) was given via intraperitoneal injection. Then blood glucose concentration was measured 15, 30, 60, and 120 min after glucose administration. An insulin tolerance test (ITT) was performed after 7 weeks of diet feeding. Food was removed for 2 h in the morning, and blood glucose was measured before and 15, 30, 60, and 120 min after intraperitoneal injection of human insulin (1 unit/kg BW; Humulin R). ## 2.3. Acute Insulin Injection to Determine Tissue-Specific Insulin Sensitivity After 8 weeks of alcohol-containing liquid diet feeding, mice were fasted for 6 h and injected intraperitoneally with either saline or human insulin (5 units/kg BW; Humulin R). 10 min later, liver and epididymal white adipose tissue (eWAT) were rapidly removed and snap-frozen in liquid nitrogen and stored at −80 °C until analysis. ## 2.4. Western Blotting Liver and eWAT (pooled samples from the same genotype and treatment, $$n = 2$$–4) were homogenized in lysis buffer containing $1\%$ NP-40, $1\%$ Triton-X 100, $1\%$ SDS, 5 mM EDTA (pH 8.0), 50 mM Tris– HCl (pH 7.4), and protease inhibitor (P8340, Sigma, St. Louis, MO, USA) and phosphatase inhibitor cocktails (P5726 and P0044, Sigma, St. Louis, MO, USA). Protein concentrations were determined by BCA kit (Pierce, Thermo Scientific™, Waltham, MA, USA). Then, 10–20 µg tissue lysates were separated by $8\%$ gel in SDS-PAGE and transferred to nitrocellulose membranes (Trans-Blot, Bio-Rad, Hercules, CA, USA). Membranes were blocked with $5\%$ nonfat dried milk or $3\%$ BSA (for phosphorylated proteins) for 1 h at room temperature and then incubated with primary antibodies overnight at 4 °C in $5\%$ nonfat dried milk or $3\%$ BSA (for phosphorylated proteins). Primary antibodies were purchased from Cell Signaling Technology, phospho-AKT (Ser473) (#9271), phospho-GSK3β (#8466), total AKT (pan) (#4691), and total GSK3β (#9315). Membranes were washed with TBS containing $0.1\%$ (vol/vol) Tween 20, and incubated with 1:10,000 dilution of goat anti-rabbit horseradish peroxidase antibody (Jackson ImmunoResearch, West Grove, PA, USA) for 1 h at room temperature. Blots were visualized with enhanced chemiluminescence (Bio-Rad, Hercules, CA, USA). The band densities were quantified using Image J software (NIH). ## 2.5. Blood Glucose and Plasma Insulin Levels After 8 weeks of liquid diet feeding, blood was collected from 6 h fasted mice via tail bleeding. Blood glucose was measured using a glucometer (Contour, Bayer, Parsippany, NJ, USA). Plasma insulin concentration was determined using an ELISA (Crystal Chem, Elk Grove Village, IL, USA) according to the manufacturer’s instructions. ## 2.6. Measurement of Liver Triglyceride Contents Frozen liver tissues (60–80 mg) were thawed, minced, and weighed in glass tubes. Lipids were extracted in 3 mL of chloroform/methanol (2:1) at room temperature overnight. After centrifugation, lipid extract was transferred to a clean glass tube, and dilute H2SO4 ($0.05\%$) was added to separate the phases by vortex and centrifugation. The aqueous upper phase was removed, and an aliquot of the bottom phase (100 µL) was transferred to a new glass tube and dried down under N2. Then, 1 mL of $1\%$ Triton X-100 in chloroform was added. After the evaporation of the solvent, deionized water (0.5 mL) was added to each tube and vortexed until the solution was clear. Triglyceride standards (Verichem Laboratories, Providence, RI, USA) was prepared by adding $1\%$ Triton X-100 in chloroform, evaporating, and dissolving in deionized water. Triglyceride content in liver samples was quantified using the Infinity Triglycerides Reagent (Thermo Scientific™, Waltham, MA, USA). ## 2.7. Primary Hepatocyte Isolation and Real-Time PCR Primary hepatocyte was isolated from chow-fed Tlr4LoxTB and Tlr4LoxTB × Alb-Cre mice as described previously [22]. Total RNAs from the hepatocytes were extracted using RNA STAT60 (Tel-Test, Friendswood, TX, USA). Complementary DNA was synthesized using the High Capacity cDNA Kit (Applied Biosystems, Waltham, MA, USA) and qPCR was performed using a Bio-Rad sequence detection system (Bio-Rad). Primers for Tlr4 (forward, CAGCAAAGTCCCTGATGACA and reverse, AGAGGTGGTGTAAGCCATGC) and 18s (forward, ACCGCAGCTAGGAATAATGGA and reverse, GCCTCAGTTCCGAAAACCA) were purchased from Integrated DNA Technologies. The relative amounts of Tlr4 mRNAs were calculated using the ΔΔCT assay. ## 2.8. Statistical Analysis Data are expressed as means ± SEM. Statistical analysis was performed using a t-test when there was only one variable or analysis of variance (two-way ANOVA) with Post hoc Tukey’s multiple comparisons test or Šídák’s multiple comparisons tests when two variables (genotype and treatment) were present (GraphPad Prism, Boston, FL, USA). $p \leq 0.05$ is considered significant. ## 3.1. Alcohol-Fed Hepatocyte TLR4 Deficient Mice Exhibited Improved Systemic Insulin Sensitivity To examine whether hepatocyte TLR4 regulates chronic alcohol-drinking-associated glucose dysregulation and insulin insensitivity, both Tlr4fl/fl and Tlr4LKO mice were fed either a control or $5\%$ alcohol-containing liquid diet chronically for 8 weeks. We found that Tlr4fl/fl and Tlr4LKO mice on the control diet exhibited similar blood glucose and plasma insulin levels after a 6 h fast (Figure 1A,B). Consistent with a previous report [28], excessive alcohol drinking led to reduced blood glucose levels in both genotypes (Figure 1A). However, fasting plasma insulin concentrations were not affected by alcohol feeding or hepatocyte TLR4 deficiency (Figure 1B). To determine the systemic glucose homeostasis and insulin sensitivity in these mice, glucose and insulin tolerance tests (IPGTT and ITT) were performed. After chronic control diet feeding, Tlr4fl/fl and Tlr4LKO mice responded similarly to glucose administration and insulin injection (Figure 1C,F). Chronic alcohol feeding caused glucose intolerance in both Tlr4fl/fl and Tlr4LKO mice (Figure 1D,E). Although alcohol-fed Tlr4fl/fl mice developed systemic insulin resistance (Figure 1G,H), Tlr4LKO mice were insulin sensitive after excessive alcohol drinking as evidenced by lower normalized blood glucose levels after 30 min of insulin injection (Figure 1G), which was comparable to mice pair-fed control liquid diet (Figure 1F,H). The analysis of the area under the curve (AUC) also showed that Tlr4LKO mice had improved systemic insulin sensitivity after chronic alcohol feeding (Figure 1H). ## 3.2. Alcohol-Fed Hepatocyte TLR4 Deficient Mice Showed Improved Insulin Sensitivity in the Liver To examine the tissue-specific insulin sensitivity, Tlr4fl/fl and Tlr4LKO mice were fed an alcohol-containing liquid diet chronically for 8 weeks. Then saline or insulin was injected intraperitoneally after a 6 h fast. Liver whole cell lysates were prepared for the expression of proteins involved in the insulin signaling pathway. Regardless of the genotypes, insulin administration greatly elevated the expression of AKT phosphorylation of Ser473 (p-AKT on Ser473) compared with saline treatment (Figure 2A,B). Although comparable expression of total GSK3β was observed in saline and insulin-treated livers, phosphorylated GSK3β (p-GSK3β on Ser9) levels were higher in the hepatic whole cell lysates of alcohol-fed Tlr4LKO mice (Figure 2A,C). Impaired insulin signaling in epididymal white adipose tissue (eWAT) has been reported in alcohol-fed rodent models [12]. Therefore, Western blotting of total AKT and phosphorylated AKT (p-AKT) was performed on eWAT samples from mice fed an alcohol-containing liquid diet for 8 weeks. In response to insulin administration, similarly elevated AKT phosphorylation in the eWAT was observed in Tlr4fl/fl and Tlr4LKO mice (Figure 2D,E). Insulin stimulation of GSK phosphorylation on Ser9 in the eWAT tended to be increased in alcohol-fed Tlr4LKO mice (Figure 2D). However, the densitometry quantification did not show a significant difference (Figure 2F). These findings suggest that hepatocyte TLR4 deficiency protects mice from chronic alcohol-induced insulin resistance in the liver. ## 3.3. Restoration of Endogenous TLR4 Expression in Hepatocytes Did Not Affect Blood Glucose and Plasma Insulin Levels after Chronic Alcohol Feeding To investigate the sufficiency of hepatocyte TLR4 in alcohol-induced insulin resistance, we used a mouse model that is conditionally null for TLR4 (Tlr4LoxTB) [24]. The utility of this mouse model has been widely tested [24,25,26]. To re-express Tlr4 selectively in hepatocytes, we crossed the Tlr4LoxTB mice to Alb-Cre mice (Tlr4LoxTB × Alb-Cre). As shown in Figure 3A, primary hepatocytes isolated from Tlr4LoxTB × Alb-Cre mice had dramatically elevated endogenous TLR4 mRNA expression. Then blood glucose and plasma insulin levels were determined in fasted Tlr4LoxTB × Alb-Cre and littermate Tlr4LoxTB mice after chronic liquid diet feeding. We found that alcohol-fed Tlr4LoxTB and Tlr4LoxTB × Alb-Cre mice showed similarly reduced blood glucose levels (Figure 3B). In addition, there were no differences in plasma insulin levels between Tlr4LoxTB and Tlr4LoxTB × Alb-Cre littermates fed a control liquid diet (Figure 3C). After chronic alcohol feeding, reduced plasma insulin concentrations were observed in Tlr4LoxTB mice, but not in Tlr4LoxTB × Alb-Cre littermates (Figure 3C). ## 3.4. Restoration of Endogenous TLR4 Expression in Hepatocytes Exacerbated Glucose Intolerance in Alcohol-Fed Mice Next, we performed IPGTTs and ITTs in Tlr4LoxTB and Tlr4LoxTB × Alb-Cre mice after chronic liquid diet feeding. On control liquid diets, both genotypes showed similar blood glucose responses after intraperitoneal injection of glucose (Figure 4A,C). After chronic alcohol feeding, Tlr4LoxTB mice had elevated normalized blood glucose levels, indicating glucose intolerance in these mice (Figure 4B,C). Interestingly, compared with Tlr4LoxTB mice, alcohol-fed Tlr4LoxTB × Alb-Cre mice exhibited even higher normalized blood glucose levels observed 15 min after glucose administration (Figure 4B). In addition, the AUC analysis showed that reactivation of TLR4 in hepatocytes tended to reduce systemic glucose tolerance in mice after chronic alcohol feeding (Figure 4C). For ITTs, Tlr4LoxTB mice maintained similar blood glucose levels on both control and alcohol-containing diets, suggesting that whole-body TLR4 knockout mice were resistant to alcohol-induced systemic insulin resistance (Figure 4D,F). After chronic alcohol feeding, Tlr4LoxTB × Alb-Cre mice tended to have elevated blood glucose levels 15 and 30 min after insulin injection compared with Tlr4LoxTB mice (Figure 4E). However, the increases were not significantly different (Figure 4E,F). These findings suggest that reactivation of hepatocyte TLR4 promotes alcohol-induced glucose intolerance in mice. ## 3.5. Restoration of Hepatocyte TLR4 in Mice Led to Impaired Phosphorylation of AKT in Both Liver and Adipose Tissue To determine whether TLR4 reactivation in hepatocytes is sufficient to impair insulin signaling in the liver and adipose tissue, we performed acute saline and insulin injection in alcohol-fed Tlr4LoxTB, Tlr4LoxTB × Alb-Cre, and Tlr4LoxTB × Zp3-Cre mice. Compared to saline treatment, insulin administration caused increased AKT phosphorylation of Ser473 (p-AKT on Ser473) in the liver of Tlr4LoxTB mice (Figure 5A,B). Interestingly, this increase was dramatically blunted in livers prepared from Tlr4LoxTB × Alb-Cre mice. However, restoration of TLR4 in hepatocytes and in the whole body did not suppress hepatic p-GSK3β expression (Figure 5A,C). In addition, western blot analyses were performed in whole cell lysates of eWAT from saline and insulin-treated mice after chronic alcohol drinking. Figure 5D showed that insulin treatment greatly increased AKT phosphorylation of Ser473 in the eWAT of Tlr4LoxTB mice. However, eWAT from Tlr4LoxTB × Alb-Cre and Tlr4LoxTB × Zp3-Cre mice showed dramatically reduced p-AKT (Ser473) expression (Figure 5D,E). Regardless of the genotypes, comparably increased p-GSK3β expression was observed in the eWAT following insulin administration (Figure 5D,F). ## 3.6. Hepatocyte TLR4 Regulates Liver Triglyceride Contents in Mice after Excessive Alcohol Intake It has been widely reported that chronic alcohol administration leads to hepatic steatosis in rodents and human subjects [29]. Consistently, we found that 8 weeks of alcohol intake caused significantly increased liver triglyceride contents in Tlr4fl/fl mice (Figure 6A). However, mice lacking hepatocyte TLR4 tended to accumulate less triglyceride in the liver after chronic alcohol feeding (Figure 6A). Interestingly, global TLR4 knockout mice (Tlr4LoxTB) were resistant to alcohol overconsumption-induced hepatic steatosis (Figure 6B). In contrast, reactivation of endogenous TLR4 in hepatocytes and in the whole body promoted more triglyceride accumulation in their livers (Figure 6B). These data suggest that hepatocyte TLR4 plays an important role in regulating chronic alcohol-induced fatty liver development. ## 4. Discussion The relationship between alcohol consumption and insulin sensitivity is complicated. Accumulating epidemiological studies suggest that light to moderate alcohol intake is associated with enhanced insulin sensitivity [1,30,31,32]; however, heavy drinking promotes the development of insulin resistance [2,30]. Consistent with the impaired insulin action observed in individuals drinking a large amount of alcohol, experimental animals with chronic alcohol intake develop glucose intolerance [4,6,12] and/or insulin resistance [5,6,12]. Carr et al. reported that 4 weeks of chronic alcohol feeding leads to glucose intolerance and insulin resistance in C57BL/6 mice. In addition, significantly increased hepatic glucose production and reduced peripheral glucose disposal were observed in mice fed an alcohol-containing liquid diet for 8 weeks [7]. Collectively, these findings suggest a correlation between excessive alcohol consumption and insulin resistance. In agreement with previous reports, we also observed that alcohol overconsumption led to reduced glucose tolerance and impaired insulin sensitivity in Tlr4fl/fl mice (Figure 1D,G). Interestingly, alcohol-fed hepatocyte TLR4 deficient mice (Tlr4LKO) were insulin sensitive during ITT (Figure 1G) and showed increased insulin-stimulated phosphorylation of GSK3β in the liver (Figure 2A). Taking advantage of the cell-type specific TLR4 reactivatable mouse model (Tlr4LoxTB), we generated mice that restore endogenous TLR4 expression only in hepatocytes under TLR4 null background (Tlr4LoxTB × Alb-Cre). This reactivation model could prevent the changes in TLR4 expression in other cell types and limit their potential contributions to disease development. We found that, compared with Tlr4LoxTB mice, Tlr4LoxTB × Alb-Cre mice tended to develop glucose intolerance after chronic alcohol feeding (Figure 4B). In addition, insulin-stimulated AKT phosphorylation was dramatically reduced in the livers of Tlr4LoxTB × Alb-Cre mice following heavy drinking, which was comparable to Tlr4LoxTB × Zp3-Cre mice (Figure 5A). Furthermore, Tlr4LoxTB × Alb-Cre mice exhibited the lowest expression of p-AKT on Ser473 in the eWAT (Figure 5D). These data suggest that hepatocyte TLR4 is sufficient to impair insulin sensitivity in alcohol-fed mice. Our IPGTT experiments showed that global TLR4 deficient mice (Tlr4LoxTB) following chronic alcohol-containing liquid diet feeding developed systemic glucose intolerance (Figure 4B,C). This is not surprising since it has been reported that mice lacking intestinal epithelial TLR4 showed impaired glucose tolerance after a HFD feeding [33]. Therefore, TLR4 signaling in different cell types may trigger opposite effects on the development of the metabolic syndrome. In contrast to previous findings that adenovirus-mediated overexpression of TLR4 in the liver caused insulin resistance in chow-fed wild-type mice [21], in the current study, similar glucose tolerance and insulin sensitivity were observed in Tlr4LoxTB × Alb-Cre and littermate Tlr4LoxTB mice after chronic control liquid diet feeding (Figure 4A,D). The discrepancy between these two studies could be due to the differences in mouse models and diets. In addition, we cannot rule out the possibility that the combined effect of adenovirus-induced liver injury and TLR4-mediated inflammation could contribute to impaired insulin sensitivity in wild-type mice on a chow diet [21]. Despite the consistent observation that heavy drinking causes systemic insulin resistance in rodents, the discrepancy remains regarding the alterations of proteins involved in insulin signaling pathways. For example, reduced p-AKT expression has been observed in alcohol-exposed mouse livers [4,7,34]. However, He et al. reported increased AKT phosphorylation at Ser473 in the livers of alcohol-fed rats [9]. In eWAT isolated from alcohol-fed mice, both reduced [35] and increased [36] p-AKT protein levels were reported. Moreover, Poirier et al. showed that insulin-stimulated AKT phosphorylation was not affected in alcohol-treated adipocytes [15]. GSK3β is negatively regulated by insulin via AKT phosphorylation. Phosphorylated GSK3β at Ser9 inhibits its kinase activity and enhances insulin-stimulated glycogen synthesis [37,38]. In addition, GSK3β has been shown to directly phosphorylate insulin receptor substrate-1 and impair insulin signaling [37]. Consistent with the concept that alcohol drinking suppresses insulin sensitivity, reduced GSK3β phosphorylation has been reported in the liver and eWAT of alcohol-fed rodents [9,35]. In contrast, increased GSK3β Ser9 phosphorylation was observed in mouse livers after 4 weeks of chronic alcohol feeding [4]. These inconsistent observations could be due to different animal models (rat vs. mouse) and/or experimental conditions (ex vivo vs. in vitro, and acute vs. chronic treatment). In the current study, we examined the protein levels of AKT and GSK3β in alcohol-fed mice after acute insulin stimulation. Interestingly, hepatocyte TLR4 deficient mice showed improved insulin sensitivity and elevated GSK3β Ser9 phosphorylation in the liver after chronic alcohol feeding. While reactivation of TLR4 in hepatocytes decreased AKT Ser473 phosphorylation but did not affect p-GSK3β. It is possible that other TLR4-expressing cells in Tlr4LKO mice could influence insulin signaling molecules and contribute to alcohol-induced insulin resistance. Global TLR4 knockout mice have been reported to protect against alcohol-induced fatty liver disease [17]. Consistently, we found that Tlr4LoxTB mice, which lack TLR4 in the whole body, are resistant to alcohol-induced hepatic triglyceride accumulation (Figure 6B). Interestingly, reactivation of TLR4 in hepatocytes and in the whole body similarly promoted the development of hepatic steatosis in mice after chronic alcohol feeding (Figure 6B). In contrast, alcohol-fed hepatocyte TLR4 deficient mice tended to accumulate less triglyceride in the liver (Figure 6A). These findings indicate that hepatocyte TLR4 plays a role in mediating alcohol-induced fatty liver disease. In addition, the effect of hepatocyte TLR4 on liver triglyceride contents could partially contribute to its role in altering insulin sensitivity in the context of chronic alcohol drinking. It has been reported that hepatocyte TLR4 deficient mice have reduced systemic and adipose tissue inflammation after chronic HFD and alcohol-containing liquid diet feeding [19,22]. Considering the relationship between inflammation and insulin resistance [39] and the important role of TLR4 in mediating inflammatory response, we speculate that hepatocyte TLR4 is required in mediating tissue and systemic inflammatory response, and its deficiency is correlated with decreased adipose tissue inflammation and enhanced insulin sensitivity. Furthermore, significant attention has been paid to the critical role of crosstalk between the liver and other tissues in disease development. Several hepatokines have been shown to greatly affect adipose tissue inflammatory response and insulin sensitivity [40]. However, the specific hepatokine(s) that mediate hepatocyte TLR4-induced inflammation and insulin resistance following chronic alcohol feeding are largely unknown and require further investigation. The limitation of the current study was that a relatively small number of Tlr4fl/fl and Tlr4LKO mice were fed the control liquid diet and used in some experiments. For example, we performed IPGTT and ITT experiments in Tlr4fl/fl and Tlr4LKO mice ($$n = 3$$–4) after 8 weeks of control liquid diet feeding and found that regardless of the genotypes, mice exhibited similar blood glucose levels. Previously we have reported that Tlr4fl/fl and Tlr4LKO mice following a long-term chow diet feeding responded similarly to intraperitoneal injection of either glucose or insulin [22]. Therefore, under conditions that do not cause diseases, such as control liquid diet or chow diet feeding, TLR4 deletion in hepatocytes does not affect glucose metabolism or insulin sensitivity. In addition, comparable liver triglyceride content was observed in control-fed Tlr4fl/fl and Tlr4LKO mice ($$n = 3$$–4). Consistently, we have observed that hepatocyte TLR4 deficiency did not influence hepatic triglyceride levels when mice were fed a control liquid diet for 4 weeks or acutely treated with maltose dextrin via oral gavage [19]. Nevertheless, the small sample size of control-fed mice should be considered during data interpretation. In summary, these findings provide in vivo evidence that hepatocyte TLR4 is a significant contributor to the development of excessive alcohol intake-induced insulin resistance. Targeting hepatocyte TLR4 could have therapeutic potential in the prevention and treatment of metabolic syndrome induced by alcohol overconsumption. ## References 1. Bell D.S.. Alcohol and the NIDDM patient. Diabetes Care (1996) 19 509-513. DOI: 10.2337/diacare.19.5.509 2. Roh W.G., Shin H.C., Choi J.H., Lee Y.J., Kim K.. Alcohol consumption and higher incidence of impaired fasting glucose or type 2 diabetes in obese Korean men. Alcohol (2009) 43 643-648. DOI: 10.1016/j.alcohol.2009.10.005 3. Shmueli E., Walker M., Alberti G., Record C.O.. Normal splanchnic but impaired peripheral insulin-stimulated glucose uptake in cirrhosis. Hepatology (1993) 18 86-95. DOI: 10.1002/hep.1840180115 4. Liangpunsakul S., Rahmini Y., Ross R.A., Zhao Z., Xu Y., Crabb D.W.. Imipramine blocks ethanol-induced ASMase activation, ceramide generation, and PP2A activation, and ameliorates hepatic steatosis in ethanol-fed mice. Am. J. Physiol. Gastrointest. Liver Physiol. (2012) 302 G515-G523. DOI: 10.1152/ajpgi.00455.2011 5. Zhong W., Zhao Y., Tang Y., Wei X., Shi X., Sun W., Sun X., Yin X., Sun X., Kim S.. Chronic alcohol exposure stimulates adipose tissue lipolysis in mice: Role of reverse triglyceride transport in the pathogenesis of alcoholic steatosis. Am. J. Pathol. (2012) 180 998-1007. DOI: 10.1016/j.ajpath.2011.11.017 6. Onishi Y., Honda M., Ogihara T., Sakoda H., Anai M., Fujishiro M., Ono H., Shojima N., Fukushima Y., Inukai K.. Ethanol feeding induces insulin resistance with enhanced PI 3-kinase activation. Biochem. Biophys. Res. Commun. (2003) 303 788-794. DOI: 10.1016/S0006-291X(03)00407-8 7. Carr R.M., Dhir R., Yin X., Agarwal B., Ahima R.S.. Temporal effects of ethanol consumption on energy homeostasis, hepatic steatosis, and insulin sensitivity in mice. Alcohol. Clin. Exp. Res. (2013) 37 1091-1099. DOI: 10.1111/acer.12075 8. de la Monte S.M., Yeon J.E., Tong M., Longato L., Chaudhry R., Pang M.Y., Duan K., Wands J.R.. Insulin resistance in experimental alcohol-induced liver disease. J. Gastroenterol. Hepatol. (2008) 23 e477-e486. DOI: 10.1111/j.1440-1746.2008.05339.x 9. He L., Simmen F.A., Mehendale H.M., Ronis M.J., Badger T.M.. Chronic ethanol intake impairs insulin signaling in rats by disrupting Akt association with the cell membrane. Role of TRB3 in inhibition of Akt/protein kinase B activation. J. Biol. Chem. (2006) 281 11126-11134. DOI: 10.1074/jbc.M510724200 10. Denucci S.M., Tong M., Longato L., Lawton M., Setshedi M., Carlson R.I., Wands J.R., de la Monte S.M.. Rat strain differences in susceptibility to alcohol-induced chronic liver injury and hepatic insulin resistance. Gastroenterol. Res. Pract. (2010) 2010 312790. DOI: 10.1155/2010/312790 11. Ramirez T., Longato L., Dostalek M., Tong M., Wands J.R., de la Monte S.M.. Insulin resistance, ceramide accumulation and endoplasmic reticulum stress in experimental chronic alcohol-induced steatohepatitis. Alcohol Alcohol. (2013) 48 39-52. DOI: 10.1093/alcalc/ags106 12. Kang L., Sebastian B.M., Pritchard M.T., Pratt B.T., Previs S.F., Nagy L.E.. Chronic ethanol-induced insulin resistance is associated with macrophage infiltration into adipose tissue and altered expression of adipocytokines. Alcohol. Clin. Exp. Res. (2007) 31 1581-1588. DOI: 10.1111/j.1530-0277.2007.00452.x 13. Patel B.C., D’Arville C., Iwahashi M., Simon F.R.. Impairment of hepatic insulin receptors during chronic ethanol administration. Am. J. Physiol. (1991) 261 G199-G205. DOI: 10.1152/ajpgi.1991.261.2.G199 14. Wilkes J.J., DeForrest L.L., Nagy L.E.. Chronic ethanol feeding in a high-fat diet decreases insulin-stimulated glucose transport in rat adipocytes. Am. J. Physiol. (1996) 271 E477-E484. DOI: 10.1152/ajpendo.1996.271.3.E477 15. Poirier L.A., Rachdaoui N., Nagy L.E.. GLUT4 vesicle trafficking in rat adipocytes after ethanol feeding: Regulation by heterotrimeric G-proteins. Biochem. J. (2001) 354 323-330. DOI: 10.1042/bj3540323 16. Adachi Y., Moore L.E., Bradford B.U., Gao W., Thurman R.G.. Antibiotics prevent liver injury in rats following long-term exposure to ethanol. Gastroenterology (1995) 108 218-224. DOI: 10.1016/0016-5085(95)90027-6 17. Hritz I., Mandrekar P., Velayudham A., Catalano D., Dolganiuc A., Kodys K., Kurt-Jones E., Szabo G.. The critical role of toll-like receptor (TLR) 4 in alcoholic liver disease is independent of the common TLR adapter MyD88. Hepatology (2008) 48 1224-1231. DOI: 10.1002/hep.22470 18. Inokuchi S., Tsukamoto H., Park E., Liu Z.X., Brenner D.A., Seki E.. Toll-like receptor 4 mediates alcohol-induced steatohepatitis through bone marrow-derived and endogenous liver cells in mice. Alcohol. Clin. Exp. Res. (2011) 35 1509-1518. DOI: 10.1111/j.1530-0277.2011.01487.x 19. Jia L., Chang X., Qian S., Liu C., Lord C.C., Ahmed N., Lee C.E., Lee S., Gautron L., Mitchell M.C.. Hepatocyte toll-like receptor 4 deficiency protects against alcohol-induced fatty liver disease. Mol. Metab. (2018) 14 121-129. DOI: 10.1016/j.molmet.2018.05.015 20. Fulham M.A., Ratna A., Gerstein R.M., Kurt-Jones E.A., Mandrekar P.. Alcohol-induced adipose tissue macrophage phenotypic switching is independent of myeloid Toll-like receptor 4 expression. Am. J. Physiol. Cell Physiol. (2019) 317 C687-C700. DOI: 10.1152/ajpcell.00276.2017 21. Uchimura K., Hayata M., Mizumoto T., Miyasato Y., Kakizoe Y., Morinaga J., Onoue T., Yamazoe R., Ueda M., Adachi M.. The serine protease prostasin regulates hepatic insulin sensitivity by modulating TLR4 signalling. Nat. Commun. (2014) 5 3428. DOI: 10.1038/ncomms4428 22. Jia L., Vianna C.R., Fukuda M., Berglund E.D., Liu C., Tao C., Sun K., Liu T., Harper M.J., Lee C.E.. Hepatocyte Toll-like receptor 4 regulates obesity-induced inflammation and insulin resistance. Nat. Commun. (2014) 5 3878. DOI: 10.1038/ncomms4878 23. Shan B., Shao M., Zhang Q., Hepler C., Paschoal V.A., Barnes S.D., Vishvanath L., An Y.A., Jia L., Malladi V.S.. Perivascular mesenchymal cells control adipose-tissue macrophage accrual in obesity. Nat. Metab. (2020) 2 1332-1349. DOI: 10.1038/s42255-020-00301-7 24. Jia L., Lee S., Tierney J.A., Elmquist J.K., Burton M.D., Gautron L.. TLR4 Signaling Selectively and Directly Promotes CGRP Release from Vagal Afferents in the Mouse. eNeuro (2021) 8. DOI: 10.1523/ENEURO.0254-20.2020 25. Szabo-Pardi T.A., Barron L.R., Lenert M.E., Burton M.D.. Sensory Neuron TLR4 mediates the development of nerve-injury induced mechanical hypersensitivity in female mice. Brain Behav. Immun. (2021) 97 42-60. DOI: 10.1016/j.bbi.2021.06.011 26. Ursino G., Ramadori G., Hofler A., Odouard S., Teixeira P.D.S., Visentin F., Veyrat-Durebex C., Lucibello G., Firnkes R., Ricci S.. Hepatic non-parenchymal S100A9-TLR4-mTORC1 axis normalizes diabetic ketogenesis. Nat. Commun. (2022) 13 4107. DOI: 10.1038/s41467-022-31803-5 27. Bertola A., Mathews S., Ki S., Wang H., Gao B.. Mouse model of chronic and binge ethanol feeding (the NIAAA model). Nat. Protocols. (2013) 8 627-637. DOI: 10.1038/nprot.2013.032 28. Field J.B., Williams H.E., Mortimore G.E.. Studies on the mechanism of ethanol-induced hypoglycemia. J. Clin. Investig. (1963) 42 497-506. DOI: 10.1172/JCI104738 29. You M., Arteel G.E.. Effect of ethanol on lipid metabolism. J. Hepatol. (2019) 70 237-248. DOI: 10.1016/j.jhep.2018.10.037 30. Fuchs C.S., Stampfer M.J., Colditz G.A., Giovannucci E.L., Manson J.E., Kawachi I., Hunter D.J., Hankinson S.E., Hennekens C.H., Rosner B.. Alcohol consumption and mortality among women. N. Engl. J. Med. (1995) 332 1245-1250. DOI: 10.1056/NEJM199505113321901 31. Facchini F., Chen Y.D., Reaven G.M.. Light-to-moderate alcohol intake is associated with enhanced insulin sensitivity. Diabetes Care (1994) 17 115-119. DOI: 10.2337/diacare.17.2.115 32. Rimm E.B., Chan J., Stampfer M.J., Colditz G.A., Willett W.C.. Prospective study of cigarette smoking, alcohol use, and the risk of diabetes in men. BMJ (1995) 310 555-559. DOI: 10.1136/bmj.310.6979.555 33. Lu P., Sodhi C.P., Yamaguchi Y., Jia H., Prindle T., Fulton W.B., Vikram A., Bibby K.J., Morowitz M.J., Hackam D.J.. Intestinal epithelial Toll-like receptor 4 prevents metabolic syndrome by regulating interactions between microbes and intestinal epithelial cells in mice. Mucosal Immunol. (2018) 11 727-740. DOI: 10.1038/mi.2017.114 34. Zhang Y., Lu J., Zhong Y.J., Yang C.F., Chen L., Wu D., Song M.W., Shi L., Ma Z.H., Li L.. Methyl ferulic acid ameliorates alcohol-induced hepatic insulin resistance via miR-378b-mediated activation of PI3K-AKT pathway. Biomed. Pharmacother. (2022) 145 112462. DOI: 10.1016/j.biopha.2021.112462 35. Li Y., Chao X., Wang S., Williams J.A., Ni H.M., Ding W.X.. Role of Mechanistic Target of Rapamycin and Autophagy in Alcohol-Induced Adipose Atrophy and Liver Injury. Am. J. Pathol. (2020) 190 158-175. DOI: 10.1016/j.ajpath.2019.09.023 36. Zhao C., Liu Y., Xiao J., Liu L., Chen S., Mohammadi M., McClain C.J., Li X., Feng W.. FGF21 mediates alcohol-induced adipose tissue lipolysis by activation of systemic release of catecholamine in mice. J. Lipid Res. (2015) 56 1481-1491. DOI: 10.1194/jlr.M058610 37. Leng S., Zhang W., Zheng Y., Liberman Z., Rhodes C.J., Eldar-Finkelman H., Sun X.J.. Glycogen synthase kinase 3 beta mediates high glucose-induced ubiquitination and proteasome degradation of insulin receptor substrate 1. J. Endocrinol. (2010) 206 171-181. DOI: 10.1677/JOE-09-0456 38. Krishnankutty A., Kimura T., Saito T., Aoyagi K., Asada A., Takahashi S.I., Ando K., Ohara-Imaizumi M., Ishiguro K., Hisanaga S.I.. In vivo regulation of glycogen synthase kinase 3beta activity in neurons and brains. Sci. Rep. (2017) 7 8602. DOI: 10.1038/s41598-017-09239-5 39. Lee Y., Olefsky J.. Chronic tissue inflammation and metabolic disease. Genes Dev. (2021) 35 307-328. DOI: 10.1101/gad.346312.120 40. Watt M., Miotto P.M., De Nardo W., Montgomery M.K.. The Liver as an Endocrine Organ-Linking NAFLD and Insulin Resistance. Endocr. Rev. (2019) 40 1367-1393. DOI: 10.1210/er.2019-00034
--- title: Distinct Changes in Placental Ceramide Metabolism Characterize Type 1 and 2 Diabetic Pregnancies with Fetal Macrosomia or Preeclampsia authors: - Miira M. Klemetti - Sruthi Alahari - Martin Post - Isabella Caniggia journal: Biomedicines year: 2023 pmcid: PMC10046505 doi: 10.3390/biomedicines11030932 license: CC BY 4.0 --- # Distinct Changes in Placental Ceramide Metabolism Characterize Type 1 and 2 Diabetic Pregnancies with Fetal Macrosomia or Preeclampsia ## Abstract Disturbances of lipid metabolism are typical in diabetes. Our objective was to characterize and compare placental sphingolipid metabolism in type 1 (T1D) and 2 (T2D) diabetic pregnancies and in non-diabetic controls. Placental samples from T1D, T2D, and control pregnancies were processed for sphingolipid analysis using tandem mass spectrometry. Western blotting, enzyme activity, and immunofluorescence analyses were used to study sphingolipid regulatory enzymes. Placental ceramide levels were lower in T1D and T2D compared to controls, which was associated with an upregulation of the ceramide degrading enzyme acid ceramidase (ASAH1). Increased placental ceramide content was found in T1D complicated by preeclampsia. Similarly, elevated ceramides were observed in T1D and T2D pregnancies with poor glycemic control. The protein levels and activity of sphingosine kinases (SPHK) that produce sphingoid-1-phosphates (S1P) were highest in T2D. Furthermore, SPHK levels were upregulated in T1D and T2D pregnancies with fetal macrosomia. In vitro experiments using trophoblastic JEG3 cells demonstrated increased SPHK expression and activity following glucose and insulin treatments. Specific changes in the placental sphingolipidome characterize T1D and T2D placentae depending on the type of diabetes and feto-maternal complications. Increased exposure to insulin and glucose is a plausible contributor to the upregulation of the SPHK-S1P-axis in diabetic placentae. ## 1. Introduction Proper placental function is a vital determinant of maternal adaptation to pregnancy, fetal development and growth, and long-term health outcomes in both the mother and child [1,2]. Vascular and metabolic complications, such as hypertensive disorders and abnormal fetal growth, are common in pregnancies affected by type 1 (T1D) or type 2 diabetes (T2D), and the placenta is known to be centrally involved in their pathogenesis [3,4,5]. Complex disturbances of the lipidome of highly metabolic tissues and plasma are typical in diabetes [6,7,8]; however, our understanding of the changes in lipid metabolism that occur in the placenta due to maternal diabetes is incomplete. Sphingolipids are bioactive lipids that, apart from their function as essential building blocks of the cell membrane, are critically involved in a plethora of cell signaling pathways. For example, specific sphingolipid species are implicated in cell fate, inflammatory, immune and stress responses, cell adhesion and migration, angiogenesis, vascular function, and mitochondrial bioenergetics [9,10]. In addition, they are also important regulators of cell membrane dynamics and vesicular trafficking [9]. The precursor of all sphingolipids is ceramide, the “hub” of complex sphingolipid metabolism, which is produced and degraded in a compartmentalized fashion by a multitude of enzyme variants and interconnected pathways [11]. Briefly, the de novo synthesis pathway of ceramide begins in the endoplasmic reticulum (ER) with the production of sphinganine from palmitoyl-CoA and serine in a rate-limiting reaction catalyzed by serine palmitoyltranferase. Subsequently, acyl chains of variable length (14 to 34 carbon atoms) are added to the sphingoid backbone by specific ceramide synthases (CerS1-6). Alternatively, sphinganine can be phosphorylated into sphinganine-1-phosphate (Sa1P) by sphingosine kinases (SPHK) [12]. Ceramide can also be produced by hydrolysis of sphingomyelin catalyzed by sphingomyelinases (hydrolysis pathway), or via re-acylation from sphingosine (salvage pathway) [11]. Ceramide biogenesis by these pathways can be triggered in different intracellular compartments by various stressors, such as oxidative insults, hypoxia, inflammation, or an overload of saturated fatty acids [13]. Ceramide is hydrolyzed into sphingosine (SPH) and fatty acids by ceramidases. SPH can then be phosphorylated by SPHKs to form the lysosphingolipid sphingosine-1-phosphate (S1P). Both ceramide and S1P govern various signaling pathways related to cell death, proliferation, migration, survival, and senescence [10,14]. Numerous studies have reported that disturbances of sphingolipid homeostasis are key to the onset and progression of diabetes, obesity, and associated cardiovascular conditions [8,13]. We have shown that preeclampsia, a life-threatening pregnancy hypertensive syndrome typified by maternal systemic endothelial dysfunction, is associated with placental lysosomal ceramide build-up, characterizing preeclampsia as a sphingolipid storage disorder [15]. Furthermore, we have reported that placental ceramide accumulation in preeclampsia results in increased trophoblast autophagy and necroptosis [15,16], and tilts the mitochondrial dynamic balance towards fission, leading to increased mitophagy [17]. In contrast, in fetal growth restriction, placental ceramide levels are decreased due to the upregulation of acid ceramidase (ASAH1). In addition, SPHK expression and activity are reduced, suggesting that the aberrant placental ceramide/sphingoid-1-phosphate axis might contribute to abnormal fetal growth [18]. Interestingly, we recently demonstrated a similar reduction in ceramides, in conjunction with upregulated mitochondrial fusion, in placentae affected by gestational diabetes (GDM) [19]. Collectively, these observations suggest that specific alterations in the placental sphingolipidome distinguish vasculo-metabolic complications of pregnancy and are interlinked with major alterations in placental cell metabolism and bioenergetics. In the context of T1D and T2D pregnancies, the significance of sphingolipids has received less attention. Herein, we examined placental ceramide and sphingoid-1-phosphate content and metabolism in women with T1D and T2D. Additionally, we investigated whether common complications that burden mothers with diabetes (preeclampsia) and their fetuses (macrosomia/large-for-gestational age) are characterized by distinct placental sphingolipid profiles that could also shed light on the placental metabolic derangements implicated in the pathogenesis of these complex outcomes. ## 2.1. Collection of Placental Samples and Clinical Information Pregnant women with T1D ($$n = 28$$) and T2D ($$n = 21$$) were recruited by the Research Centre for Women’s and Infants’ Health (RCWIH) Biobank, Mount Sinai Hospital (MSH), Toronto, Canada. Healthy women without diabetes and with a delivery between 34+0- and 40+5-weeks’ gestation ($$n = 49$$) were recruited as controls. Only singleton pregnancies were included in the study. Smokers (self-reported, any time of gestation) and pregnancies affected by substance abuse, fetal malformations, chromosomal aberrations, chorioamnionitis, or small-for-gestational age (relative birth weight <−2.0 SD units) in the absence of preeclampsia were excluded. All participants provided written informed consent. Placental tissue was collected according to the ethical guidelines of the University of Toronto, Faculty of Medicine and MSH, immediately after delivery, and snap-frozen in liquid nitrogen. The study protocol was approved by the MSH Research Ethics Board (REB number: 11-0287-E) and carried out in agreement with the Declaration of Helsinki. Available information on maternal age, parity and gravidity, pre-pregnancy weight and height, gestational weight gain (GWG), age at diabetes diagnosis, blood pressure, glycated hemoglobin (HbA1c), and obstetric and perinatal outcomes were extracted from the MSH patient records. Obesity was defined as maternal pre-pregnancy body mass index (BMI) ≥ 30 kg/m2. GWG was categorized according to the Institute of Medicine (IOM) criteria as adequate, below, or above recommendations [20]. Preeclampsia was defined according to the criteria of the American College of Obstetricians and Gynecologists [21]. Fetal macrosomia (i.e., large-for-gestational age (LGA)) was defined as a relative birth weight (birth weight z-score) exceeding + 2 SD units (>97.7th percentile) using a Canadian standard population, standardized for sex and gestational age [22]. ## 2.2. Lipid Mass Spectral Analysis Placental tissue from 42 controls, 26 women with T1D, and 21 women with T2D were processed for lipid analysis as described previously [17]. Following lipid extraction, sphingolipid species were quantified utilizing high-performance liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) at the Analytical Facility for Bioactive Molecules, Hospital for Sick Children, Toronto, ON, Canada. ## 2.3. Western Blotting Approximately 100–200 mg of snap-frozen placentae were crushed, dissolved in 400–600 μL of RIPA buffer containing proteasome inhibitor, and homogenized at 4 °C to generate placentae tissue lysates that were centrifuged at 12,000 rpm for 10 min. Protein concentration was quantified using Bradford protein assay (Bio-Rad®, Mississauga, ON, Canada). Placental lysates containing 30 μg of protein were subjected to SDS-PAGE on $7.5\%$ (SPHK1 protein analysis), $10\%$ (SPHK2 protein analysis), or $12\%$ (ASAH1 protein analysis) BioRad FastCast acrylamide gels (Bio-Rad®, Mississauga, ON, Canada). After each run, the gels were imaged for a stain-free profile of total protein using the BioRad Chemidoc XRS+ System. This stain-free imaging technology is based on the detection of a fluorescence signal generated when trihalo compounds in the polyacrylamide gels modify tryptophan residues on exposure to ultraviolet light [23]. Subsequently, proteins were transferred onto polyvinylidene fluoride (PVDF) membranes using Trans-Blot Turbo transfer buffer 5×. Membranes were blocked with $5\%$ (w/v) non-fat milk dissolved in Tris-buffered saline containing $0.1\%$ (v/v) Tween 20 (TBS-T) for 1 h, before probing with respective antibodies diluted in $5\%$ (w/v) non-fat milk at 4 °C overnight. Next day, membranes were washed 3 × 10 min in TBS-T, incubated with appropriate secondary antibody in $5\%$ (w/v) non-fat milk for 1 h at room temperature, and washed 3 × 10 min in TBS-T. Finally, membranes were simultaneously visualized for immunoreactivity following the addition of chemiluminescence ECL reagent (Bio-Rad®, Mississauga, ON, Canada) and equal exposure time using Bio-Rad Chemidoc XRS+ System. Densitometric analysis was performed using ImageLab software, with data normalized to total protein in stain-free gels. Total protein normalization was reported to be the most appropriate normalization method [23], as diabetes and preeclampsia have both been demonstrated to affect the placental expression of various proteins that are commonly used as loading controls/housekeeping proteins [24]. ## 2.4. Immunofluorescence Analysis Immunofluorescence staining for SPHK1 and SPHK2 as well as calreticulin (CRT; ER marker), TOM20 (mitochondria marker), and zonula occludens-1 (ZO-1; plasma membrane marker) in JEG3 cells treated with 25 mM of glucose (GLU) and 0.85 μM of insulin solution, or with EMEM vehicle (VEH) alone, was performed as previously described [17]. IF intensity and co-localization (by Pearson’s correlation coefficient) were quantified using ImageJ as reported [25]. ## 2.5. Antibodies Rabbit anti-ASAH1 was from Aviva Systems Biology (Cedarlane, Burlington, ON, Canada (OAPB00726; WB 1:750). Mouse monoclonal anti-SPHK1 (SC-365401; IF 1:200), goat anti-SPHK2 (SC-22704; IF 1:200, WB 1:500), rabbit anti-ZO-1 (SC-10804; IF 1:200), and rabbit anti-TOM20 (SC-11415; IF 1:200) were purchased from Santa Cruz Biotechnology (Mississauga, ON, Canada). Rabbit monoclonal anti-SPHK1 (ab109522; WB 1:1000) and mouse monoclonal anti-calreticulin (ab22683; IF 1:300) were from Abcam (Cedarlane, Burlington, ON, Canada). Secondary antibodies used were goat anti-rabbit IgG-HRP (sc-2054; WB 1:2000), goat anti-mouse IgG-HRP (sc-2005; WB 1:2000), and donkey anti-goat IgG (H+L) (WB 1:2000) from Jackson Laboratory, Bar Harbor, Maine, USA. For IF experiments, Alexa Fluor 488 donkey anti-rabbit IgG (A21206) and Alexa Fluor 594 donkey anti-mouse IgG (A21203) were obtained from ThermoFisher Scientific (Mississauga, ON, Canada). ## 2.6. SPHK Activity Assay Total SPHK enzyme activity (SPHK1+SPHK2) was measured in placental and JEG3 lysates using an SPHK Assay Kit (Echelon Bioscience, K-3500, Salt Lake City, UT, USA) as previously described [18]. ## 2.7. RNA Analysis Total RNA was extracted from 30 mg of frozen placental tissue using a QIAGEN RNeasy Plus Mini Kit, according to the manufacturer’s protocol. RNA quantity and purity were analyzed using a Thermo Scientific Nanodrop 1000. Next, 1 µg of RNA was reverse transcribed using the Quantabio qScript cDNA SuperMix kit (VWR International, Mississauga, ON, Canada). cDNA was then added to Quantabio PerfeCTa FastMix II and TaqMan human primers and probes for ASAH1, SPHK1, and RPLPO (Applied Biosystems, ThermoFisher Scientific, Waltham, MA, USA). qPCR was performed in duplicate on a Bio-Rad CFX96 Real-Time System, and Ct values were obtained in the Bio-Rad CFX Manager 3.1 program using regression as the Ct determination mode. Average Ct values were normalized against average expression of RPLPO (ΔCt values), then fold change between groups was analyzed using the 2−ΔΔCt calculation [26]. Samples with a Ct value > 0.5 between duplicates for a certain probe were reanalyzed. ## 2.8. Cell Line Culture and Treatments Choriocarcinoma JEG3 cells (ATCC® HTB-36™, ATCC), authenticated by short tandem repeat genotyping, were cultured on coverslips in 6-well plates in standard conditions (ambient air) at 37 °C in Eagle’s Minimum Essential Medium (EMEM) (ATCC, 30–2003) containing $10\%$ (v/v) FBS and penicillin–streptomycin (Wisent Inc., St Bruno, QC, Canada). Upon reaching $60\%$–$80\%$ confluency, cells were treated with either 25 mM of glucose (GLU), 0.85 μM of insulin solution (INS; 10 mg/mL in 25 mM HEPES, Catalog No. I9278, Sigma-Aldrich, Munich, Germany), glucose plus insulin (GLU+INS), or EMEM alone (VEH) for 24 h. After treatment, the cells were washed with phosphate-buffered saline (PBS) on ice and either collected at 4 °C in 1μM phenylmethylsulfonyl fluoride (PMSF) for analysis of SPHK enzyme activity and snap-frozen at −80 °C or fixed with $3.7\%$ (v/v) formaldehyde for immunofluorescence (IF) microscopy. ## 2.9. Statistics Statistical analyses were performed using GraphPad Prism 8 for Windows 64-bit (Version 8.1.2 [332], 6 May 2019) and IBM® SPSS® Statistics Version 25.0. Categorical variables were analyzed with the Chi-square test. For the comparison of continuous variables in two groups, Student’s t test, or Mann-Whitney U-test was utilized, as appropriate. In the case of hypotheses involving more than two comparison groups, One-way Analysis of Variance or Kruskal—Wallis tests were applied followed by Tukey’s or Dunn’s post-hoc tests, respectively. Data are presented as means (SD) or medians (IQR or range). Values of $p \leq 0.05$ were considered statistically significant. ## 3.1. Maternal and Perinatal Characteristics Maternal characteristics and perinatal outcomes of pregnant individuals with T1D and T2D and control pregnancies are displayed in Table 1. Maternal age and parity were similar in all groups. Age at diabetes diagnosis was lower and diabetes duration longer in women with T1D compared to those with T2D. Maternal pre-pregnancy BMI was higher in T2D pregnancies than in control women. Most individuals in all study groups gained more weight during pregnancy than recommended by the IOM. As expected, blood pressure levels after 20 weeks’ gestation and the frequencies of hypertensive complications, such as preeclampsia, were higher in women with diabetes. Of the T1D pregnancies with preeclampsia (T1DPE), two were early-onset syndromes occurring <34 weeks’ gestation and four were late-onset cases appearing ≥34 weeks’ gestation. In contrast, among the T2D women with preeclampsia (T2DPE), all patients had the early-onset subtype. Glycated hemoglobin (HbA1c) levels were similar in women with T1D and T2D. Most women were delivered by cesarean section without preceding labor in all groups. However, the median gestational age at birth was slightly lower in women with diabetes. Both the mean relative birth weight of the offspring and the frequency of fetal macrosomia (i.e., large-for-gestational age (LGA); birth weight z-score > +2.0 SD units) were higher in pregnancies affected by diabetes, whereas no macrosomic offspring were born to control women. In T1D pregnancies with LGA (T1DLGA), the average (range) relative birth weight of the fetuses was +2.95 (2.32–4.39) SD units, and in T2D pregnancies with LGA (T2DLGA), the average (range) relative birth weight was +3.29 (2.91–4.47) SD units. ## 3.2. Pre-Gestational Diabetes Is Associated with Decreased Placental Ceramide Considering our previous studies showing that pregnancy disorders are associated with changes in placental sphingolipid content [15,18,19,27], we first examined placental sphingolipids in T1D and T2D pregnancies. Since LC-MS/MS analysis revealed an overall lower total ceramide content in both T1D and T2D placentae relative to controls (Table 2), we pooled the data from all diabetes cases and observed a significant decrease in total placental ceramide concentrations in diabetes (T1D+T2D: diabetes mellitus (DM)) vs. control pregnancies (Figure 1A). Correspondingly, placental concentrations of sphingosine, a bioactive product of ceramide hydrolysis by ASAH1, displayed a trend toward increased levels, although the difference was not significant (Figure 1B). With respect to distinct ceramide species (14–24 carbon acyl chains), both T1D and T2D placentae were characterized by significantly lower concentrations of CER 18:0, 22:0, and 24:1 as compared to those from control pregnancies (Figure 1C). Despite lower ceramide levels in diabetic placentae, no changes in placental sphinganine or dihydroceramides, intermediates of the de novo pathway of ceramide synthesis, were observed between control vs. diabetic pregnancies (Supplementary Figure S1A,B), suggesting that de novo ceramide pathway was unaffected by diabetes. ## 3.3. Preeclampsia and Fetal Macrosomia Result in Distinct Alterations of the Placental Sphingolipidome in Type 1 and Type 2 Diabetic Pregnancies Placental concentrations of various ceramide species in T1D, T1DPE, and control pregnancies were next examined. No changes in ceramide species were found in placentae from T1D relative to CTR, except for the CER 24:1 concentration that was significantly lower in T1D placentae (Table 2, upper panel). In line with our earlier observations in preeclamptic pregnancies [15], total placental ceramide content as well as CER 22:0 and 24:0 species concentrations were increased in T1DPE, compared to T1D. CER 18:0 also showed elevated levels in T1DPE, when compared with T1D and CTR ($$p \leq 0.040$$), but pairwise post-hoc analyses yielded $p \leq 0.05$ (Table 2, upper panel). Furthermore, we found that when T1D pregnancies with a large-for-gestational age fetus (T1DLGA) pregnancies were excluded, also the median (IQR) CER 16:0 concentration was higher in T1DPE [9.73 ($\frac{5.38}{9.24}$); $$n = 6$$] as compared to T1D placentae [6.48 ($\frac{5.56}{7.06}$); $$n = 13$$] ($$p \leq 0.002$$). In contrast to T1DPE, placental levels of long- and very-long-chain ceramide species, including CER 18:0, CER 24:0, and CER 24:1, were significantly lower in T2DPE than in controls (Table 2, lower panel). Furthermore, when only preeclamptic diabetes patients were compared, lower total placental ceramide content (median (IQR)) was observed in T2DPE (35.60 ($\frac{18.90}{45.43}$)) vs. T1DPE (66.99 ($\frac{58.87}{71.28}$)) ($$p \leq 0.017$$). Since we have previously reported that pregnancies with fetal growth restriction exhibit reduced placental CER levels [18], we next explored whether placental ceramide concentrations are affected by fetal macrosomia. T1DPE and T2DPE placentae were excluded from the analysis. LC-MS/MS analyses revealed increased placental CER 14:0 and CER 16:0 content in T1DLGA pregnancies as compared to controls and T1D, respectively (Table 3). In T2D, no differences were seen in placental ceramide levels between T2DLGA pregnancies, T2D pregnancies without macrosomia, and controls (Table 3). ## 3.4. Placental Acid Ceramidase Is Upregulated in Type 1 Diabetes and in Type 2 Diabetes with and without Preeclampsia In line with our observation of lower placental ceramide levels in both types of pre-gestational diabetes, immunoblotting for ASAH1, the enzyme that breaks down ceramide to sphingosine, revealed significantly greater levels of this enzyme in both T1D and T2D (Figure 1D), compared to control pregnancies. ASAH1 levels varied markedly in the individual T1DPE samples, but on average trended higher; however, no statistical changes in ASAH1 protein levels were found in placental lysates from T1DPE patients relative to controls or T1D (Figure 1D). In contrast, ASAH1 protein content was upregulated in T2DPE placentae relative to controls (Figure 1D), in keeping with lower levels of very-long-chain ceramides in T2DPE. No changes in placental ASAH1 levels were observed with respect to fetal macrosomia (mean (SD) protein fold changes in CTR vs. T1DLGA: 1.00 (0.20 vs. 1.14 (0.45), $$p \leq 0.54$$; in T1D vs. T1DLGA: 1.41 (0.44) vs. 1.14 (0.45), $$p \leq 0.19$$; in CTR vs. T2DLGA: 1.00 (0.17) vs. 1.24 (0.41), $$p \leq 0.17$$; and in T2D vs. T2DLGA: 1.39 (0.37) vs. 1.24 (0.41), $$p \leq 0.55$$). Real-time PCR analysis demonstrated no changes in ASAH1 mRNA expression in diabetes vs. control placentae (Supplementary Figure S2A), suggesting that the increases in ASAH1 protein in diabetes are likely due to a reduced turnover of the enzyme. ## 3.5. Poor Maternal Glycemic Control Is Associated with Elevated Levels of Placental Ceramides Due to the observed differences in placental ceramide levels in diabetic pregnancies with and without preeclampsia and fetal macrosomia, we next explored the associations between common clinical risk factors of these disorders—obesity, gestational weight gain, and glycemic control—and placental ceramide. Since excess adiposity is known to associate with alterations of sphingolipid metabolism [28], we compared placental sphingolipids in non-diabetic obese (BMI 32.1–64.1 kg/m2) vs. non-obese control women (17.2–27.7 kg/m2). No differences in specific ceramide species nor in total placental ceramide concentrations were found in placentae from obese vs. non-obese normoglycemic women (Supplementary Table S1). Similarly, regarding gestational weight gain, we did not find placental ceramide differences between diabetic women who gained more or less than the IOM-recommended amount of weight during pregnancy. Finally, we examined sphingolipid changes in placentae from diabetic women with poor vs. better glycemic control in late pregnancy as identified by third-trimester HbA1c concentrations (HbA1c ≥ 64 mmol/L (≥$7.5\%$) vs. <64 mmol/L (<$7.5\%$)). Interestingly, we observed increased placental CER 16:0, 18:0, and 24:0 species in diabetic pregnancies with poor late-pregnancy glycemic control, which accounted for a significant increase in total placental ceramides in these pregnancies (Table 4). ## 3.6. Placental Production of Sphingoid-1-Phosphates Is Upregulated in Diabetic Pregnancies Complicated by Preeclampsia and Fetal Macrosomia Despite their low tissue concentrations, sphingoid-1-phosphates, namely sphingosine-1-phosphate (S1P) and sphinganine-1-phosphate (Sa1P), are highly bioactive lysosphingolipids that play an integral part in the sphingolipid cycle. Hence, we extended our analyses into the quantification of these lipid species in our placental samples using high-performance LC-MS/MS. No changes in total sphingoid-1-phosphates (S1P+Sa1P) were found in T1D or T2D placentae relative to controls (Table 2). However, when preeclampsia cases were excluded from the analysis, higher median (IQR) levels of sphingoid-1-phosphates were observed in T2D (0.112 ($\frac{0.054}{0.282}$)) vs. T1D (0.087 ($\frac{0.079}{0.125}$)) placentae ($$p \leq 0.041$$). In T1DPE, on the other hand, heightened total sphingoid-1-phosphate levels were seen as compared to T1D (Table 2). T2DPE placentae displayed a similar trend towards increased sphingoid-1-phosphate levels as compared to T2D (Table 2). Combined analysis showed increased sphingoid-1-phosphates in both T1DPE+T2DPE (0.270 ($\frac{0.167}{0.494}$)) vs. T1D+T2D (0.089 ($\frac{0.060}{0.250}$), $$p \leq 0.041$$) and vs. controls (0.093 ($\frac{0.067}{0.138}$), $$p \leq 0.026$$). Likewise, examination of sphingoid-1-phosphate concentrations revealed an enrichment of sphingoid-1-phosphates in T1DLGA placentae compared to controls or T1D (Table 3). No changes in sphingoid-1-phosphates content were seen with respect to LGA in T2D placentae. ## 3.7. Placental Sphingosine Kinase Activity Is Increased in Type 2 Diabetic Pregnancies and in Diabetic Pregnancies with Fetal Macrosomia Considering our findings of decreased placental ceramides in both T1D and T2D, and increased placental sphingoid-1-phosphate levels in T2D and both types of diabetes with LGA, we next examined the concentrations and activity of sphingosine kinase isoenzymes (SPHK$\frac{1}{2}$), which have the potential to turn the placental sphingolipid rheostat towards S1P production (Supplementary Figure S3). Western blotting showed increased placental SPHKs (SPHK1+SPHK2) concentrations in T2D, but not T1D, and in both T1DLGA, and T2DLGA relative to controls (Figure 2A). Since SPHK isoforms elicit partly overlapping and compensatory roles [29,30], we next quantified total SPHK levels in diabetic and non-diabetic placental tissue. Densitometric analyses confirmed a significant increase in SPHKs in T1DLGA and T2DLGA placentae compared to controls and in T2DLGA relative to T2D (Figure 2B). Despite increased sphingoid-1-phosphate levels in diabetic pregnancies with preeclampsia, no changes were seen in placental SPHK1 or SPHK2 enzyme levels in either diabetes type in the presence of preeclampsia. In line with increased placental SPHK protein levels and total sphingoid-1-phosphates in T2D, SPHK activity was significantly increased in T2D placentae, but not in T1D, as compared to controls (Figure 2C). Again, no differences in SPHK1 mRNA expression were observed between control, T1D, T1DLGA, T2D, or T2DLGA placentae (Supplementary Figure S2B). Thus, like ASAH1, the increase in SPHK1 protein in T2D and T1DLGA/T2DLGA placentae is likely due to a slower turnover of the enzyme. ## 3.8. In Vitro Exposure of Trophoblast Cells to High Glucose and Insulin Augments SPHK Activity To assess the effect of diabetic milieu on SPHK activity in vitro, we treated choriocarcinoma JEG-3 cells with 0.85 μM insulin (INS), 25 mM glucose (GLU), INS+GLU, or control vehicle EMEM (VEH) for 24 h. In line with the observations in T2D placentae, SPHK activity was markedly increased in cells treated with GLU or GLU+INS (Figure 3A). SPHK1 has been reported to primarily reside in the cytosol and plasma membrane, while SPHK2 is largely restricted to the mitochondria, ER, and nuclear compartments [31]. Consistent with increased enzyme activity, IF revealed an enhanced positive signal of both SPHK1 and SPHK2 in JEG3 cells following GLU+INS exposure (Figure 3B). SPHK1 signal co-localization with the plasma membrane marker ZO-1 was increased following GLU+INS treatment (Pearson’s correlation coefficient of 0.47 in GLU+INS treatment vs. 0.26 in vehicle-treated controls; Figure 3B), indicating a translocation to the plasma membrane following enzyme activation. Similarly, SPHK2 showed markedly enhanced co-localization with the mitochondrial marker TOM20 (Pearson’s correlation coefficient of 0.39 in GLU+INS treatment vs. 0.28 in Vehicle-treated controls; Figure 3C) and the ER marker calreticulin (Pearson’s correlation coefficient of 0.3350 in GLU+INS treatment vs. 0.2386 in vehicle-treated controls; Figure 3D). Taken together, these data indicate augmented expression and activation of SPHK enzymes in placental cells upon in vitro exposure to hyperglycemic and hyperinsulinemic conditions. ## 4. Discussion In the present study, we provide evidence of changes in placental ceramide metabolism in pregnancies complicated by pre-existing maternal diabetes. We found that T1D and T2D pregnancies are characterized by reduced placental ceramide content; however, in the presence of preeclampsia, T1D pregnancies exhibit placental ceramide enrichment. Of clinical relevance, we report that poor glycemic control, an important risk factor of fetal overgrowth and other adverse pregnancy outcomes [4], is associated with increased placental ceramides. In agreement with this, elevated levels of pro-apoptotic CER 16:0 were also observed in placentae from T1D pregnancies with a macrosomic fetus. Moreover, we show that placental SPHKs are upregulated in T2D and in both types of diabetes with fetal macrosomia, in line with augmented levels of sphingoid-1-phosphates. Finally, our in vitro experiments demonstrate that hyperglycemic and hyperinsulinemic conditions, typical of T1D and T2D, are plausible contributors to the upregulation of SPHK activity in diabetic placentae. To the best of our knowledge, our study provides the first examination of placental ceramide and lysosphingolipid metabolism in the context of pregnancies affected by T1D and T2D. Proper sphingolipid metabolism and signaling are vital for embryo implantation, organogenesis, and fetal growth [30,32,33,34,35,36]. In well-controlled GDM, we have recently reported decreased placental CER 16:0, 18:0, and 24:0 levels in conjunction with an upregulation of the ASAH1 enzyme [19]. Similarly, a recent study, using high-resolution mass spectrometry, showed lower total ceramides, and particularly reduced levels of CER 14:0, 16:0, and 18:0 species, in placentae from GDM vs. non-GDM pregnancies [37]. In contrast, another study using immunohistochemical analysis suggested increased ceramides in placental villous trophoblasts of insulin-treated GDM patients [38]. Our present findings on decreased placental ceramides in T1D and T2D and these published observations in GDM placentae collectively suggest that changes in this core sphingolipid typify diabetic placental metabolism. In the present study, placental ASAH1 upregulation was most prominent in T2D, which shares pathogenesis with GDM and obesity. However, our analysis of placental sphingolipid profiles in obese vs. non-obese control pregnancies suggested that the reduction in placental ceramides is not due to obesity but rather caused by mechanisms related to diabetes. This is in accordance with a previous report indicating that GDM is associated with a decrease in placental ceramides irrespective of maternal BMI [37]. The metabolic features of T2D are particularly likely to “fuel” the production of ceramide [39], and active compensatory mechanisms are needed to optimize feto-placental well-being. Interestingly, outside pregnancy, in other metabolically active tissues, the overexpression of ASAH1 [40] and SPHK [41] has been demonstrated to prevent ceramide accretion, leading to improved glucose and lipid metabolism. Considering that in GDM reduced placental ceramides were associated with the dominance of mitochondrial fusion [19], our current results warrant further investigations into the relevance of ceramide in placental mitochondrial function and dynamics in pre-gestational diabetes. Emerging evidence has elucidated the diverse roles of different sphingolipid species, which are dependent on the length of their carbon chain and intracellular localization, in human metabolism and its disorders [42,43]. Although tissue-specific exceptions exist, CER 16:0 and 18:0 species are generally pro-apoptotic, CER 16:0, 18:0, and 22:0 anti-proliferative, and CER 14:0 and 16:0 pro-autophagic in different cellular processes [43]. Our data indicate that these species were increased in the placentae of T1D women with preeclampsia. This is in line with our previous reports in early-onset preeclampsia, featuring CER-dependent increased trophoblast cell death, autophagy, and necroptosis rates [15,16]. Placental oxidative stress, which burdens both preeclamptic and diabetic placentae [44], likely promotes ceramide accumulation, as we have previously shown that the induction of oxidative stress in placental villous explants and JEG-3 cells upregulates CER 16:0 and 18:0 [15]. Since severe hyperglycemia in T1D increases the likelihood of oxidative stress [45], it could contribute to increased CER 16:0, 18:0, and 24:0 levels in women with poor late-pregnancy glycemic control. In contrast to our finding on T1DPE, placental ceramide enrichment did not characterize T2DPE, but, instead, lower levels of very-long-chain ceramides (CER 24:0 and 24:1) were found. This could be due to the upregulation of ASAH1 and SPHK which reduce ceramide and sphingosine levels, respectively. Even in non-diabetic pregnancies, preeclampsia is characterized by hyperinsulinemia, and other features of the metabolic syndrome [46], and it is plausible that these metabolic characteristics in T2DPE could contribute to ASAH1 [19] and SPHK [29] upregulation. On the other hand, since CER 24:0 and 24:1 have been reported to have pro-proliferative effects in some tissues, their reduced levels could also have opposite effects [9,43]. Decreased plasma concentrations of very-long-chain ceramides have also been implicated in diabetic nephropathy [47], which shares risk factors and pathophysiological background with preeclampsia, as both disorders feature endothelial dysfunction and glomerular damage leading to hypertension and proteinuria. Sphingosine kinases (SPHKs) are key regulators of sphingolipid metabolism, as they are responsible for the phosphorylation of sphingosine and sphinganine into potent signaling lysosphingolipids, S1P and Sa1P, respectively. S1P is a powerful mediator of cell proliferation and survival, vascularization, and angiogenesis, with many effects opposed to those of ceramide [14,48]. The functions of Sa1P are less clear, but mouse studies have demonstrated tissue-protective effects in ischemia and reperfusion injury of the liver [49]. SPHK1 and SPHK2 subcellular distribution contribute to the diverse functions [10] and compartmentalization of sphingoid-1-phosphate production [14,50]. Interestingly, while modest increases in SPHK2 levels, like those observed in the present study in diabetic placentae, have been reported to promote cell proliferation and survival, high levels of this enzyme may lead to pro-apoptotic signaling [50,51]. Congruent with our IF data in trophoblast cells, in other tissues, SPHK1 has been reported to locate in the cytosol from where it shuttles to the plasma membrane to elicit its effects, whereas SPHK2 resides mostly in the mitochondria, ER, and nucleus [31]. SPHKs are involved in a multitude of disease processes, especially those of hyperproliferative and inflammatory nature, e.g., cancer, insulin resistance, and diabetic vascular complications [7,14,48]. Although limited data is available on the significance of the SPHK-S1P axis in normal and complicated pregnancies, studies are accumulating on their crucial roles, e.g., in decidualization, placentation, placental vascularization, and early pregnancy success [36]. Intriguingly, we show enhanced SPHK expression and activity in T2D placentae and increased placental SPHK levels in both types of diabetic pregnancies affected by fetal macrosomia. As the cytotrophoblast layer is the most metabolically active cell layer of the placenta [52] and S1P has been reported to downregulate cytotrophoblast differentiation into syncytiotrophoblasts, increased S1P signaling could aid in preserving the metabolic capacity of these cells, to meet the high oxidative phosphorylation demands and ensure active substrate transfer to the fetus [35]. In contrast, no changes in SPHK protein were found in pregnancies of women with pre-existing diabetes who develop preeclampsia, despite elevated levels of sphingoid-1-phosphates. In preeclampsia, a high turnover of ceramide substrates into sphingoid-1-phosphates could be a protective mechanism preventing the accumulation of pro-death lipid metabolites. Despite the observed SPHK upregulation in both types of diabetic pregnancies with fetal macrosomia, increased sphingoid-1-phosphate levels were only recorded in T1DLGA placentae. This could result from the relatively low sample sizes available for these analyses. On the other hand, this finding is also in line with the high turnover rate and compartmentalization that characterize S1P synthesis [48]. Furthermore, in contrast to high S1P blood concentrations, S1P tissue concentrations are generally low, as much of the synthesized S1P is secreted [48] or broken down by S1P lyase, which is also an important regulator of S1P activity and tissue gradients [53,54]. In support of this notion, mouse studies suggest that SPHK overexpression may be coupled to enhanced S1P breakdown [41,55]. At least three out of the five G-protein coupled receptors that mediate S1P and Sa1P signaling have so far been identified in the placenta (S1PR1-S1PR3) [35], and the SPHK-S1P-S1PR axis has been implicated in a multitude of processes, from placental and embryonic development to trophoblast differentiation [56,57,58] and epigenetic modification [59]. Hence, the functions of both intracellular and excreted S1P in diabetic placentae and their potential significance in offspring health and development are an interesting avenue for future studies. Our in vitro experiments suggest that exposure of placental cells to hyperglycemia and, to a lesser extent, hyperinsulinemia, may in part contribute to the upregulation of the SPHK-sphingoid-1-phosphate axis. Although data are scarce in trophoblast cells, increased SPHK activity has been previously reported in the aorta, heart [60], and kidney [61] of streptozotocin-induced diabetic rats, and high glucose conditions have been shown to upregulate SPHK1 in human umbilical vein endothelial cells [60] and glomerular mesangial cells [61]. Moreover, a study in breast cancer cells has shown that insulin can also mediate mitogenic effects via both SPHK1 and SPHK2 [29]. These studies are in accordance with our findings in placental JEG3 cells. Placental hypervascularization and surface area enlargement are typical of diabetic placentae, especially in association with fetal macrosomia [62,63], and evidence suggests that high fetal and maternal insulin levels could be among the stimuli leading to these morphological changes [63,64]. Hence, it is plausible that SPHK upregulation dominated T2D pregnancies and pregnancies associated with fetal macrosomia, i.e., pregnancies characterized by high fetal and/or maternal insulin levels in addition to hyperglycemia. It is possible that in diabetic pregnancies affected by processes that restrict fetal growth (poorer placental vascularization, hypertension, diabetic vasculopathy), downregulation of SPHK triggers opposing processes, in line with what has been previously demonstrated in IUGR [18]. A strength of our study is the comparably large total number of placental samples obtained from control and diabetic pregnancies. On the other hand, considering that diabetes is a heterogenous condition, a clear limitation is that the sample sizes per complication subgroup (preeclampsia, macrosomia) were smaller. All patients were diagnosed and followed up at the same hospital (MSH) according to uniform clinical guidelines, and standardized BioBank protocols were utilized in the collection of clinical specimens. For the lipidomic analyses, we utilized a highly sensitive and selective LC-MS/MS technique, which enabled even the measurement of lipid species with low tissue concentrations. In western blotting experiments, we used the currently recommended approach of total protein normalization with stain-free technology, which has been shown more reliable as compared to housekeeping proteins as loading controls [23]. Among the weaknesses of our study is the limited availability of information on maternal glycaemic control and no data on maternal blood lipid profile or other metabolic parameters, such as those reflecting insulin resistance. Due to the use of pre-collected biobank samples, we also lacked fetal (cord) blood samples, which could have been analyzed for markers reflecting fetal metabolism, and information on neonatal anthropometric variables, apart from relative birth weight. We also lacked information on placental histopathological and morphological changes. ## 5. Conclusions In conclusion, our study provides novel evidence for distinct alterations of placental sphingolipid metabolism in T1D and T2D pregnancies. Collectively, our results suggest that the placental milieu in diabetes tilts the sphingolipid rheostat toward lower levels of ceramides, favoring the production of sphingoid-1-phosphates, especially in T2D (Graphical abstract). Furthermore, our results demonstrate that placental ceramide metabolism–with close linkages to central cellular signaling pathways such as those regulating cell fate and energy metabolism–is remarkably heterogenous in maternal pre-gestational diabetes, depending on the type of diabetes, associated complications, and clinical characteristics such as preeclampsia, fetal macrosomia, and poor glycemic control. Hence, changes in sphingolipid metabolism during pregnancy in a spectrum of diabetes may serve as footprints for predicting the onset of associated complications such as preeclampsia. This could be achieved by analyzing cargo of placental extracellular vesicles to identify changes in sphingolipids across gestation that could be used to predict diabetes-associated adverse outcomes. Considering the significance of in utero metabolic milieu in the programming of offspring metabolism and cardiovascular health, our results call for future studies addressing the role of these placental metabolic alterations in specific short- and long-term outcomes. ## References 1. Burton G.J., Fowden A.L., Thornburg K.L.. Placental Origins of Chronic Disease. Physiol. Rev. (2016) 96 1509-1565. DOI: 10.1152/physrev.00029.2015 2. Jauniaux E., Poston L., Burton G.J.. Placental-related diseases of pregnancy: Involvement of oxidative stress and implications in human evolution. Hum. Reprod. Update (2006) 12 747-755. DOI: 10.1093/humupd/dml016 3. Murphy H.R., Bell R., Cartwright C., Curnow P., Maresh M., Morgan M., Sylvester C., Young B., Lewis-Barned N.. Improved pregnancy outcomes in women with type 1 and type 2 diabetes but substantial clinic-to-clinic variations: A prospective nationwide study. Diabetologia (2017) 60 1668-1677. DOI: 10.1007/s00125-017-4314-3 4. Klemetti M.M., Laivuori H., Tikkanen M., Nuutila M., Hiilesmaa V., Teramo K.. White’s classification and pregnancy outcome in women with type 1 diabetes: A population-based cohort study. Diabetologia (2016) 59 92-100. DOI: 10.1007/s00125-015-3787-1 5. Desoye G.. The Human Placenta in Diabetes and Obesity: Friend or Foe? The 2017 Norbert Freinkel Award Lecture. Diabetes Care (2018) 41 1362-1369. DOI: 10.2337/dci17-0045 6. Yaribeygi H., Bo S., Ruscica M., Sahebkar A.. Ceramides and diabetes mellitus: An update on the potential molecular relationships. Diabet. Med. (2019) 37 11-19. DOI: 10.1111/dme.13943 7. Ng M.L., Wadham C., Sukocheva O.A.. The role of sphingolipid signalling in diabetes-associated pathologies (Review). Int. J. Mol. Med. (2017) 39 243-252. DOI: 10.3892/ijmm.2017.2855 8. Chavez J.A., Summers S.A.. A ceramide-centric view of insulin resistance. Cell Metab. (2012) 15 585-594. DOI: 10.1016/j.cmet.2012.04.002 9. Hannun Y.A., Obeid L.M.. Sphingolipids and their metabolism in physiology and disease. Nat. Rev. Mol. Cell Biol. (2018) 19 175-191. DOI: 10.1038/nrm.2017.107 10. Hatoum D., Haddadi N., Lin Y., Nassif N.T., McGowan E.M.. Mammalian sphingosine kinase (SphK) isoenzymes and isoform expression: Challenges for SphK as an oncotarget. Oncotarget (2017) 8 36898-36929. DOI: 10.18632/oncotarget.16370 11. Hannun Y.A., Obeid L.M.. Many ceramides. J. Biol. Chem. (2011) 286 27855-27862. DOI: 10.1074/jbc.R111.254359 12. Baranowski M., Blachnio-Zabielska A.U., Charmas M., Helge J.W., Dela F., Ksiazek M., Dlugolecka B., Klusiewicz A., Chabowski A., Gorski J.. Exercise increases sphingoid base-1-phosphate levels in human blood and skeletal muscle in a time- and intensity-dependent manner. Eur. J. Appl. Physiol. (2015) 115 993-1003. DOI: 10.1007/s00421-014-3080-x 13. Ying L., Tippetts T.S., Chaurasia B.. Ceramide dependent lipotoxicity in metabolic diseases. Nutr. Healthy Aging (2019) 5 1-12. DOI: 10.3233/NHA-170032 14. Pyne S., Adams D.R., Pyne N.J.. Sphingosine 1-phosphate and sphingosine kinases in health and disease: Recent advances. Prog. Lipid Res. (2016) 62 93-106. DOI: 10.1016/j.plipres.2016.03.001 15. Melland-Smith M., Ermini L., Chauvin S., Craig-Barnes H., Tagliaferro A., Todros T., Post M., Caniggia I.. Disruption of sphingolipid metabolism augments ceramide-induced autophagy in preeclampsia. Autophagy (2015) 11 653-669. DOI: 10.1080/15548627.2015.1034414 16. Bailey L.J., Alahari S., Tagliaferro A., Post M., Caniggia I.. Augmented trophoblast cell death in preeclampsia can proceed via ceramide-mediated necroptosis. Cell Death Dis. (2017) 8 e2590. DOI: 10.1038/cddis.2016.483 17. Ausman J., Abbade J., Ermini L., Farrell A., Tagliaferro A., Post M., Caniggia I.. Ceramide-induced BOK promotes mitochondrial fission in preeclampsia. Cell Death Dis. (2018) 9 298. DOI: 10.1038/s41419-018-0360-0 18. Chauvin S., Yinon Y., Xu J., Ermini L., Sallais J., Tagliaferro A., Todros T., Post M., Caniggia I.. Aberrant TGFbeta Signalling Contributes to Dysregulation of Sphingolipid Metabolism in Intrauterine Growth Restriction. J. Clin. Endocrinol. Metab. (2015) 100 E986-E996. DOI: 10.1210/jc.2015-1288 19. Abbade J., Klemetti M.M., Farrell A., Ermini L., Gillmore T., Sallais J., Tagliaferro A., Post M., Caniggia I.. Increased placental mitochondrial fusion in gestational diabetes mellitus: An adaptive mechanism to optimize feto-placental metabolic homeostasis?. BMJ Open Diabetes Res. Care (2020) 8 e000923. DOI: 10.1136/bmjdrc-2019-000923 20. ACOG Committee opinion no. 548: Weight gain during pregnancy. Obstet. Gynecol. (2013) 121 210-212. DOI: 10.1097/01.AOG.0000425668.87506.4c 21. Hypertension in pregnancy. Report of the American College of Obstetricians and Gynecologists’ Task Force on Hypertension in Pregnancy. Obstet. Gynecol. (2013) 122 1122-1131. DOI: 10.1097/01.AOG.0000437382.03963.88 22. Kramer M.S., Platt R.W., Wen S.W., Joseph K.S., Allen A., Abrahamowicz M., Blondel B., Breart G.. A new and improved population-based Canadian reference for birth weight for gestational age. Pediatrics (2001) 108 E35. DOI: 10.1542/peds.108.2.e35 23. Ghosh R., Gilda J.E., Gomes A.V.. The necessity of and strategies for improving confidence in the accuracy of western blots. Expert Rev. Proteom. (2014) 11 549-560. DOI: 10.1586/14789450.2014.939635 24. Lanoix D., St-Pierre J., Lacasse A.A., Viau M., Lafond J., Vaillancourt C.. Stability of reference proteins in human placenta: General protein stains are the benchmark. Placenta (2012) 33 151-156. DOI: 10.1016/j.placenta.2011.12.008 25. Farrell A., Alahari S., Ermini L., Tagliaferro A., Litvack M., Post M., Caniggia I.. Faulty oxygen sensing disrupts angiomotin function in trophoblast cell migration and predisposes to preeclampsia. JCI Insight (2019) 4 127009. DOI: 10.1172/jci.insight.127009 26. Livak K.J., Schmittgen T.D.. Analysis of relative gene expression data using real-time quantitative PCR and the 2. Methods (2001) 25 402-408. DOI: 10.1006/meth.2001.1262 27. Ermini L., Ausman J., Melland-Smith M., Yeganeh B., Rolfo A., Litvack M.L., Todros T., Letarte M., Post M., Caniggia I.. A Single Sphingomyelin Species Promotes Exosomal Release of Endoglin into the Maternal Circulation in Preeclampsia. Sci. Rep. (2017) 7 12172. DOI: 10.1038/s41598-017-12491-4 28. Meikle P.J., Summers S.A.. Sphingolipids and phospholipids in insulin resistance and related metabolic disorders. Nat. Rev. Endocrinol. (2017) 13 79-91. DOI: 10.1038/nrendo.2016.169 29. Dai L., Qi Y., Chen J., Kaczorowski D., Di W., Wang W., Xia P.. Sphingosine kinase (SphK) 1 and SphK2 play equivalent roles in mediating insulin’s mitogenic action. Mol. Endocrinol. (2014) 28 197-207. DOI: 10.1210/me.2013-1237 30. Mizugishi K., Li C., Olivera A., Bielawski J., Bielawska A., Deng C.X., Proia R.L.. Maternal disturbance in activated sphingolipid metabolism causes pregnancy loss in mice. J. Clin. Investig. (2007) 117 2993-3006. DOI: 10.1172/JCI30674 31. Neubauer H.A., Pitson S.M.. Roles, regulation and inhibitors of sphingosine kinase 2. FEBS J. (2013) 280 5317-5336. DOI: 10.1111/febs.12314 32. Kono M., Mi Y., Liu Y., Sasaki T., Allende M.L., Wu Y.P., Yamashita T., Proia R.L.. The sphingosine-1-phosphate receptors S1P1, S1P2, and S1P3 function coordinately during embryonic angiogenesis. J. Biol. Chem. (2004) 279 29367-29373. DOI: 10.1074/jbc.M403937200 33. Mizugishi K., Inoue T., Hatayama H., Bielawski J., Pierce J.S., Sato Y., Takaori-Kondo A., Konishi I., Yamashita K.. Sphingolipid pathway regulates innate immune responses at the fetomaternal interface during pregnancy. J. Biol. Chem. (2015) 290 2053-2068. DOI: 10.1074/jbc.M114.628867 34. Eliyahu E., Park J.H., Shtraizent N., He X., Schuchman E.H.. Acid ceramidase is a novel factor required for early embryo survival. FASEB J. (2007) 21 1403-1409. DOI: 10.1096/fj.06-7016com 35. Nagamatsu T., Iwasawa-Kawai Y., Ichikawa M., Kawana K., Yamashita T., Osuga Y., Fujii T., Schust D.J.. Emerging roles for lysophospholipid mediators in pregnancy. Am. J. Reprod. Immunol. (2014) 72 182-191. DOI: 10.1111/aji.12239 36. Fakhr Y., Brindley D.N., Hemmings D.G.. Physiological and pathological functions of sphingolipids in pregnancy. Cell. Signal. (2021) 85 110041. DOI: 10.1016/j.cellsig.2021.110041 37. Pinto G.D.A., Murgia A., Lai C., Ferreira C.S., Goes V.A., Guimarães D.A.B., Ranquine L.G., Reis D.L., Struchiner C.J., Griffin J.L.. Sphingolipids and acylcarnitines are altered in placentas from women with gestational diabetes mellitus. Br. J. Nutr. (2022) 1-12. DOI: 10.1017/S000711452200397X 38. Mejia J.F., Hirschi K.M., Tsai K.Y.F., Long M.G., Tullis B.C., Bitter E.E.K., Bikman B.T., Reynolds P.R., Arroyo J.A.. Differential placental ceramide levels during gestational diabetes mellitus (GDM). Reprod. Biol. Endocrinol. (2019) 17 81. DOI: 10.1186/s12958-019-0523-6 39. Field B.C., Gordillo R., Scherer P.E.. The Role of Ceramides in Diabetes and Cardiovascular Disease Regulation of Ceramides by Adipokines. Front. Endocrinol. (2020) 11 569250. DOI: 10.3389/fendo.2020.569250 40. Xia J.Y., Holland W.L., Kusminski C.M., Sun K., Sharma A.X., Pearson M.J., Sifuentes A.J., McDonald J.G., Gordillo R., Scherer P.E.. Targeted Induction of Ceramide Degradation Leads to Improved Systemic Metabolism and Reduced Hepatic Steatosis. Cell Metab. (2015) 22 266-278. DOI: 10.1016/j.cmet.2015.06.007 41. Bruce C.R., Risis S., Babb J.R., Yang C., Kowalski G.M., Selathurai A., Lee-Young R.S., Weir J.M., Yoshioka K., Takuwa Y.. Overexpression of sphingosine kinase 1 prevents ceramide accumulation and ameliorates muscle insulin resistance in high-fat diet-fed mice. Diabetes (2012) 61 3148-3155. DOI: 10.2337/db12-0029 42. Turpin-Nolan S.M., Brüning J.C.. The role of ceramides in metabolic disorders: When size and localization matters. Nat. Rev. Endocrinol. (2020) 16 224-233. DOI: 10.1038/s41574-020-0320-5 43. Grosch S., Schiffmann S., Geisslinger G.. Chain length-specific properties of ceramides. Prog. Lipid Res. (2012) 51 50-62. DOI: 10.1016/j.plipres.2011.11.001 44. Lappas M., Hiden U., Desoye G., Froehlich J., Hauguel-de Mouzon S., Jawerbaum A.. The role of oxidative stress in the pathophysiology of gestational diabetes mellitus. Antioxid. Redox Signal. (2011) 15 3061-3100. DOI: 10.1089/ars.2010.3765 45. Shah M.S., Brownlee M.. Molecular and Cellular Mechanisms of Cardiovascular Disorders in Diabetes. Circ. Res. (2016) 118 1808-1829. DOI: 10.1161/CIRCRESAHA.116.306923 46. Kaaja R., Laivuori H., Laakso M., Tikkanen M.J., Ylikorkala O.. Evidence of a state of increased insulin resistance in preeclampsia. Metab. Clin. Exp. (1999) 48 892-896. DOI: 10.1016/S0026-0495(99)90225-1 47. Klein R.L., Hammad S.M., Baker N.L., Hunt K.J., Al Gadban M.M., Cleary P.A., Virella G., Lopes-Virella M.F.. Decreased plasma levels of select very long chain ceramide species are associated with the development of nephropathy in type 1 diabetes. Metab. Clin. Exp. (2014) 63 1287-1295. DOI: 10.1016/j.metabol.2014.07.001 48. Cartier A., Hla T.. Sphingosine 1-phosphate: Lipid signaling in pathology and therapy. Science (2019) 366. DOI: 10.1126/science.aar5551 49. Park S.W., Kim M., Chen S.W., Brown K.M., D’Agati V.D., Lee H.T.. Sphinganine-1-phosphate protects kidney and liver after hepatic ischemia and reperfusion in mice through S1P1 receptor activation. Lab. Investig. (2010) 90 1209-1224. DOI: 10.1038/labinvest.2010.102 50. Maceyka M., Sankala H., Hait N.C., Le Stunff H., Liu H., Toman R., Collier C., Zhang M., Satin L.S., Merrill A.H.. SphK1 and SphK2, sphingosine kinase isoenzymes with opposing functions in sphingolipid metabolism. J. Biol. Chem. (2005) 280 37118-37129. DOI: 10.1074/jbc.M502207200 51. Neubauer H.A., Pham D.H., Zebol J.R., Moretti P.A., Peterson A.L., Leclercq T.M., Chan H., Powell J.A., Pitman M.R., Samuel M.S.. An oncogenic role for sphingosine kinase 2. Oncotarget (2016) 7 64886-64899. DOI: 10.18632/oncotarget.11714 52. Kolahi K.S., Valent A.M., Thornburg K.L.. Cytotrophoblast, Not Syncytiotrophoblast, Dominates Glycolysis and Oxidative Phosphorylation in Human Term Placenta. Sci. Rep. (2017) 7 42941. DOI: 10.1038/srep42941 53. Saba J.D.. Fifty years of lyase and a moment of truth: Sphingosine phosphate lyase from discovery to disease. J. Lipid Res. (2019) 60 456-463. DOI: 10.1194/jlr.S091181 54. Dixit D., Okuniewska M., Schwab S.R.. Secrets and lyase: Control of sphingosine 1-phosphate distribution. Immunol. Rev. (2019) 289 173-185. DOI: 10.1111/imr.12760 55. Oyama O., Sugimoto N., Qi X., Takuwa N., Mizugishi K., Koizumi J., Takuwa Y.. The lysophospholipid mediator sphingosine-1-phosphate promotes angiogenesis in vivo in ischaemic hindlimbs of mice. Cardiovasc. Res. (2008) 78 301-307. DOI: 10.1093/cvr/cvn002 56. Johnstone E.D., Chan G., Sibley C.P., Davidge S.T., Lowen B., Guilbert L.J.. Sphingosine-1-phosphate inhibition of placental trophoblast differentiation through a G(i)-coupled receptor response. J. Lipid Res. (2005) 46 1833-1839. DOI: 10.1194/jlr.M500095-JLR200 57. Zhang J., Dunk C.E., Lye S.J.. Sphingosine signalling regulates decidual NK cell angiogenic phenotype and trophoblast migration. Hum. Reprod. (2013) 28 3026-3037. DOI: 10.1093/humrep/det339 58. Hemmings D.G., Hudson N.K., Halliday D., O’Hara M., Baker P.N., Davidge S.T., Taggart M.J.. Sphingosine-1-phosphate acts via rho-associated kinase and nitric oxide to regulate human placental vascular tone. Biol. Reprod. (2006) 74 88-94. DOI: 10.1095/biolreprod.105.043034 59. Hait N.C., Allegood J., Maceyka M., Strub G.M., Harikumar K.B., Singh S.K., Luo C., Marmorstein R., Kordula T., Milstien S.. Regulation of histone acetylation in the nucleus by sphingosine-1-phosphate. Science (2009) 325 1254-1257. DOI: 10.1126/science.1176709 60. Wang L., Xing X.P., Holmes A., Wadham C., Gamble J.R., Vadas M.A., Xia P.. Activation of the sphingosine kinase-signaling pathway by high glucose mediates the proinflammatory phenotype of endothelial cells. Circ. Res. (2005) 97 891-899. DOI: 10.1161/01.RES.0000187469.82595.15 61. Lan T., Liu W., Xie X., Xu S., Huang K., Peng J., Shen X., Liu P., Wang L., Xia P.. Sphingosine kinase-1 pathway mediates high glucose-induced fibronectin expression in glomerular mesangial cells. Mol. Endocrinol. (2011) 25 2094-2105. DOI: 10.1210/me.2011-0095 62. Huynh J., Dawson D., Roberts D., Bentley-Lewis R.. A systematic review of placental pathology in maternal diabetes mellitus. Placenta (2015) 36 101-114. DOI: 10.1016/j.placenta.2014.11.021 63. Cvitic S., Desoye G., Hiden U.. Glucose, insulin, and oxygen interplay in placental hypervascularisation in diabetes mellitus. BioMed Res. Int. (2014) 2014 145846. DOI: 10.1155/2014/145846 64. Lassance L., Miedl H., Absenger M., Diaz-Perez F., Lang U., Desoye G., Hiden U.. Hyperinsulinemia stimulates angiogenesis of human fetoplacental endothelial cells: A possible role of insulin in placental hypervascularization in diabetes mellitus. J. Clin. Endocrinol. Metab. (2013) 98 E1438-E1447. DOI: 10.1210/jc.2013-1210
--- title: Stability Determination of Intact Humanin-G with Characterizations of Oxidation and Dimerization Patterns authors: - Mustafa Ozgul - Anthony B. Nesburn - Nader Nasralla - Benjamin Katz - Enes Taylan - Baruch D. Kuppermann - Maria Cristina Kenney journal: Biomolecules year: 2023 pmcid: PMC10046509 doi: 10.3390/biom13030515 license: CC BY 4.0 --- # Stability Determination of Intact Humanin-G with Characterizations of Oxidation and Dimerization Patterns ## Abstract Humanin is the first identified mitochondrial-derived peptide. Humanin-G (HNG) is a variant of Humanin that has significantly higher cytoprotective properties. Here, we describe the stability features of HNG in different conditions and characterize HNG degradation, oxidation, and dimerization patterns over short-term and long-term periods. HNG solutions were prepared in high-performance liquid chromatography (HPLC) water or MO formulation and stored at either 4 °C or 37 °C. Stored HNG samples were analyzed using HPLC and high-resolution mass spectrometry (HRMS). Using HPLC, full-length HNG peptides in HPLC water decreased significantly with time and higher temperature, while HNG in MO formulation remained stable up to $95\%$ at 4 °C on day 28. HNG peptides in HPLC water, phosphate-buffered saline (PBS) and MO formulation were incubated at 37 °C and analyzed at day 1, day 7 and day 14 using HRMS. Concentrations of full-length HNG peptide in HPLC water and PBS declined over time with a corresponding appearance of new peaks that increased over time. These new peaks were identified to be singly oxidized HNG, doubly oxidized HNG, homodimerized HNG, singly oxidized homodimerized HNG, and doubly oxidized homodimerized HNG. Our results may help researchers improve the experimental design to further understand the critical role of HNG in human diseases. ## 1. Introduction Mitochondrial DNA (mtDNA) is double-stranded, circular DNA comprised of 16,569 nucleotide pairs that represents 37 genes encoding for 13 peptides, 22 transfer RNAs, and 2 ribosomal RNAs [1,2]. Mitochondria-derived peptides (MDPs), encoded by the human mtDNA, play essential roles in many cellular physiological processes that can affect aging and disease progression [3,4,5,6,7,8,9]. Exploring mitochondrial biology, several MDPs, consisting of 16–38 amino acids, have been identified [10]. Humanin (HN), the first identified MDP, contains 24 amino acids (2687.3 Da) [5] and has neuroprotective and anti-apoptotic properties in in vitro and in vivo models [5,11,12,13,14]. The serum HN levels decrease significantly with age and are associated with age-related diseases in rodent animal models and human clinical studies [15,16]. HN peptides protect against neurotoxicity in Alzheimer’s disease and suppress amyloid-beta-induced neuronal death in vitro [17]. The administration of exogenous HN peptides provides cytoprotective effects in Type-2 diabetes rat models [15] as well as myocardial and cerebral ischemia [18,19] and atherosclerosis [20] mouse models [3]. Humanin-G (HNG) is an HN derivative with a S14G substitution exhibiting 1000-fold more potent cytoprotective properties than HN, and it also demonstrates therapeutic potential for multiple diseases [5,21]. Similar to Humanin, HNG (2657.3 Da) inhibits amyloid-beta (Aβ)-induced death in primary neurons in vitro and demonstrates cytoprotective effects for myocardial ischemia-reperfusion injury in animal models [18,22,23,24]. HNG also has antitumor effects as shown in neuroblastoma tumor xenograft experiments [25]. The effect was linked to reduced angiogenesis and increased tumor cell apoptosis [25]. Recently, we investigated the effect of HNG in a transmitochondrial cybrid model for age-related macular degeneration (AMD), which is the most common cause of visual impairment in the elderly population. Cybrids are cell lines with identical nuclei but with mitochondria from different individuals with AMD or age-matched normal subjects. The AMD cybrids treated with HNG showed significantly increased levels of humanin receptor proteins and decreased levels of RNA/proteins involved in apoptosis, autophagy, and ER stress pathways [1]. HNG-treated AMD cybrids showed significantly lower levels of cell death and improved functions in vitro [1]. However, conducting in vivo studies has been challenging due to the instability of the HNG peptide because of its tendency to rapidly degrade, oxidize and dimerize. The development of novel formulations to enhance the stability of HNG peptides is a critical first step toward the therapeutic delivery of HNG in retinal degeneration models in vivo and for future clinical investigations to treat several age-related diseases such as AMD, Alzheimer’s disease, and diabetic retinopathy. To the best of our knowledge, this is the first study that accurately analyzes the stability features of HNG and identifies its fragments and their therapeutic potential using high-performance liquid chromatography (HPLC) and high-resolution mass spectrometry (HRMS). We also developed a stabilization formula (MO formulation) that significantly improves the HNG peptide stability when stored long term and at 37 °C. ## 2.1. Chemicals and Materials The HNG peptide (Catalog No: AS-60887) was purchased from AnaSpec Inc. (Fremont, CA, US). Acetonitrile, HPLC water, LC-MS water, and formic acid were purchased from Fisher Scientific (Waltham, MA, USA). Analytical grade solvents were used in all experiments. ## 2.2. Physiochemical Properties An ExPASy ProtParam bioinformatics software tool was used to determine structural prediction including the instability index value, grand average of hydropathy value (GRAVY), and theoretical isoelectric point (pI). The instability index represents the prediction of peptide instability. When a peptide’s instability index is less than 40, the peptide is classified as stable, and if it is higher than 40, the peptide is designated as unstable. The GRAVY method predicts peptide hydrophilicity and hydrophobicity. GRAVY’s positive values and negative values represent the hydrophobic and hydrophilic structures, respectively [26,27,28,29]. The ExPASy PeptideCutter bioinformatics software tool was utilized to predict potential cleave sites, cleaving enzymes, and chemicals in the HNG peptide [30]. Theoretical charge of HNG peptide over pH change was analyzed using the peptide analysis tool in the Thermo-Fisher Scientific website. ## 2.3. HNG Solution Preparation and Storage For HPLC studies, HNG solutions were prepared at 125 μg/mL in HPLC water and at 112.5 μg/mL in the stabilization formula (MO formulation). The MO formulation, a proprietary solution, is a colorless liquid and includes organic acid (pH = 2.4–2.5) that has been found to be non-toxic to cells. We prepared duplicate HNG peptide solutions that were stored at two different temperatures (4 °C and 37 °C). For long-term stability analyses, HNG solutions were stored for 11 months at 4 °C. HNG solutions were filtered using 0.22 μm filters before HPLC analysis. The stability features of the stored HNG solutions were evaluated by HPLC at seven different time-points (6 h, 21 h, 33 h, day 3, day 7, day 14, and day 28). For HRMS studies, samples with a concentration of 30 μM HNG were prepared from the stored HPLC water and MO-formulation to analyze 11-month-old HNG products. ## 2.4. High-Performance Liquid Chromatography The Agilent 1100 system with an Agilent Eclipse XDB-C8 5 μm, 4.6 × 150 mm HPLC column was used to achieve liquid chromatographic separation. HNG was monitored at a wavelength of 200 nm using a Diode Array Detector. Gradient elution was performed with solvent A (water with $0.05\%$ trifluoroacetic acid) and solvent B (acetonitrile with $0.05\%$ trifluoroacetic acid). The gradient started at $30\%$ solvent B with a ramp to $60\%$ solvent B in a period of 10 min. At 12 min, the gradients begin to return to $30\%$ solvent B in 0.1 min. The column was equilibrated from 12.1 min to 20 min at $30\%$ solvent B (Table S1). The flow rate was set to 1 mL/min, and 25 μL of the sample was injected into the column. The column temperature was set at 40 °C. ## 2.5. High-Resolution Mass Spectrometry The Waters® Acquity H-class ultra-performance liquid chromatography (UPLC) method was run on a Waters BEH C4 column 300 Å, 1.7 μm, 50 mm × 2.1 using 25 min gradient at 0.3 mL/min from $97\%$ A to $97\%$ B, where A is $0.1\%$ formic acid in water and B is $100\%$ ACN (Table S2). Mass spectrometric analysis was performed using a XEVO G2-XS Quadrupole Time-of-Flight (QTof) mass spectrometer equipped with StepWave ion optics (Waters Corp., MA, USA). The positive electrospray ionization mode was utilized. Measurements were conducted using an ion source desolvation temperature of 350 °C and a cone voltage of 40 V. Argon was utilized as damping gas in the Collision-Induced Dissociation (CID) experiments. A capillary transfer temperature of 300 °C and a spray voltage of 3.0 kV were used to accomplish ionization. A resolution of 30,000 Full Width at Half Maximum (FWHM) was used for a full scan experiment within a range of m/z 100–2000 in addition to 15,000 FWHM with an isolation window adjusted to m/z 2.0 for Parallel Reaction Monitoring (PRM) mode. The instrument was operated in MSE continuum mode, which alternates low-energy (6 V) and high-energy (40 V) scans every 0.5 sec. Leucine Enkephalin was used as a lock mass for nominal mass correction, and a CsNaI ladder was used for detector calibration. Mass to Charge Calculation Formula for Dimerized Form HNG and their Fragments. The disulfide bridge causes a mass shift of −2 Da. The monoisotopic HNG molecular weight (Mw) is 2657.3. Mass to Charge Calculation Formula for oxidized form HNG and their fragments. Oxidation causes a mass shift of +16 Da. The monoisotopic HNG molecular weight (Mw) is 2657.3. Mass to Charge Ratio=Peptide Mw+Oxidation Mass+Number of ProtonationCharge State of Oxidized Peptide ## 2.6. Data Analysis Collected data were analyzed using MassLynx (version 4.2, 2016) and BiopharmaLynx (version 4.0.27.10, 2015) software programs provided by the Waters Company (Milford, MA, USA). ## 3. Results The molecular structure, molecular weights, charges, and amino acid sequences of HNG peptides were characterized using UCSF Chimera software and the ExPASy ProtParam bioinformatics software tool (Figure 1). In Figure 1A, the predicted structure of HNG shows the proposed length and width as approximately 4.8 nm and approximately 1.8 nm, respectively. In Figure 1B, the isoelectric point is 10. 1, indicating it is a basic peptide. The net charge of HNG peptide at pH 7 is 1.9, indicating it is a soluble peptide in neutral water. In Figure 1C, HNG has an Instability Index of 91.33, suggesting it is an unstable peptide. The GRAVY value is 0.358, indicating a hydrophobic property. We used the ExPASy PeptideCutter bioinformatics software to predict potential cleave sites, cleaving enzymes, and chemicals in the HNG peptide [30]. The functions, hydropathicity, name of cleaving enzymes/chemicals, and properties of each amino acid of HNG peptide are given in Table 1. ## 3.1. Short-Term Stability of HNG Peptide in Different Conditions We analyzed the stability of the HNG peptide stored in HPLC water (Figure 2A–I, Left Panel) and MO formulation (Figure 2J–Q, Right Panel) at 4 °C and 37 °C using HPLC at day 0, day 1, day 3, day 7, day 14, and day 28. The concentration of full-length HNG decreases over time, while the concentration of the HNG products (HNG-Pd) simultaneously increased (Figure 2A–R). Compared to the MO formulation, we found that the HNG peptide was sensitive to storage temperature and duration. At 21 h after storage, the full-length HNG peptide level in HPLC water stored at 4 °C was $90\%$ compared to $67\%$ at 37 °C (Figure 3). After 33 h of storage, nearly half of the full-length HNG peptides ($52\%$) was found at 37 °C in HPLC water, indicating its half-life at body temperature (Figure 3). By day 28, the full-length HNG stored in HPLC water was further declined to concentrations of $11\%$ at 4 °C and $5\%$ at 37 °C (Figure 3). When we evaluated the HNG stability stored in the MO formulation, remarkably, we found that the full-length HNG peptide remained stable up to $95\%$ at 4 °C on day 28. Even at 37 °C, the full-length HNG peptide concentration was significantly higher when stored in the MO formulation compared to HPLC water ($67\%$ versus $11\%$, respectively) (Figure 3). These results show that the HNG peptide is highly unstable when stored in HPLC water, and the stability of the HNG peptide can be successfully improved when stored in our newly developed MO formulation. Overall, the stability of the full-length HNG peptide in HPLC water and MO formulation was measured using HPLC at 4 °C and 37 °C over a 28-day period, and the results showed that the full-length HNG peptide in HPLC-grade water is not stable at 4 °C and 37 °C. Based on the results of these HPLC studies, we saw indications, represented by the other peaks in the graph, of unidentified amino acid sequences of the HNG products. To identify the composition of HNG products that occurred in the three different solutions (HPLC-grade water, PBS, and MO formulation), we performed experiments using UPLC-HRMS. Collected UPLC-HRMS data were analyzed using the BiopharmaLynx program to identify the peptide sequence of each peak (Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13, Tables S3 and S4). ## 3.2. Characterization of Full-Length HNG, Its Oxidized and Dimerized Forms To characterize HNG products using UPLC-HRMS, we prepared a solution of HNG peptide in HPLC-grade water, PBS and MO formulation. Using UPLC-HRMS, the full-length HNG, singly oxidized full-length HNG (SOx-HNG), doubly oxidized full-length HNG (Dox-HNG), singly oxidized dimerized HNG (SOx-DM-HNG) and doubly oxidized dimerized HNG (DOx-DM-HNG) were identified. The amino acid sequences and m/z ratio of full-length HNG, singly oxidized HNG, doubly oxidized HNG in PBS on day 14 at 37 °C are represented (Figure 4). In Figure 4A,B, multiple charged full-length HNG peptides (MAPRGFSCLLLLTGEIDLPVKRRA) were observed at m/z 532.31 ($z = 5$), 665.13 ($z = 4$), and 886.50 ($z = 3$). In Figure 4C,D, multiple charged singly oxidized (methionine) full-length HNG peptides (SOx-HNG) were observed at m/z 535.50 ($z = 5$), 669.13 ($z = 4$), and 891.83 ($z = 3$). In Figure 4E,F, multiple charged doubly oxidized (methionine and cysteine) full-length HNG peptides (Dox-HNG) were observed at m/z 538.70 ($z = 5$), 673.13 ($z = 4$), and 897.17 ($z = 3$). In Figure 5, the amino acid sequences and m/z ratio of intact homodimerized (cysteine-cysteine disulfide bone) HNG (DM-HNG), singly oxidized homodimerized HNG, and doubly oxidized homodimerized HNG in PBS on day 14 at 37 °C are represented. In Figure 5A,B, multiple charged DM-HNG peptides were observed at 532.21 m/z (DM-HNG 1-24, $z = 10$), 591.22 m/z (DM-HNG 1-24, $z = 9$), 665.01 m/z (DM-HNG 1-24, $z = 8$), 759.86 m/z (DM-HNG 1-24, $z = 7$), 886.33 m/z (DM-HNG 1-24, $z = 6$), and 1063.41 m/z (DM-HNG 1-24, $z = 5$). Figure 5: (A) Amino acid sequence and m/z ratio of DM-HNG are represented. (B) Representative HRMS precursor ion mass spectra of DM-HNG in PBS. (C) Amino acid sequence and m/z ratio of SOX-DM-HNG are represented. (D) Representative HRMS precursor ion mass spectra of SOX-DM-HNG in PBS. (E) Amino acid sequence and m/z ratio of DOX-DM-HNG are represented. (F) Representative HRMS precursor ion mass spectra of DOX-DM-HNG in PBS. In Figure 5C,D, multiple charged homodimerized HNG with methionine oxidation (singly oxidized homodimerized HNG, SOx-DM-HNG) were observed at 533.81 m/z ($z = 10$), 593.01 m/z ($z = 9$), 667.01 m/z ($z = 8$), 762.14 m/z ($z = 7$), 888.99 m/z ($z = 6$), and 1066.80 m/z ($z = 5$). In Figure 5E,F, multiple charged homodimerized HNG with methionine oxidized and cysteine disulfide bonds (doubly oxidized homodimerized HNG, DOx-DM-HNG) were observed at 532.30 m/z ($z = 10$), 594.77 m/z ($z = 9$), 669.01 m/z ($z = 8$), 764.43 m/z ($z = 7$), 891.67 m/z ($z = 6$), and 1069.80 m/z ($z = 5$). HNG solutions were incubated at 37 °C and analyzed at day 1 (Figure 6), day 7 (Figure 7) and day 14 (Figure 8) using ultra-performance liquid chromatography coupled with high-resolution mass spectrometry (Waters® Xevo G2-XS QTof). The presence of full-length HNG and DM-HNG in PBS, HPLC-grade water and MO formulation at the different time periods was analyzed (Figure 6, Figure 7 and Figure 8). The retention time frames of full-length HNG and DM-HNG ranged from 20.75 to 21.25 min and 22.50 to 23.25 min, respectively (Figure 6, Figure 7 and Figure 8). Figure 6: Ion chromatogram results of incubated HNG at 37 °C for 1 day in PBS (A), HPLC-grade water (B), and MO formulation (C). Figure 7: Ion chromatogram results of incubated HNG at 37 °C for 7 days in PBS (A), HPLC-grade water (B), and MO formulation (C). Figure 8: Ion chromatogram results of incubated HNG at 37 °C for 14 days in PBS (A), HPLC-grade water (B), and MO formulation (C). Concentrations of full-length HNG peptide declined over time with a corresponding appearance of new peaks that increased over time (Figure 6, Figure 7 and Figure 8). These new peaks were identified as oxidized and/or dimerized HNG products. The DM-HNG was the dominant HNG-Pd at all time points. We found that the full-length HNG peptide had oxidized and dimerized at 37 °C in the PBS, the HPLC-grade water, and MO formulation at day 1 (Figure 6A–C), day 7 (Figure 7A–C) and day 14 (Figure 8A–C). The concentration of the DM-HNG simultaneously increased over time, while the HNG stored in the MO formulation remained mostly intact. ( Figure 6, Figure 7 and Figure 8). The full-length HNG, SOx-HNG and DOx-HNG were evaluated over time in PBS, HPLC-grade water and MO formulation. In Figure 9A, at day 1, SOx-HNG and DOx-HNG were detected in the HPLC-grade water as well as in the PBS and the MO formulation at 37 °C. The highest intensities of full-length HNG, and SOx-HNG were detected in the MO-formula, and next were those in the HPLC-grade water, with the lowest in the PBS. The highest intensity of DOx-HNG was detected in the PBS solution, while that in the HPLC-grade water was lower and that in the MO formulation was the lowest. In Figure 9B, at day 7, higher intensities of SOx-HNG peptides and full-length HNG were detected in the MO formulation, next were those in the HPLC-grade water and the lowest were in the PBS at 37 °C. The highest intensity of DOx-HNG was detected in the PBS, that in the HPLC-grade water was lower and that in the MO formulation was lowest. In Figure 9C, on day 14 at 37 °C, higher intensities of full-length HNG peptides were detected in the MO formulation, next were those in the HPLC-grade water and the lowest were in the PBS at 37 °C. Higher intensities of SOx-HNG peptides were detected in the MO formulation, next were those in the PBS and the lowest were in the HPLC-grade water. The highest intensity of DOx-HNG was detected in the HPLC-grade water, that in the PBS was lower and that in the MO formulation was the lowest. The homodimerized form of HNG, SOx-DM-HNG and DOx-DM-HNG were evaluated over time in PBS, HPLC-grade water and MO formula at 37 °C. In Figure 10A, at day 1, a lower intensity of DM-HNG was detected in MO formula than HPLC-grade water and PBS. Figure 9: Overlay ion chromatogram (19 min–22.5 min) results of incubated HNG at 37 °C in PBS, HPLC-grade water and MO formulation at day 1 (A), at day 7 (B), and at day 14 (C). Figure 10: Overlay ion chromatogram (22.5 min–24 min) results of incubated HNG at 37 °C in PBS, HPLC-grade water and MO formulation at day 1 (A), at day 7 (B), at day 14 (C). In Figure 10B, on day 7 at 37 °C, higher intensities of SOx-DM-HNG peptides and DM-HNG were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO formulation at 37 °C. The highest intensity of DOx-DM-HNG was detected in the HPLC-grade water, that in the PBS was lower and that in the MO formulation was lowest. In Figure 10C, on day 14 at 37 °C, higher intensities of SOx-DM-HNG peptides and DM-HNG were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO-formula at 37 °C. The highest intensity of DOx-DM-HNG was detected in the PBS, that in the HPLC-grade water was lower and that in the MO-formula was lowest. In Figure 11A, intensities of SOx-HNG peptides were evaluated at 37 °C in various solutions at day 1, day 7 and day 14 using UPLC-HRMS. At days 1 and 7, higher intensity of SOx-HNG were detected in the MO formulation, next were those in the HPLC-grade water, with the lowest in the PBS. At day 14, higher intensities of SOx-HNG peptides were detected in the MO formulation, next were those in the PBS and the lowest were in the HPLC-grade water. Figure 11: Evaluation of intensity changes of SOx-HNG (A), DOX-HNG (B) at day 1, 7 and 14. In Figure 11B, intensities of Dox-HNG peptides were evaluated at 37 °C in various solutions at day 1, day 7 and day 14 using UPLC-HRMS. At days 1 and 7, higher intensity of DOx-HNG was detected in the PBS solution, that in the HPLC-grade water was lower and that in the MO formulation was the lowest. At day 14, higher intensities of DOx-HNG peptides were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO formulation. In Figure 12A, intensities of DM-HNG peptides were evaluated at 37 °C in various solutions at day 1, day 7 and day 14 using UPLC-HRMS. At day 1, lower intensities of DM-HNG were detected in MO formulation than HPLC-grade water and PBS. At day 7, higher intensities of DM-HNG were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO formulation at 37 °C. At day 14, a higher intensity of DM-HNG was detected in the HPLC-grade water, next was that in the PBS and the lowest was in the MO formulation at 37 °C. Figure 12: Evaluation of intensity changes of homodimerized form of HNG (A), SOX-DM-HNG (B), DOX-DM-HNG (C) at day 1, 7 and 14. In Figure 12B, intensities of SOX-DM-HNG were evaluated at 37 °C in various solutions at day 1, day 7 and day 14 using UPLC-HRMS. At day 1, SOX-DM-HNG was detected in HPLC-grade water and PBS and not detected in MO formulation. At day 7, a higher intensity of SOx-DM-HNG was detected in the HPLC-grade water, next was that in the PBS and the lowest was in the MO formulation at 37 °C. At day 14, higher intensities of SOx-DM-HNG were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO formulation at 37 °C. In Figure 12C, intensities of DOX-DM-HNG were evaluated at 37 °C in various solutions at day 1, day 7 and day 14 using UPLC-HRMS. At day 1, a lower intensity of DOX-DM-HNG was detected in MO formulation than HPLC-grade water and PBS. At day 7, higher intensities of DOx-DM-HNG peptides were detected in the HPLC-grade water, next were those in the PBS and the lowest were in the MO formulation at 37 °C. At day 14, the highest intensity of DOx-DM-HNG was detected in the PBS, that in the HPLC-grade water was lower and that in the MO formulation was the lowest. ## 3.3. Long-Term Stability of HNG Peptide in Different Conditions We evaluated the long-term stability of HNG peptide in the HPLC water and MO formulation stored at 4 °C for 11 months. Collected HRMS data were analyzed using the BiopharmaLynx program to identify peptide sequences (Tables S3 and S4). Identified peptide sequences and HRMS data collected from 11 months old HNG in HPLC water (Table S3) and MO formulation (Table S4) show the full-length HNG peptide, its fragments, and dimerized forms. Mass spectrometry analysis showed that the HNG peptide in HPLC water degraded into multiple fragments (Figure 13B), while in MO formulation, HNG remained mostly intact (Figure 13E). The retention time frames of 11-month-old DM-HNG in HPLC water ranged from 24.2 to 24.6 min (Figure 13A,B, Upper Panel A) and 11-month-old HNG in MO formulation ranged from 22.4 to 22.6 min (Figure 13D,E, Lower Panel B). The 11-month-old HNG peptides in HPLC water showed multiple charged states 532.2 m/z (homodimerized-HNG 1-24, $z = 10$), 591.0 m/z (DM-HNG 1-24, $z = 9$), 664.9 m/z (DM-HNG 1-24, $z = 8$), 759.9 m/z (DM-HNG 1-24, $z = 7$), 886.2 m/z (DM-HNG 1-24, $z = 6$), and 1063.4 m/z (DM-HNG 1-24, $z = 5$) as shown in Figure 13C, Upper Panel A. Many fewer multiple charged intact molecules of 11-month-old HNG peptides in MO formulation were observed at m/z 532.3 (HNG 1-24, $z = 5$), 665.1 (HNG 1-24, $z = 4$), and 886.5 (HNG 1-24, $z = 3$) (Figure 13F, Lower Panel B). Figure 13: Measurements of HNG in HPLC-grade water (Upper Panel A) and MO formulation (Lower Panel B) at 4 °C for 11 months using HRMS. UPPER PANEL A, (A) Amino acid sequence and m/z ratio of dimerized form HNG is represented; (B,C) Representative ion chromatogram and HRMS product ion mass spectra of HNG in HPLC water, respectively. LOWER PANEL B, (D) Amino acid sequence and m/z ratio of HNG is represented; (E,F) Representative ion chromatogram and HRMS product ion mass spectra of HNG in MO formulation. As seen in Figure 13, the HNG peptides without a stabilizing formulation were oxidized and dimerized continuously in the HPLC-grade water while the peptide oxidation and dimerization were much slower in the MO formulation. The HNG peptides were stored in the HPLC water (Table S3) and MO formulation (Table S4) over 11 months, and many dimers and oxidized products were identified in both solutions. However, the highest intensities of full-length HNG were found in MO formulation (Table S4), suggesting the potential of the MO formulation for general use in enhancing peptide stability and preventing peptide oxidation. In summary, our results show that the full-length HNG peptide (24 amino acids) is highly susceptible to chemical modification when placed in HPLC water and PBS at either 4 °C or 37 °C. For example, when placed in HPLC water for 28 days at 4 °C, less than $11\%$ was found in the full-length HNG peptide form (Figure 3), but the HNG peptide was stabilized when placed in the MO formulation ($67\%$ and $95\%$ remained full-length HNG at 37 °C or 4 °C, respectively). Using UPLC-HRMS, the full-length HNG was found in both the singly oxidized and doubly oxidized forms (Figure 4); the ionized mass spectra of the SOX-HNG and DOX-HNG showed multiple charged states (Figure 4). The ion chromatography of the full-length HNG incubated for 1 to 14 days in PBS, HPLC-grade water or the MO formulations showed increasing loss in the full-length HNG in the PBS and HPLC-grade water, but surprisingly, the full-length HNG in the MO formulation remains mostly stable (Figure 6). Finally, when the full-length HNG was stored for 11 months in HPLC-grade water at 4 °C, we found full-length HNG, along with dimerization and degradation products of HNG, which included three different HNG fragments and four different dimerized forms of HNG fragments (Table S3). When the full-length HNG was stored for 11 months in the MO formulation at 4 °C, there were the full-length HNG, SOX-HNG, DOX-HNG and degradation products of HNG, which includes 28 different HNG fragments and 64 dimerized forms of HNG fragments (Table S4). These data demonstrate that the full-length HNG is fragmented and modified chemically in HPLC-grade water and the MO formulation. Future studies will investigate the biological features of the HNG fragments, the dimerized HNG and oxidized HNG, since these forms may have signaling functions, reflecting increased mitochondrial DNA damage and/or perhaps a positive, rescuing effect for damaged cells. ## 4. Discussion Despite the promising results that demonstrated the key cellular protective role of HNG, the in vitro and in vivo stability and half-life properties of HNG peptide have not been well studied, and the majority of the reported studies investigating the role of HNG were mostly limited to in vitro cell cultures. Therefore, determining and enhancing the molecular stability properties of HNG is essential to better translate to proper in vivo therapeutic studies. Understanding the stability properties of full-length HNG peptide can help us more accurately determine the dosing and frequency for HNG administration in in vivo animal studies and for possible future clinical studies investigating its role in physiology and disease. Water and moisture have many effects on peptide degradation [31]. A 28-amino acid Vasoactive Intestinal Peptide, a 29-amino acid peptide Glucagon, vaso-active intestinal peptide and a tricyclic glycopeptide Vancomycin are unstable in aqueous solutions [32,33,34,35]. Knoop et al. have reported the instability of the MOTS-c peptide, another mitochondria-derived peptide, in the human plasma [36]. Consistent with those findings, our results showed that HNG in HPLC water is unstable at 37°C, reaching $50\%$ concentration at approximately 33 h. When stored in HPLC water at 4°C, then $54\%$ of the HNG peptide remained stable at day 7, indicating peptide instability even in cold storage conditions in HPLC water. Therefore, researchers investigating the effects of the HNG peptide should consider this newly identified short half-life when determining treatment doses and frequency in cell cultures or in vivo administration to the systemic circulation. To overcome the low stability issue of the HNG peptide, we developed a special solution (MO formulation) to improve the stability of the molecule. The MO formulation has an acidic property to stabilize the HNG peptide structure. Our proprietary solution demonstrated significant efficiency resulting in $95\%$ full-length HNG peptides after 28 days of storage at 4 °C. Moreover, the MO formulation could provide a $95\%$ stable HNG peptide concentration for up to 7 days at 37 °C. Hence, the MO formulation may significantly improve the efficacy of HNG treatment, and it could reduce administration frequency and costs as well. The increased stability with the MO formulation may provide further processing opportunities such as infusion of HNG and/or potentially other MDPs such as small humanin-like peptides into microspheres for various applications. Oxidative mechanisms play critical roles in aging and age-related diseases such as ischemia, atherosclerosis, Alzheimer’s disease, cataracts, and AMD [37,38,39]. Peptide oxidation decreases enzymatic activity, accumulates with age, and is related to numerous diseases [38]. Cysteine, methionine, histidine, and tryptophan amino acids are most susceptible to oxidation [38]. Oxidation causes a mass shift of +16 Da. HNG includes methionine and cysteine amino acids, which are susceptible to oxidation. Cysteine oxidation in HNG is responsible for the dimerization of HNG fragments via disulfide bridges. The disulfide bridge causes a mass shift of −2 Da and produces stable, covalently bonded dimers. The high-resolution tandem mass spectrometer provides highly sensitive and accurate results that can identify oxidation sites and disulfide bridges in peptides and dimerized peptides. Finally, there is a lack of knowledge regarding the degradation products of HNG peptide and their oxidized and dimerized forms. Our study demonstrates that HNG fragments formed homodimers and heterodimers via disulfide bridge interactions (Tables S3 and S4). In the long-term study, our results show that dimerization provides increased stability for the intact HNG and its fragments (Figure 13D,F, Lower Panel B). Consistent with the other studies, several proteins have been shown to increase stability and have functions in dimerized forms, such as human IgG antibody [40], HLA-G dimers on cell surfaces [41], human superoxide dismutase enzymes [42], and glial cell line-derived neurotrophic factor [43]. Disulfide bonds contribute to the structure, functionality stability, and dimerization of peptides and proteins [40,41,42,43]. ## 5. Conclusions For the first time, the short- and long-term stability properties of HNG peptide and its oxidation and degradation products have been analyzed in detail using advanced HPLC and HRMS technologies. Our findings may provide insight for understanding key features in the HNG peptide sequence that define its stability via disulfide bonds. It is currently unknown whether dimerized HNG fragments possess any biological activity. Additionally, we have identified various HNG fragments that may possess different cellular functionalities and/or receptor activities. Future studies will investigate whether there are any such biological functions of these HNG fragments. We also developed a new chemical solution that significantly improves the stability of the HNG peptide in both 4 °C and 37 °C media conditions for up to 28 days. Our results may help researchers design better in vitro and in vivo experimental parameters to further understand the critical role of Humanin and HNG in physiological conditions and human diseases. ## References 1. Nashine S., Cohen P., Chwa M., Lu S., Nesburn A.B., Kuppermann B.D., Kenney M.C.. Humanin G (HNG) Protects Age-Related Macular Degeneration (AMD) Transmitochondrial ARPE-19 Cybrids from Mitochondrial and Cellular Damage. Cell Death Dis. (2017) 8 e2951. DOI: 10.1038/cddis.2017.348 2. Riazi-Esfahani M., Kuppermann B., Kenney M.. The Role of Mitochondria in AMD: Current Knowledge and Future Applications. J. Ophthalmic Vis. Res. (2017) 12 424. DOI: 10.4103/jovr.jovr_182_17 3. Gong Z., Tas E., Muzumdar R.. Humanin and Age-Related Diseases: A New Link?. Front. Endocrinol. (2014) 5 210. DOI: 10.3389/fendo.2014.00210 4. Guo B., Zhai D., Cabezas E., Welsh K., Nouraini S., Satterthwait A.C., Reed J.C.. Humanin Peptide Suppresses Apoptosis by Interfering with Bax Activation. Nature (2003) 423 456-461. DOI: 10.1038/nature01627 5. Hashimoto Y., Niikura T., Tajima H., Yasukawa T., Sudo H., Ito Y., Kita Y., Kawasumi M., Kouyama K., Doyu M.. A Rescue Factor Abolishing Neuronal Cell Death by a Wide Spectrum of Familial Alzheimer’s Disease Genes and Aβ. Proc. Natl. Acad. Sci. USA (2001) 98 6336-6341. DOI: 10.1073/pnas.101133498 6. Nishimoto I., Matsuoka M., Niikura T.. Unravelling the Role of Humanin. Trends Mol. Med. (2004) 10 102-105. DOI: 10.1016/j.molmed.2004.01.001 7. Hashimoto Y., Ito Y., Niikura T., Shao Z., Hata M., Oyama F., Nishimoto I.. Mechanisms of Neuroprotection by a Novel Rescue Factor Humanin from Swedish Mutant Amyloid Precursor Protein. Biochem. Biophys. Res. Commun. (2001) 283 460-468. DOI: 10.1006/bbrc.2001.4765 8. Sreekumar P.G., Ishikawa K., Spee C., Mehta H.H., Wan J., Yen K., Cohen P., Kannan R., Hinton D.R.. The Mitochondrial-Derived Peptide Humanin Protects RPE Cells From Oxidative Stress, Senescence, and Mitochondrial Dysfunction. Invest. Ophthalmol. Vis. Sci. (2016) 57 1238. DOI: 10.1167/iovs.15-17053 9. Lee C., Zeng J., Drew B.G., Sallam T., Martin-Montalvo A., Wan J., Kim S.-J., Mehta H., Hevener A.L., de Cabo R.. The Mitochondrial-Derived Peptide MOTS-c Promotes Metabolic Homeostasis and Reduces Obesity and Insulin Resistance. Cell Metab. (2015) 21 443-454. DOI: 10.1016/j.cmet.2015.02.009 10. Minasyan L., Sreekumar P.G., Hinton D.R., Kannan R.. Protective Mechanisms of the Mitochondrial-Derived Peptide Humanin in Oxidative and Endoplasmic Reticulum Stress in RPE Cells. Oxidative Med. Cell. Longev. (2017) 2017 1-11. DOI: 10.1155/2017/1675230 11. Lee C., Wan J., Miyazaki B., Fang Y., Guevara-Aguirre J., Yen K., Longo V., Bartke A., Cohen P.. IGF-I Regulates the Age-Dependent Signaling Peptide Humanin. Aging Cell (2014) 13 958-961. DOI: 10.1111/acel.12243 12. Tajima H., Niikura T., Hashimoto Y., Ito Y., Kita Y., Terashita K., Yamazaki K., Koto A., Aiso S., Nishimoto I.. Evidence for in Vivo Production of Humanin Peptide, a Neuroprotective Factor against Alzheimer’s Disease-Related Insults. Neurosci. Lett. (2002) 324 227-231. DOI: 10.1016/S0304-3940(02)00199-4 13. Men J., Zhang X., Yang Y., Gao D.. An AD-Related Neuroprotector Rescues Transformed Rat Retinal Ganglion Cells from CoCl2-Induced Apoptosis. J. Mol. Neurosci. (2012) 47 144-149. DOI: 10.1007/s12031-011-9701-5 14. Niikura T., Chiba T., Aiso S., Matsuoka M., Nishimoto I.. Humanin: After the Discovery. MN (2004) 30 327-340. DOI: 10.1385/MN:30:3:327 15. Muzumdar R.H., Huffman D.M., Atzmon G., Buettner C., Cobb L.J., Fishman S., Budagov T., Cui L., Einstein F.H., Poduval A.. Humanin: A Novel Central Regulator of Peripheral Insulin Action. PLoS ONE (2009) 4. DOI: 10.1371/journal.pone.0006334 16. Bachar A.R., Scheffer L., Schroeder A.S., Nakamura H.K., Cobb L.J., Oh Y.K., Lerman L.O., Pagano R.E., Cohen P., Lerman A.. Humanin Is Expressed in Human Vascular Walls and Has a Cytoprotective Effect against Oxidized LDL-Induced Oxidative Stress. Cardiovasc. Res. (2010) 88 360-366. DOI: 10.1093/cvr/cvq191 17. Hashimoto Y., Niikura T., Ito Y., Sudo H., Hata M., Arakawa E., Abe Y., Kita Y., Nishimoto I.. Detailed Characterization of Neuroprotection by a Rescue Factor Humanin against Various Alzheimer’s Disease-Relevant Insults. J. Neurosci. (2001) 21 9235-9245. DOI: 10.1523/JNEUROSCI.21-23-09235.2001 18. Muzumdar R.H., Huffman D.M., Calvert J.W., Jha S., Weinberg Y., Cui L., Nemkal A., Atzmon G., Klein L., Gundewar S.. Acute Humanin Therapy Attenuates Myocardial Ischemia and Reperfusion Injury in Mice. ATVB (2010) 30 1940-1948. DOI: 10.1161/ATVBAHA.110.205997 19. Xu X., Chua C.C., Gao J., Hamdy R.C., Chua B.H.L.. Humanin Is a Novel Neuroprotective Agent Against Stroke. Stroke (2006) 37 2613-2619. DOI: 10.1161/01.STR.0000242772.94277.1f 20. Oh Y.K., Bachar A.R., Zacharias D.G., Kim S.G., Wan J., Cobb L.J., Lerman L.O., Cohen P., Lerman A.. Humanin Preserves Endothelial Function and Prevents Atherosclerotic Plaque Progression in Hypercholesterolemic ApoE Deficient Mice. Atherosclerosis (2011) 219 65-73. DOI: 10.1016/j.atherosclerosis.2011.06.038 21. Takeshita Y., Hashimoto Y., Nawa M., Uchino H., Matsuoka M.. SH3-Binding Protein 5 Mediates the Neuroprotective Effect of the Secreted Bioactive Peptide Humanin by Inhibiting c-Jun NH2-Terminal Kinase. J. Biol. Chem. (2013) 288 24691-24704. DOI: 10.1074/jbc.M113.469692 22. Klein L.E., Cui L., Gong Z., Su K., Muzumdar R.. A Humanin Analog Decreases Oxidative Stress and Preserves Mitochondrial Integrity in Cardiac Myoblasts. Biochem. Biophys. Res. Commun. (2013) 440 197-203. DOI: 10.1016/j.bbrc.2013.08.055 23. Jin H., Liu T., Wang W.-X., Xu J.-H., Yang P.-B., Lu H.-X., Sun Q.-R., Hu H.-T.. Protective Effects of [Gly14]-Humanin on β-Amyloid-Induced PC12 Cell Death by Preventing Mitochondrial Dysfunction. Neurochem. Int. (2010) 56 417-423. DOI: 10.1016/j.neuint.2009.11.015 24. Yamagishi Y., Hashimoto Y., Niikura T., Nishimoto I.. Identification of Essential Amino Acids in Humanin, a Neuroprotective Factor against Alzheimer’s Disease-Relevant Insults. Peptides (2003) 24 585-595. DOI: 10.1016/S0196-9781(03)00106-2 25. Eriksson E., Wickström M., Perup L.S., Johnsen J.I., Eksborg S., Kogner P., Sävendahl L.. Protective Role of Humanin on Bortezomib-Induced Bone Growth Impairment in Anticancer Treatment. JNCI J. Natl. Cancer Inst. (2014) 106 djt459. DOI: 10.1093/jnci/djt459 26. Kyte J., Doolittle R.F.. A Simple Method for Displaying the Hydropathic Character of a Protein. J. Mol. Biol. (1982) 157 105-132. DOI: 10.1016/0022-2836(82)90515-0 27. Chang K.Y., Yang J.-R.. Analysis and Prediction of Highly Effective Antiviral Peptides Based on Random Forests. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0070166 28. Enany S.. Structural and Functional Analysis of Hypothetical and Conserved Proteins of Clostridium Tetani. J. Infect. Public Health (2014) 7 296-307. DOI: 10.1016/j.jiph.2014.02.002 29. Li R.-F., Lu Z.-F., Sun Y.-N., Chen S.-H., Yi Y.-J., Zhang H.-R., Yang S.-Y., Yu G.-H., Huang L., Li C.-N.. Molecular Design, Structural Analysis and Antifungal Activity of Derivatives of Peptide CGA-N46. Interdiscip. Sci. Comput. Life Sci. (2016) 8 319-326. DOI: 10.1007/s12539-016-0163-x 30. Bernardo D., Sánchez B., Al-Hassi H.O., Mann E.R., Urdaci M.C., Knight S.C., Margolles A.. Microbiota/Host Crosstalk Biomarkers: Regulatory Response of Human Intestinal Dendritic Cells Exposed to Lactobacillus Extracellular Encrypted Peptide. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0036262 31. Bell L.N.. Peptide Stability in Solids and Solutions. Biotechnol. Prog. (1997) 13 342-346. DOI: 10.1021/bp970057y 32. Mody R., Tramontano A., Paul S.. Spontaneous Hydrolysis of Vasoactive Intestinal Peptide in Neutral Aqueous Solution. Int. J. Pept. Protein Res. (1994) 44 441-447. PMID: 7896502 33. Taleb N., Coriati A., Khazzaka C., Bayonne J., Messier V., Rabasa-Lhoret R.. Stability of Commercially Available Glucagon Formulation for Dual-Hormone Artificial Pancreas Clinical Use. Diabetes Technol. Ther. (2017) 19 589-594. DOI: 10.1089/dia.2017.0204 34. Jackson M.A., Caputo N., Castle J.R., David L.L., Roberts C.T., Ward W.K.. Stable Liquid Glucagon Formulations for Rescue Treatment and Bi-Hormonal Closed-Loop Pancreas. Curr. Diabetes Rep. (2012) 12 705-710. DOI: 10.1007/s11892-012-0320-5 35. Cao M., Feng Y., Zhang Y., Kang W., Lian K., Ai L.. Studies on the Metabolism and Degradation of Vancomycin in Simulated in Vitro and Aquatic Environment by UHPLC-Triple-TOF-MS/MS. Sci. Rep. (2018) 8 1-13. DOI: 10.1038/s41598-018-33826-9 36. Knoop A., Thomas A., Thevis M.. Development of a Mass Spectrometry Based Detection Method for the Mitochondrion-Derived Peptide MOTS-c in Plasma Samples for Doping Control Purposes. Rapid Commun. Mass Spectrom. (2019) 33 371-380. DOI: 10.1002/rcm.8337 37. Ramkumar S., Fan X., Wang B., Yang S., Monnier V.M.. Reactive Cysteine Residues in the Oxidative Dimerization and Cu2+ Induced Aggregation of Human ΓD-Crystallin: Implications for Age-Related Cataract. Biochim. Biophys. Acta-Mol. Basis Dis. (2018) 1864 3595-3604. DOI: 10.1016/j.bbadis.2018.08.021 38. Schey K.L., Finley E.L.. Identification of Peptide Oxidation by Tandem Mass Spectrometry. Acc. Chem. Res. (2000) 33 299-306. DOI: 10.1021/ar9800744 39. Reeg S., Grune T.. Protein Oxidation in Aging: Does It Play a Role in Aging Progression?. Antioxid. Redox Signal. (2015) 23 239-255. DOI: 10.1089/ars.2014.6062 40. McAuley A., Jacob J., Kolvenbach C.G., Westland K., Lee H.J., Brych S.R., Rehder D., Kleemann G.R., Brems D.N., Matsumura M.. Contributions of a Disulfide Bond to the Structure, Stability, and Dimerization of Human IgG1 Antibody CH3 Domain. Protein Sci. (2007) 17 95-106. DOI: 10.1110/ps.073134408 41. Boyson J.E., Erskine R., Whitman M.C., Chiu M., Lau J.M., Koopman L.A., Valter M.M., Angelisova P., Horejsí V., Strominger J.L.. Disulfide Bond-Mediated Dimerization of HLA-G on the Cell Surface. Proc. Natl. Acad. Sci. USA (2002) 99 16180-16185. DOI: 10.1073/pnas.212643199 42. Banci L., Bertini I., Durazo A., Girotto S., Gralla E.B., Martinelli M., Valentine J.S., Vieru M., Whitelegge J.P.. Metal-Free Superoxide Dismutase Forms Soluble Oligomers under Physiological Conditions: A Possible General Mechanism for Familial ALS. Proc. Natl. Acad. Sci. USA (2007) 104 11263-11267. DOI: 10.1073/pnas.0704307104 43. Li T., Yamane H., Arakawa T., Narhi L.O., Philo J.. Effect of the Intermolecular Disulfide Bond on the Conformation and Stability of Glial Cell Line-Derived Neurotrophic Factor. Protein Eng. Des. Sel. (2002) 15 59-64. DOI: 10.1093/protein/15.1.59
--- title: Phase 1 Study to Evaluate the Safety of Reducing the Prophylactic Dose of Dexamethasone around Docetaxel Infusion in Patients with Prostate and Breast Cancer authors: - Rieneke T. Lugtenberg - Stefanie de Groot - Danny Houtsma - Vincent O. Dezentjé - Annelie J. E. Vulink - Maarten J. Fischer - Johanneke E. A. Portielje - Jacobus J. M. van der Hoeven - Hans Gelderblom - Hanno Pijl - Judith R. Kroep journal: Cancers year: 2023 pmcid: PMC10046524 doi: 10.3390/cancers15061691 license: CC BY 4.0 --- # Phase 1 Study to Evaluate the Safety of Reducing the Prophylactic Dose of Dexamethasone around Docetaxel Infusion in Patients with Prostate and Breast Cancer ## Abstract ### Simple Summary Docetaxel has been approved as an anti-cancer agent in 1995. High rates of hypersensitivity reactions (HSR) and fluid retention were observed when this agent was first introduced. The use of high dose systemic corticosteroids around docetaxel infusion appeared to decrease the incidence of HSR and fluid retention and has been applied in daily practice ever since. However, there is little evidence that supports this high dose of dexamethasone. Furthermore, the application of high-dosed corticosteroids can lead to undesirable adverse effects. In this phase 1 study, we aim to evaluate the impact of reducing the dose of dexamethasone as an adjunct to docetaxel on the incidence of HSR and fluid retention in patients with prostate or breast cancer. ### Abstract Background: There is little evidence that supports the registered high dose of dexamethasone used around docetaxel. However, this high dose is associated with considerable side effects. This study evaluates the feasibility of reducing the prophylactic oral dosage of dexamethasone around docetaxel infusion. Patients and methods: Eligible patients had a histologically confirmed diagnosis of prostate or breast cancer and had received at least three cycles of docetaxel as monotherapy or combination therapy. Prophylactic dexamethasone around docetaxel infusion was administered in a de-escalating order per cohort of patients. Primary endpoint was the occurrence of grade III/IV fluid retention and hypersensitivity reactions (HSRs). Results: Of the 46 enrolled patients, 39 were evaluable (prostate cancer ($$n = 25$$), breast cancer ($$n = 14$$). In patients with prostate cancer, the dosage of dexamethasone was reduced to a single dose of 4 mg; in patients with breast cancer, the dosage was reduced to a 3-day schedule of 4 mg–8 mg–4 mg once daily, after which no further reduction has been tested. None of the 39 patients developed grade III/IV fluid retention or HSR. One patient ($2.6\%$) had a grade 1 HSR, and there were six patients ($15.4\%$) with grade I or II edema. There were no differences in quality of life (QoL) between cohorts. Conclusions: It seems that the prophylactic dose of dexamethasone around docetaxel infusion can be safely reduced with respect to the occurrence of grade III/IV HSRs or the fluid retention syndrome. ## 1. Introduction Docetaxel—a semisynthetic analog of paclitaxel, causing cell-cycle arrest and apoptosis through interference with microtubular function—has been registered as an anticancer agent in 1995 [1,2]. In the early clinical trials, high rates of hypersensitivity reactions (HSRs) were observed during taxane infusion. The occurrence of HSRs decreased to less than $10\%$ after prophylactic medication with H1 and H2 antihistamines and when systemic corticosteroids became part of cancer treatment protocols [3,4]. Additionally, a fluid retention syndrome, characterized by weight gain, edema, and pleural effusion, was observed after docetaxel administration, which resulted in treatment discontinuation in 30–$70\%$ of patients. It was found to be a cumulative, dose-limiting, and slowly reversible toxicity [5,6,7,8,9,10,11]. Corticosteroids, first given to prevent HSR, appeared to prevent fluid retention associated with docetaxel as well [5,6,7,8]. Docetaxel is approved for the treatment of breast cancer with the concomitant use of a 3-day schedule of dexamethasone 8 mg, bi-daily (bid), starting on the day before chemotherapy, with the purpose to decrease the severity of fluid retention and HSRs [12]. The commonly used dosage of docetaxel in treatment schedules for breast cancer is 100 mg/m2. For the treatment of prostate cancer, a lower dosage of 75 mg/m2 docetaxel is registered. Due to this lower dosage and the concurrent use of low-dose prednisone in prostate cancer treatment schedules, this is accompanied by another prophylactic regimen: 3 times of 8 mg of dexamethasone on the day of docetaxel infusion [12]. There is little evidence that supports the high doses of dexamethasone used in both schedules. For example, the use of the 3-day 8 mg bid schedule for the docetaxel dosage of 100 mg/m2 is based on a conference abstract [13]. Dexamethasone potentially has severe side effects and can evoke manifestations of diabetes mellitus, weight gain, gastro-esophageal reflux disease, personality changes, irritability, insomnia, agitation, euphoria, mania, and mood swings [14,15,16]. The use of corticosteroids can induce immunosuppression with an increased risk of infection—the risk is even higher when myelosuppressive chemotherapy is given simultaneously [16]. In addition, there is increasing evidence that the occurrence of diabetes, causing high values of glucose and insulin, can worsen the prognosis of cancer patients [17,18]. Data from pre-clinical and clinical studies suggest that corticosteroids can induce treatment resistance in solid tumors [19]. In recent years, it has been established that glucocorticoid receptors (GRs) may be involved in the development of castration-resistant prostate cancer (CRPC) [20,21]. The upregulation of the GR may drive tumor proliferation and possibly lead to resistance to antiandrogen therapies [22]. As a consequence, the use of dexamethasone and other corticosteroids may contribute to tumor progression in prostate cancer [23,24]. Finally, dexamethasone is a CYP3A4 inducer, which might increase docetaxel clearance [25,26]. Thus, there is a need to re-evaluate the optimal dose of prophylactic dexamethasone. In this phase 1 study, we evaluated the impact of reducing the dose of dexamethasone as an adjunct to docetaxel on HSR and fluid retention in patients with prostate or breast cancer. ## 2.1. Study Design This study is a multicenter, open label, dose-de-escalating, non-randomized phase 1 study. Patients received docetaxel infusion every 3 weeks for a minimum of 3 cycles, depending on the regimen, until progressive disease or unacceptable toxicity. Prophylactic dexamethasone co-medication was administered in a de-escalating order (Table 1) per cohort (Table 1). Six patients were enrolled per dose level initially. Each patient within a cohort received the same dose of dexamethasone in every subsequent cycle. The last patients of a cohort were observed for 2 cycles of docetaxel treatment before accrual to the next lower dose level was started. Patients were replaced within a cohort if they left the study within 3 weeks for reasons other than toxicity. If no grade III/IV HSR or fluid retention reaction occurred in the six patients within on cohort, the next cohort was treated with the next dose level. If one grade III/IV HSR or fluid retention reaction occurred in one of the six patients within one cohort, then three additional patients were treated at that dose level. If there were no additional grade III/IV HSR or fluid retention in the additional 3 patients, accrual to the next lower dose level was started. If a grade III/IV HSR or fluid retention occurred in at least $\frac{2}{6}$ or $\frac{2}{9}$ patients, that dose was not considered as safe. Each patient within a cohort received the same dose of dexamethasone in every subsequent cycle. Patients were replaced within a cohort if they left the study within 3 weeks for reasons other than toxicity. Initially, for the breast cancer group, 3 additional cohorts were planned (cohort 4: day 0: 8 mg and day 1: 4 mg; cohort 5: day 0: 8 mg; and cohort 6 day 0: 4 mg). However, the inclusion of patients with breast cancer was stopped after cohort 3, as inclusion was falling behind due to an increase in the use of weekly paclitaxel instead of docetaxel in this group of patients. ## 2.2. Patient Eligible patients had histologically confirmed the diagnosis of prostate cancer or breast cancer and a treatment plan with a minimum of 3 cycles of docetaxel monotherapy or combination therapy. Patients with prostate cancer can be treated with or without bi-daily (bid) 5 mg prednisone continuously. Patients had an adequate bone marrow function (i.e., white blood counts > 3.0 × 109/L, absolute neutrophil count ≥ 1.5 × 109/L, and platelet count ≥ 100 × 109/L) no signs of liver damage (i.e., bilirubin ≤ 1.5 × upper limit of normal (UNL) range, ALAT and/or ASAT ≤ 2.5 × UNL, and Alkaline Phosphatase ≤ 5 × UNL), adequate renal function (i.e., calculated creatinine clearance ≥ 50 mL/min), a WHO performance status of 0–2, age ≥ 18 years, a survival expectation of >3 months, an absence of diabetes mellitus, an absence of steroid use for other conditions, an absence of pregnancy or current lactation, an absence of existing edema, and written informed consent. Patients were excluded if they had a known hypersensitivity for docetaxel, paclitaxel, other chemotherapeutic agents, products containing polysorbate 80, or an earlier experience of anaphylaxis for food, insect bites, medication, or another foreign substance. ## 2.3. Endpoint The primary endpoint was HSR or fluid retention syndrome grade III/IV. During each cycle, toxicity was documented by the physician and graded according to the Common Terminology Criteria for Adverse Events version 4.03 (CTCAE v.4.03) [26]. ## 2.4. Quality of Life Quality of life was assessed with the European Organization for Research and Treatment of Cancer—Core Quality of Life Questionnaire (EORTC-QLQ C30) before the start of treatment, after 3 cycles and after 6 cycles of docetaxel. This 30-item test comprises one global health scale, five function scales (physical, emotional, cognitive, social, and role), three symptom scales (fatigue, nausea, and pain), and six single items. All scores were transformed to a 0–100 scale. ## 2.5. Statistical Analysis Other trials with the monotherapy of docetaxel showed that 5–$6\%$ of patients experienced grade III/IV fluid retention [27,28,29] and $2.5\%$ of patients experienced grade III/IV HSR [27] with concomitant prophylactic high-dose dexamethasone. Therefore, the combined endpoint of grade III/IV fluid retention or HSR was expected to occur in approximately $8\%$ of patients. We deemed doubling in the occurrence of grade III/IV fluid retention and HSR to be acceptable. Thus, one out of six patients within the one-dose cohort who experienced grade III/IV fluid retention or HSR would be above the maximal accepted doubling of side effects. In that case, three additional patients were treated at the same dose level. If there was no further occurrence of grade III/IV fluid retention or HSR within that cohort, the toxicity remained under the accepted doubling (one out of nine patients, $11\%$). Grade III/IV HSR or fluid retention occurred in at least $\frac{2}{6}$ or $\frac{2}{9}$ patients, thus, the estimation of occurrence of grade III/IV fluid retention or HSR was unacceptably high in the study, with $33\%$ or $22\%$, respectively. Therefore, the last high dose level of dexamethasone will be the recommended dose for a phase III study. The differences between cohorts on serum levels for glucose, insulin, and IGF-1 levels were tested with the Mann–Whitney U test. The differences between cohorts on the different QoL scales were estimated using linear mixed models, with an unstructured covariance matrix including cohorts, time, and the interaction between cohorts and time. For each scale, all scores over time were used as the dependent outcome in the models. All tests were two-tailed with a significance level of $p \leq 0.05.$ All data were analyzed using IBM SPSS Statistics for Windows (Version 25.0. Armonk, NY, USA: IBM Corp). ## 3.1. Patient Characteristics From April 2016 to June 2020, 28 patients with prostate cancer and 18 patients with breast cancer from three participating Dutch centers were included. Patient characteristics are shown in Table 2. A total of $\frac{39}{46}$ patients were evaluable for toxicity. Three patients used the normal dosage of dexamethasone and violated study protocol. One patient declined to participate in the study and withdrew informed consent, two patients stopped the docetaxel treatment early, and one patient switched to paclitaxel treatment (Figure 1). ## 3.2. Toxicity The percentage of patients in whom a hypersensitivity reaction (HSR) and/or fluid retention was observed is shown in Table 3. No grade III/IV toxicity occurred in any of the cohorts. One patient developed a grade I HSR. After this mild reaction, the patient decided to use the normal dosage of dexamethasone in the consequent cycles and left the study. Six patients ($15\%$) had grade I or II fluid retention, consisting of mild-to-moderate edema; no pleural effusions were observed. Febrile neutropenia, nausea, and hyperglycemia occurred in up to $33\%$ (Table 4). No differences were found in the median levels of glucose, insulin, or IGF-1 between cohorts (Supplementary Material Table S1). ## 3.3. Quality of Life The QLQ-C30 was completed in all 24 patients in the prostate cancer cohort and $\frac{10}{14}$ of the patients in the breast cancer cohort before the start of docetaxel treatment. The scores were comparable between the different cohorts (Supplementary Material Table S2). Some of the scores deteriorated similarly during docetaxel treatment, other scores had different patterns over time. However, there were no differences between cohorts on any of the EORTC-QLQ C30 scales or items during the tree time points (Supplementary Material Table S3). ## 4. Discussion This phase 1 dose-finding study evaluates the feasibility of reducing the optimal dose of dexamethasone comedication for docetaxel treatment and demonstrates the feasibility of reducing it. Our results show that reducing the dose of dexamethasone to a single dose of 4 mg before docetaxel administration in patients with prostate cancer (docetaxel dosage of 75 mg/m2)—or to at least to 4 mg on day −1, 8 mg on day 0, and 4 mg on day 1 in patients with breast cancer (docetaxel dosage of 100 mg/m2)—does not increase the incidence of hypersensitivity reactions (HSRs) or fluid retention syndrome. Furthermore, there were no differences in perceived QoL, nor did patients experience increased nausea, fatigue, or loss of appetite. High-dose dexamethasone may be associated with severe side effects such as metabolic changes, gastro-intestinal conditions, and behavior change [14,15,16]. However, the use of dexamethasone around docetaxel treatment is mostly short-term and HSRs and fluid retention syndrome can be life-threatening, so tapering should be carried out cautiously. Therefore, reducing or withholding dexamethasone premedication is nevertheless desirable, especially when evidence to support the prescription of high-dose dexamethasone co-medication is lacking. Upfront therapy with six courses of docetaxel in addition to androgen deprivation therapy (ADT) has been shown to improve overall survival in patients with metastatic hormone-sensitive prostate cancer (mHSPC) and has become the standard of care [30,31]. In the CHAARTED trial, the concomitant use of prednisone was not mandatory; therefore, nowadays, many patients with prostate cancer receive six cycles of docetaxel 75 mg/m2 without bi-daily 5 mg of prednisone. We show that even in this group of patients, dexamethasone can be safely reduced to a single dose of 4 mg before docetaxel infusion. Thus, it seems that patients can receive six courses of docetaxel treatment safely with a single dose of 4 mg dexamethasone premedication for each course, instead of three times of 8 mg, even without chronic prednisone use (cohort 3B prostate cancer). This supports a significant reduction in corticosteroid use with a beneficial reduction in associated adverse effects. Our findings were consistent with the results of a few previous studies, in which lower doses of dexamethasone during docetaxel treatment were investigated [32,33,34,35,36]. None of these studies reported an increase in the incidence of HSRs or fluid retention if dexamethasone was given in a lower dose. Chen et al. safely reduced the dose of the recommended dexamethasone as an adjunct of docetaxel (dosage 70 mg/m2) in patients with head and neck cancer from 45 to 11 mg without an increase in severe HSRs or edema [33]. Accordingly, other studies reported no differences in HSRs or fluid retention after a single dose of dexamethasone IV before docetaxel administration in weekly or 3-weekly treatment schedules for the treatment of various solid tumors [32,35], nor after a 3-day regimen with a lower dose of dexamethasone (4.5 mg once a day) in comparison with the standard regimen [33]. To our knowledge, this is the first prospective study in which a single dose of oral dexamethasone premedication before docetaxel treatment has been investigated. Chemotherapy-induced (febrile) neutropenia and infections can be life-threatening and dose-limiting adverse events. High-dose dexamethasone might increase this risk even further because it causes lymphopenia and has an immunosuppressive effect. Furthermore, steroid-induced hyperglycemia may contribute to an increased risk of infection as well. None of our patients in the cohorts with the lowest dose of dexamethasone had febrile neutropenia and only one patient had grade 1 hyperglycemia, while in the first two cohorts, up to one third of patients had grade 1–2 hyperglycemia and four patients ($17\%$) were hospitalized due to febrile neutropenia. We realize that the number of patients in each cohort is limited. Nevertheless, we believe our data do support our hypothesis and show a clear trend in favor of lower dosages of dexamethasone. Moreover, our findings are in line with similar observations reported previously. For example, in the study of Kang et al., less infectious complications were observed in patients treated with a single dose of 10 mg of dexamethasone IV in comparison with patients who received the oral bid dosage of 4 mg of dexamethasone for 2 days in addition to the 10 mg of IV before docetaxel administration [32]. The lower risk of infectious complications was also observed in studies using lower doses of dexamethasone as anti-emetic regimens [37]. The occurrence of the fluid retention syndrome has a strong correlation with the cumulative dose of administered docetaxel. In phase II studies, before the introduction of corticosteroid premedication regimens, the median cumulative dose at the onset of fluid retention was between 300 and 400 mg/m2 [38]. Nowadays, most patients with prostate or breast cancer will not receive more than 400–450 mg/m2 of docetaxel. The mean dosages in our study were 412 mg/m2 (range 75–645) in patients with prostate cancer and a mean of 394 mg/m2 (range 100–600) in patients with breast cancer. In contrast, treatment regimens given in the early phase quite often entailed dosages up to 600–700 mg/m2. Therefore, a severe fluid retention syndrome is less likely to occur in current clinical practice. Two patients with breast cancer ($14\%$) and four patients with prostate cancer ($16\%$) developed mild-to-moderate edema. However, this may also have been an adverse effect of the dexamethasone premedication, as edema also occurred in the first cohorts that used higher dexamethasone dosages. This finding is all the more reason to critically assess the necessity of (high-dose) dexamethasone as an adjunct to chemotherapy. Patients are most at risk of developing hypersensitivity reactions (HSRs) during the first or second infusion, and life-threating reactions are very rare [39]. In our study, none of the patients developed a severe HRS, despite using lower dosages of dexamethasone premedication, and the only non-severe HRS (grade 1) occurred during the first chemotherapy cycle. Parinyanitakul et al. and Barrosa-Sousa et al. showed that in patients with early breast cancer treated with paclitaxel, dexamethasone premedication could be withheld safely if patients did not experience HSR in response to the two previous cycles [40,41]. It is conceivable that the same applies for dexamethasone if prescribed as an adjunct to docetaxel administration. Our trial was limited by the small number of patients enrolled in the study, especially the number of patients with breast cancer. We initially planned to include at least 36 patients in six cohorts of patients with breast cancer. However, during the study period, breast cancer treatment protocols in the Netherlands changed rapidly to include weekly paclitaxel instead of 3-weekly docetaxel because of a more favorable side-effect profile. As a consequence, we prematurely stopped the inclusion of patients with breast cancer. In the 14 patients we included, it appeared safe to reduce the dexamethasone dose by more than half (from a cumulative dose of 48 mg to 16 mg), as it did not increase the incidence of HSRs or fluid retention. In view of the results in patients with prostate cancer, it is conceivable that the dexamethasone dose can be reduced even further. Future studies with a larger number of patients could possibly establish the “median effective dose (ED50)” of prophylactic dexamethasone for different docetaxel regimens. The lower dosage of dexamethasone could indeed be worthwhile in view of its potential side effects and the growing evidence that hyperglycemia and hyperinsulinemia (the metabolic effects of dexamethasone) may be associated with poorer outcomes in patients with cancer [42,43,44]. However, since dexamethasone is also used for its effective antiemetic properties, it may not be appropriate to omit it completely. Nevertheless, recent clinical trials have demonstrated the benefit of using prednisone, instead of dexamethasone, in conjunction with docetaxel for the treatment of advanced prostate cancer [45]. Furthermore, Tanaka et al. showed that docetaxel combined with 0.5 mg of dexamethasone orally twice a day results in a PSA response and good survival efficacy in castration-resistant prostate cancer [46,47], whereas other trials in patients with metastatic castration-resistant prostate cancer displayed benefits regarding switching corticosteroid from prednisone to dexamethasone after progression in abiraterone acetate [48,49]. This implies that particular patients with advanced prostate cancer might benefit from the use of corticosteroids. As dexamethasone is also used for its effective antiemetic properties, it may not be appropriate to omit it completely in docetaxel treatment. ## 5. Conclusions Our study strongly suggests that the prophylactic dose of dexamethasone as an adjunct of docetaxel infusion may be safely reduced without increasing the risk of hypersensitivity reactions or fluid retention syndrome. However, in view of the small number of patients included, larger studies are required to confirm our results. ## References 1. Schiff P.B., Fant J., Horwitz S.B.. Promotion of microtubule assembly in vitro by taxol. Nature (1979.0) 277 665-667. DOI: 10.1038/277665a0 2. Cortes J.E., Pazdur R.. Docetaxel. J. Clin. Oncol. (1995.0) 13 2643-2655. DOI: 10.1200/JCO.1995.13.10.2643 3. Schrijvers D., Wanders J., Dirix L., Prove A., Vonck I., Van O.A., Kaye S.. Coping with toxicities of docetaxel (Taxotere). Ann. Oncol. (1993.0) 4 610-611. DOI: 10.1093/oxfordjournals.annonc.a058599 4. Mezzano V., Giavina-Bianchi P., Picard M., Caiado J., Castells M.. Drug desensitization in the management of hypersensitivity reactions to monoclonal antibodies and chemotherapy. BioDrugs (2014.0) 28 133-144. DOI: 10.1007/s40259-013-0066-x 5. Dieras V., Chevallier B., Kerbrat P., Krakowski I., Roche H., Misset J.L., Lentz M.A., Azli N., Murawsky M., Riva A.. A multicentre phase II study of docetaxel 75 mg m-2 as first-line chemotherapy for patients with advanced breast cancer: Report of the Clinical Screening Group of the EORTC. European Organization for Research and Treatment of Cancer. Br. J. Cancer (1996.0) 74 650-656. DOI: 10.1038/bjc.1996.416 6. Hudis C.A., Seidman A.D., Crown J.P., Balmaceda C., Freilich R., Gilewski T.A., Hakes T.B., Currie V., Lebwohl D.E., Baselga J.. Phase II and pharmacologic study of docetaxel as initial chemotherapy for metastatic breast cancer. J. Clin. Oncol. (1996.0) 14 58-65. DOI: 10.1200/JCO.1996.14.1.58 7. Ravdin P.M., Burris H.A., Cook G., Eisenberg P., Kane M., Bierman W.A., Mortimer J., Genevois E., Bellet R.E.. Phase II trial of docetaxel in advanced anthracycline-resistant or anthracenedione-resistant breast cancer. J. Clin. Oncol. (1995.0) 13 2879-2885. DOI: 10.1200/JCO.1995.13.12.2879 8. Ten Bokkel Huinink W.W., Prove A.M., Piccart M., Steward W., Tursz T., Wanders J., Franklin H., Clavel M., Verweij J., Alakl M.. A phase II trial with docetaxel (Taxotere) in second line treatment with chemotherapy for advanced breast cancer. A study of the EORTC Early Clinical Trials Group. Ann. Oncol. (1994.0) 5 527-532. DOI: 10.1093/oxfordjournals.annonc.a058907 9. Chevallier B., Fumoleau P., Kerbrat P., Dieras V., Roche H., Krakowski I., Azli N., Bayssas M., Lentz M.A., Van G.M.. Docetaxel is a major cytotoxic drug for the treatment of advanced breast cancer: A phase II trial of the Clinical Screening Cooperative Group of the European Organization for Research and Treatment of Cancer. J. Clin. Oncol. (1995.0) 13 314-322. DOI: 10.1200/JCO.1995.13.2.314 10. Fumoleau P., Chevallier B., Kerbrat P., Krakowski Y., Misset J.L., Maugard-Louboutin C., Dieras V., Azli N., Bougon N., Riva A.. A multicentre phase II study of the efficacy and safety of docetaxel as first-line treatment of advanced breast cancer: Report of the Clinical Screening Group of the EORTC. Ann. Oncol. (1996.0) 7 165-171. DOI: 10.1093/oxfordjournals.annonc.a010544 11. Valero V., Jones S.E., Von Hoff D.D., Booser D.J., Mennel R.G., Ravdin P.M., Holmes F.A., Rahman Z., Schottstaedt M.W., Erban J.K.. A phase II study of docetaxel in patients with paclitaxel-resistant metastatic breast cancer. J. Clin. Oncol. (1998.0) 16 3362-3368. DOI: 10.1200/JCO.1998.16.10.3362 12. EMA Label Docetaxel 13. Riva A., Fumoleau P., Roché H.. Efficacy and safety of different corticosteroid premedications in breast cancer patients treated with Taxotere. Proc. Am. Soc Clin Oncol. (1997.0) 14. Hickish T., Astras G., Thomas P., Penfold S., Purandare L., Hickish T.F., Kerr D.. Glucose intolerance during adjuvant chemotherapy for breast cancer. J. Natl. Cancer Inst. (2009.0) 101 537. DOI: 10.1093/jnci/djp025 15. Vardy J., Chiew K.S., Galica J., Pond G.R., Tannock I.F.. Side effects associated with the use of dexamethasone for prophylaxis of delayed emesis after moderately emetogenic chemotherapy. Br. J. Cancer (2006.0) 94 1011-1015. DOI: 10.1038/sj.bjc.6603048 16. Yoo K.E., Kang R.Y., Lee J.Y., Lee Y.J., Suh S.Y., Kim K.S., Kim H.S., Lee S.H., Lee B.K.. Awareness of the adverse effects associated with prophylactic corticosteroid use during docetaxel therapy. Support Care Cancer (2015.0) 23 1969-1977. DOI: 10.1007/s00520-014-2547-y 17. Jiralerspong S., Kim E.S., Dong W., Feng L., Hortobagyi G.N., Giordano S.H.. Obesity, diabetes, and survival outcomes in a large cohort of early-stage breast cancer patients. Ann. Oncol. (2013.0) 24 2506-2514. DOI: 10.1093/annonc/mdt224 18. Marrone M.T., Selvin E., Barber J.R., Platz E.A., Joshu C.E.. Hyperglycemia, classified with multiple biomarkers simultaneously in men without diabetes, and risk of fatal prostate cancer. Cancer Prev. Res. (2019.0) 12 103-112. DOI: 10.1158/1940-6207.CAPR-18-0216 19. Herr I., Pfitzenmaier J.. Glucocorticoid use in prostate cancer and other solid tumours: Implications for effectiveness of cytotoxic treatment and metastases. Lancet Oncol. (2006.0) 7 425-430. DOI: 10.1016/S1470-2045(06)70694-5 20. Kumar R.. Emerging role of glucocorticoid receptor in castration resistant prostate cancer: A potential therapeutic target. J. Cancer (2020.0) 11 696-701. DOI: 10.7150/jca.32497 21. Zhou F., Shi Y., Zhao G., Aufderklamm S., Murray K.S., Jin B.. A narrative review of the role of glucocorticoid receptors in prostate cancer: Developments in last 5 years. Transl. Androl. Urol. (2022.0) 11 1189-1199. DOI: 10.21037/tau-22-501 22. Sakellakis M., Flores L.J.. Is the glucocorticoid receptor a key player in prostate cancer?: A literature review. Medicine (2022.0) 101 e29716. DOI: 10.1097/MD.0000000000029716 23. Hu J., Chen Q.. The role of glucocorticoid receptor in prostate cancer progression: From bench to bedside. Int. Urol. Nephrol. (2017.0) 49 369-380. DOI: 10.1007/s11255-016-1476-8 24. Puhr M., Hoefer J., Eigentler A., Ploner C., Handle F., Schaefer G., Kroon J., Leo A., Heidegger I., Eder I.. The Glucocorticoid Receptor Is a Key Player for Prostate Cancer Cell Survival and a Target for Improved Antiandrogen Therapy. Clin. Cancer Res. (2018.0) 24 927-938. DOI: 10.1158/1078-0432.CCR-17-0989 25. McCune J.S., Hawke R.L., LeCluyse E.L., Gillenwater H.H., Hamilton G., Ritchie J., Lindley C.. In vivo and in vitro induction of human cytochrome P4503A4 by dexamethasone. Clin. Pharmacol. Ther. (2000.0) 68 356-366. DOI: 10.1067/mcp.2000.110215 26. Teo Y.L., Saetaew M., Chanthawong S., Yap Y.S., Chan E.C., Ho H.K., Chan A.. Effect of CYP3A4 inducer dexamethasone on hepatotoxicity of lapatinib: Clinical and in vitro evidence. Breast Cancer Res. Treat. (2012.0) 133 703-711. DOI: 10.1007/s10549-012-1995-7 27. Common Terminology Criteria for Adverse Events (CTCAE). (2011.0) 28. Chan S., Friedrichs K., Noel D., Pinter T., Van Belle S., Vorobiof D., Duarte R., Gil Gil M., Bodrogi I., Murray E.. Prospective randomized trial of docetaxel versus doxorubicin in patients with metastatic breast cancer. J. Clin. Oncol. (1999.0) 17 2341-2354. DOI: 10.1200/JCO.1999.17.8.2341 29. Jones S.E., Erban J., Overmoyer B., Budd G.T., Hutchins L., Lower E., Laufman L., Sundaram S., Urba W.J., Pritchard K.I.. Randomized phase III study of docetaxel compared with paclitaxel in metastatic breast cancer. J. Clin. Oncol. (2005.0) 23 5542-5551. DOI: 10.1200/JCO.2005.02.027 30. James N.D., Sydes M.R., Clarke N.W., Mason M.D., Dearnaley D.P., Spears M.R., Ritchie A.W., Parker C.C., Russell J.M., Attard G.. Addition of docetaxel, zoledronic acid, or both to first-line long-term hormone therapy in prostate cancer (STAMPEDE): Survival results from an adaptive, multiarm, multistage, platform randomised controlled trial. Lancet (2016.0) 387 1163-1177. DOI: 10.1016/S0140-6736(15)01037-5 31. Sweeney C.J., Chen Y.H., Carducci M., Liu G., Jarrard D.F., Eisenberger M., Wong Y.N., Hahn N., Kohli M., Cooney M.M.. Chemohormonal Therapy in Metastatic Hormone-Sensitive Prostate Cancer. N. Engl. J. Med. (2015.0) 373 737-746. DOI: 10.1056/NEJMoa1503747 32. Kang R.Y., Yoo K.S., Han H.J., Lee J.Y., Lee S.H., Kim D.W., Lee Y.J.. Evaluation of the effects and adverse drug reactions of low-dose dexamethasone premedication with weekly docetaxel. Support Care Cancer (2017.0) 25 429-437. DOI: 10.1007/s00520-016-3420-y 33. Zhang M., Chen Z., Yang Y., Cheng H., Li C., He Q.. Pretreatment with different doses of dexamethasone in the prevention of docetaxel-induced hypersensitivity. Pak. J. Pharm. Sci. (2017.0) 30 61-65. PMID: 28603114 34. Chen N.X., Zhao F.F., Yan F., Zhang X.X.. Safe dose reduction of steroid pre-medication for docetaxel in head and neck neoplasm treatment. Acta Otolaryngol. (2016.0) 136 177-180. DOI: 10.3109/00016489.2015.1096957 35. Chouhan J.D., Herrington J.D.. Single premedication dose of dexamethasone 20 mg IV before docetaxel administration. J. Oncol. Pharm. Pract. (2011.0) 17 155-159. DOI: 10.1177/1078155210367950 36. Jacobs C., Hutton B., Mazzarello S., Smith S., Joy A., Amir E., Ibrahim M.F., Gregario N., Daigle K., Eggert L.. Optimisation of steroid prophylaxis schedules in breast cancer patients receiving docetaxel chemotherapy-a survey of health care providers and patients. Support Care Cancer (2015.0) 23 3269-3275. DOI: 10.1007/s00520-015-2731-8 37. Gjafa E., Ng K., Grunewald T., Galazi M., Skyllberg E., Wilson P., Alifrangis C., Shamash J.. Neutropenic sepsis rates in patients receiving BEP chemotherapy using olanzapine and reduced doses of dexamethasone compared to a standard antiemetic regimen. BJU Int. (2021.0) 127 205-211. DOI: 10.1111/bju.15175 38. Piccart M.J., Klijn J., Paridaens R., Nooij M., Mauriac L., Coleman R., Bontenbal M., Awada A., Selleslags J., Van V.A.. Corticosteroids significantly delay the onset of docetaxel-induced fluid retention: Final results of a randomized study of the European Organization for Research and Treatment of Cancer Investigational Drug Branch for Breast Cancer. J. Clin. Oncol. (1997.0) 15 3149-3155. DOI: 10.1200/JCO.1997.15.9.3149 39. Syrigou E., Dannos I., Kotteas E., Makrilia N., Tourkantonis I., Dilana K., Gkiozos I., Saif M.W., Syrigos K.N.. Hypersensitivity reactions to docetaxel: Retrospective evaluation and development of a desensitization protocol. Int. Arch. Allergy Immunol. (2011.0) 156 320-324. DOI: 10.1159/000324454 40. Parinyanitikul N., Tanpipattanakul W., Poovorawan N., Rattananupong T., Laoitthi P., Sithidetphaiboon P., Thanasanvimon S., Sriuranpong V.. Incidence of infusion hypersensitivity reaction after withholding dexamethasone premedication in early breast cancer patients not experiencing two previous cycles of infusion hypersensitivity reaction for weekly paclitaxel chemotherapy. Support Care Cancer (2018.0) 26 2471-2477. DOI: 10.1007/s00520-018-4087-3 41. Barroso-Sousa R., Vaz-Luis I., Di Meglio A., Hu J., Li T., Rees R., Sinclair N., Milisits L., Leone J.P., Constantine M.. Prospective Study Testing a Simplified Paclitaxel Premedication Regimen in Patients with Early Breast Cancer. Oncologist (2021.0) 26 927-933. DOI: 10.1002/onco.13960 42. Ferroni P., Riondino S., Laudisi A., Portarena I., Formica V., Alessandroni J., D’Alessandro R., Orlandi A., Costarelli L., Cavaliere F.. Pretreatment Insulin Levels as a Prognostic Factor for Breast Cancer Progression. Oncologist (2016.0) 21 1041-1049. DOI: 10.1634/theoncologist.2015-0462 43. Duggan C., Wang C.Y., Neuhouser M.L., Xiao L., Smith A.W., Reding K.W., Baumgartner R.N., Baumgartner K.B., Bernstein L., Ballard-Barbash R.. Associations of insulin-like growth factor and insulin-like growth factor binding protein-3 with mortality in women with breast cancer. Int. J. Cancer (2013.0) 132 1191-1200. DOI: 10.1002/ijc.27753 44. Derr R.L., Ye X., Islas M.U., Desideri S., Saudek C.D., Grossman S.A.. Association between hyperglycemia and survival in patients with newly diagnosed glioblastoma. J. Clin. Oncol. (2009.0) 27 1082-1086. DOI: 10.1200/JCO.2008.19.1098 45. Teply B.A., Luber B., Denmeade S.R., Antonarakis E.S.. The influence of prednisone on the efficacy of docetaxel in men with metastatic castration-resistant prostate cancer. Prostate Cancer Prostatic Dis. (2016.0) 19 72-78. DOI: 10.1038/pcan.2015.53 46. Tanaka N., Nishimura K., Okajima E., Ina K., Ogawa O., Nagata H., Akakura K., Fujimoto K., Gotoh M., Teramukai S.. Docetaxel-based chemotherapy combined with dexamethasone 1 mg daily oral administration for castration-resistant prostate cancer: Long-term outcomes. Int. J. Urol. (2019.0) 26 797-803. DOI: 10.1111/iju.14009 47. Tanaka N., Nishimura K., Okajima E., Ina K., Ogawa O., Nagata H., Akakura K., Fujimoto K., Gotoh M., Teramukai S.. The efficacy and safety of docetaxel-based chemotherapy combined with dexamethasone 1 mg daily oral administration: JMTO Pca 10-01 phase II trial. Jpn. J. Clin. Oncol. (2017.0) 47 247-251. DOI: 10.1093/jjco/hyw193 48. Yang Z., Ni Y., Zhao D., Zhang Y., Wang J., Jiang L., Chen D., Wu Z., Wang Y., He L.. Corticosteroid switch from prednisone to dexamethasone in metastatic castration-resistant prostate cancer patients with biochemical progression on abiraterone acetate plus prednisone. BMC Cancer (2021.0) 21. DOI: 10.1186/s12885-021-08670-2 49. Roviello G., Sobhani N., Corona S.P., D’Angelo A.. Corticosteroid switch after progression on abiraterone acetate plus prednisone. Int. J. Clin. Oncol. (2020.0) 25 240-246. DOI: 10.1007/s10147-019-01577-w
--- title: Construction of a Diagnostic m7G Regulator-Mediated Scoring Model for Identifying the Characteristics and Immune Landscapes of Osteoarthritis authors: - Liang Hao - Xiliang Shang - Yang Wu - Jun Chen - Shiyi Chen journal: Biomolecules year: 2023 pmcid: PMC10046530 doi: 10.3390/biom13030539 license: CC BY 4.0 --- # Construction of a Diagnostic m7G Regulator-Mediated Scoring Model for Identifying the Characteristics and Immune Landscapes of Osteoarthritis ## Abstract With the increasingly serious burden of osteoarthritis (OA) on modern society, it is urgent to propose novel diagnostic biomarkers and differentiation models for OA. 7-methylguanosine (m7G), as one of the most common base modification forms in post transcriptional regulation, through which the seventh position N of guanine (G) of messenger RNA is modified by methyl under the action of methyltransferase; it has been found that it plays a crucial role in different diseases. Therefore, we explored the relationship between OA and m7G. Based on the expression level of 18 m7G-related regulators, we identified nine significant regulators. Then, via a series of methods of machine learning, such as support vector machine recursive feature elimination, random forest and lasso-cox regression analysis, a total of four significant regulators were further identified (DCP2, EIF4E2, LARP1 and SNUPN). Additionally, according to the expression level of the above four regulators, two different m7G-related clusters were divided via consensus cluster analysis. Furthermore, via immune infiltration, differential expression analysis and enrichment analysis, we explored the characteristic of the above two different clusters. An m7G-related scoring model was constructed via the PCA algorithm. Meanwhile, there was a different immune status and correlation for immune checkpoint inhibitors between the above two clusters. The expression difference of the above four regulators was verified via real-time quantitative polymerase chain reaction. Overall, a total of four biomarkers were identified and two different m7G-related subsets of OA with different immune microenvironment were obtained. Meanwhile, the construction of m7G-related Scoring model may provide some new strategies and insights for the therapy and diagnosis of OA patients. ## 1. Introduction Being unpredictable and one of the most common chronic degenerative joint diseases, OA (OA) has a prevalence that increases with age. OA causes significant pain and disability [1,2,3]. Furthermore, inflammation or fibrosis of the infrapatellar fat pad is present in patients with OA, which is one of the well-established risk factors for the development of the pain caused by OA [4,5]. Moreover, pathological changes, including subchondral osteosclerosis, synovitis, fibrosis, and cartilage degeneration are closely associated with OA [6,7,8,9]. OA has a multifaceted etiology; and is caused by a combination of immune response, chronic inflammation, trauma, and biomechanical processes [10,11,12]. Congenital joint abnormalities, trauma, stress injury, obesity, sex, age, and knee gap narrowing [13] are all complicatedly associated with the development of OA [14].OA can have serious physical, emotional, and economic consequences; moreover, it is becoming an evolving public health issue that negatively affects the daily lives and quality of life of people [15]. With an increasing prevalence, overall, the age-standardized prevalence of OA increased by $7.5\%$ in Northern Europe between 1990 and 2015, with an annual increase of $43\%$ [16]. Clinical symptoms and imaging findings are required for making the standard diagnosis of OA; however, often, when these symptoms appear, patients have already reached the late stage of the disease [17]. Moreover, patients with OA show relatively severe symptoms and have poor treatment results because no current drug treatment has been found to reverse the progression of OA in the long term [15]. Therefore, it is even more important to look for novel diagnostic modalities that can help diagnose OA as early as possible, facilitating the timely treatment of patients. RNA modifications were not known as the “epitranscriptome” until 2015. Emerging studies on the function of these modifications have shown significant implications for human pathology [18]. Adenosine methylation is present in mRNAs and non-coding RNAs, such as circular RNA (circRNAs), microRNAs (miRNAs), and long-stranded non-coding RNAs (lncRNAs), which adjust their biogenesis and function [19]. The RNA methylation types can be divided into various modification types: m6A, m5C, m7G, and so on; of which m6A and m7G are the two most common types [20,21,22]. However, so far, more extensive studies on the association between m6A and OA are available, whereas there are few studies on m7G [23,24]. Moreover, there is a prevalence of m7G RNA modifications within mRNA, and their conservation, regulation, and dynamics as well as their roles in translational control have been shown. Modifications of the m7G cap are widely seen in OA-related mRNA [25]; they play an important role in the efficient translation, splicing, and stability of related mRNA and also affect the synthesis of related proteins. Modifications of m7G are also observed in OA-related mRNA and help in enabling the translation of OA-related mRNA. Thus, m7G, as a transcriptional marker, is important for protein translation and can be used as the basis for making diagnostic models for OA. Herein, we have analyzed numerous publicly available microarray datasets. Using differential analysis and algorithmic screening, we have acquired the most critical genes of m7G-regulators for forming an intersection. We have identified four significant regulators, including DCP2, EIF4E2, SNUPN, and LARP1, in combination with machine learning. Based on the expression of these four regulators, we have divided all the OA samples into m7G-related clusters. Next, we performed principal component analysis (PCA) to calculate the m7G score in the above two clusters. Then, we performed differential expression analysis, enrichment analysis, and immune infiltration to explore the characteristics of the two clusters. Finally, after the intersection of the differentially expressed genes (DEGs) between the above two clusters and DEGs between the normal samples (NM) and OA samples, we constructed a diagnostic model using LASSO Cox regression. Furthermore, we verified the abovementioned four regulators with the use of real-time quantitative polymerase chain reaction (RT-qPCR). ## 2.1. Data Acquisition and Processing By retrieving the keyword “OA” from the Gene Expression Omnibus database, datasets containing the synovial tissue samples of normal people and patients with OA were obtained [26]. Moreover, the standards for screening our datasets are as follows: [1] Homo sapiens Expression Profiling by array, [2] synovial tissue of OA from joint synovial biopsies, [3] datasets containing complete information about the samples, [4] one biopsy sample per subject was analyzed without replicates. The detailed information regarding the datasets used in our study is listed in Table 1. Then, “inSilicoMerging” [27] was used to merge and the “limma” package (v3.42.2) in R software [28] was used to remove the batch effect of these three data sets; finally, a data set containing the synovial tissue samples of 26 patients with OA and 20 normal people was obtained. Furthermore, 24 m7G-regulators collected in the gene sets GOMF_RNA_7_METHYLGUANOSINE_CAP_BINDING, GOMF_M7G_5_PPPN_DIPHOSPHATASE_ACTIVITY, and GOMF_RNA_CAP_BINDING, were summarized, wherein only 18 regulators were annotated in our data sets [29]. In this study, the association between the expression of these 18 genes and diseases as well as that between m7G-regulators self-expression was studied. The results were visualized using heat maps. The Wilcoxon signed-rank test was performed to select some significant regulators. ## 2.2. Enrichment Analysis Gene ontology (GO) analysis is a common method to annotate gene products and the functions of genes, including cellular component, biological pathway, and molecular function [30]. The Kyoto Encyclopedia of Genes and Genomes (KEGG) is a useful database for the systematic analysis of gene functions and associated high-level genomic functional information [31]. We used “clusterProfiler” (version 3.14.3) and R software packages “org.Hs.eg.db” (version 3.1.0) for the GO and KEGG pathway analysis. The maximum size gene sets were set to 5000 genes and the minimum to 5; the analysis results with a p-value of <0.05 were considered significant. ClueGO is an important plug-in of Cytoscape, which can be used for GO enrichment analysis. It was also used for enrichment analysis in our study [32]. ## 2.3. Construction and Verification of Prediction Model To more accurately screen the m7G-regulators associated with the occurrence of OA, significant m7G-regulators and OA were extracted using the support vector machine recursive feature elimination (SVM-RFE) algorithm and random forest algorithm (RF) [33], and the two analysis results were intersected using the Venn diagram. Finally, to determine the best prognostic characteristic regulators, the LASSO Cox regression analysis was performed on the basis of the abovementioned results [34]. Then, cluster analysis was performed on the finally screened feature genes using the “ConsensusClusterPlus” package; accordingly, the samples were divided into two categories. PCA was used for extracting PC1 and PC2 to form signature scores. Afterward, the above scores were applied to construct the m7G score: m7G score = Σ (PC1i + PC2i). ## 2.4. Immune Infiltration The genes significantly associated with 28 immune cell types from the literature were collected (Table S1). Then, the expression of these immune genes were linked with the distribution of 28 types of immune cells using the single-sample gene set enrichment analysis method [35]; combined with our m7G regulators; we finally analyzed the association of 28 immune cells with different m7G-related clusters and immune cells with m7G characteristic genes. ## 2.5. Construction of Diagnostic Model of OA To better explore the characteristic of our two m7G-related clusters, differential expression analysis was performed to assess the DEGs between the two m7G-related clusters and those between the NM and OA samples with the cutoff criteria of \|log2FC\| > 1 ($p \leq 0.05$). After the intersection of the two different types of DEGs, a diagnostic model was constructed using the overlapped DEGs via the LASSO Cox regression. The diagnostic score was as follows: Diagnostic Score = ∑i Coefficientsi * Expression level of signaturei. ## 2.6. Collection of Our Synovial Samples In our study, synovial tissue samples of OA were obtained from patients undergoing surgery due to knee OA ($$n = 15$$). The normal synovial samples were from patients who underwent surgery for meniscus laceration of the knee joint caused by trauma ($$n = 15$$). ## 2.7. RT-qPCR Trizol (Invitrogen, Waltham, MA, USA) reagent was used to extract total RNA from the synovial tissue samples of normal people and patients with OA; the Prime Script TMRT kit (Takara, RR047A) was used to reverse transcribe to obtain cDNA. Finally, the SYBR Premix Ex Taq II Kit (Takara, Japan) was used for PCR amplification according to the manufacturer’s instructions. And we used Bio-Rad (CFX96) of the UK for RT-qPCR. The primer sequences are listed in Table S2. ## 2.8. Statistical Analysis R software (version 4.2.1) and its related software packages were used to process and analyze data ($p \leq 0.05$). We used the Wilcoxon signed-rank test to assess the significance of difference between the two groups. And a t-test was used for the analysis of the result of RT-qPCR. Afterward, Sangerbox was used to visualize the results of the receiver operating characteristic curves and Wilcoxon signed-rank test; the “RMS” package in R software was used to visualize Nomograms. ## 3.1. Identification of Significant m7G-Regulators in OA We calculated the Spearman correlation coefficient among these 18 regulators, wherein several regulators demonstrated significant correlation on the basis of the expression levels of 18 m7G-regulators in our datasets which were merged with the three datasets from GEO (Figure 1A). Meanwhile, the interaction association of these 18 regulators was revealed by constructing a protein-protein interaction network (Figure 1B). Then, we used the Wilcoxon signed-rank test to identify the significant regulators in our training set. Thus, nine significant regulators ($p \leq 0.05$) were obtained, including DCP2, IFIT5, EIF4E2, NUDT11, NUDT3, LARP1, SNUPN, LSM1, and CYFIP1 (Figure 1C). The heat map shows the expression level of these nine significant regulators (Figure 1D). ## 3.2. Enrichment Analysis for Our Significant m7G-Regulators The ClueGO plug-in in Cytoscape was first used to perform enrichment analysis ($p \leq 0.05$) for comprehensively exploring the function of the above nine significant m7G-regulators in OA. Thus, the term with the largest proportion was “m7G(5′)pppN diphosphatase activity” ($53.85\%$); furthermore, the rest of the terms revealed that our significant m7G-regulators were almost involved in the pathways of RNA metabolism and translation progress (Figure 2A). Meanwhile, the MCODE plug-in was used to extract the important clusters of the ClueGO results. The clusters with a high score are shown in Figure 2B–D; the cluster with the highest score was the “m7G(5′)pppN diphosphatase activity” pathway, which was consistent with the biological function of these regulators. Moreover, the GO, KEGG, and Reactome analyses were performed to ensure the preciseness of our research ($p \leq 0.05$) (Figure 2E–I). Correspondingly, almost all results of different methods of enrichment analysis revealed that these regulators focused on RNA metabolism. Interestingly, pathways associated with the immune system and other pathways, including the viral myocarditis, interferon signaling, and HIF-1 signaling pathways, were found, indicating that these m7G-regulators played a significant role in RNA modification as well as immune and other fields. ## 3.3. Selection of Significant m7G-Regulators via Machine Learning In line with the above analysis, we explored the important and key role of OA by using several methods of machine learning to further identify some significant regulators in OA. First, SVM-RFE was performed to evaluate the diagnostic effectiveness of these regulators. Thus, seven regulators were obtained. Meanwhile, another method, RF, was used to calculate the importance of these regulators (Figure 3A,B). With a score of >2, six regulators were selected (Figure 3C,D). Then, we intersected the results of SVM-RFE and RF, through which four regulators (EIF4E2, DCP2, SNUPN, and LARP1) were considered the final crucial regulators (Figure 3E). Furthermore, the LASSO Cox regression was performed to verify the diagnostic effectiveness of these four regulators (Figure 3F,G). Thus, all four regulators were regarded as significantly diagnostic signatures. Moreover, based on the expression matrix of GSE32317, the nomogram was further exhibited to reveal the efficiency of the above four m7G-regulators in distinguishing early- and end-stage OA, and the calibration curve revealed the accuracy of our model (Figure 3H,I). ## 3.4. Identification of Two Different m7G-Related Clusters According to the expression levels of the four key regulators selected via machine learning, we divided our OA samples in the training set into two m7G-related clusters with the most appropriate K value ($K = 2$) via consensus cluster analysis (Figure 4A–C). Furthermore, the PCA diagram revealed a significant difference between clusters A and B (Figure 4D). Meanwhile, all four regulators showed a significant statistical difference between m7G-related clusters A and B (Figure 4E,F). Based on the PCA algorithm, an m7G score module was calculated to distinguish the above two clusters ($p \leq 0.05$), which was higher in m7G-related cluster B and lower in m7G-related cluster A (Figure 4G). ## 3.5. GSEA, Immune Infiltration, and Immune Checkpoint Characteristics in m7G–Related Clusters To better describe the characteristics and functions in the abovementioned m7G-related clusters, the GSEA analysis was performed. We identified three pathways with a p-value of <0.05, including TGF_BETA_SIGNALING, ALLOGRAFT_REJECTION, and ESTROGEN_RESPONSE_EARLY, which indicated that the m7G score was mainly associated with the metabolism and immune system (Figure 5A). Therefore, we performed the immune infiltration analysis and the mantel test to demonstrate the association between the four significant regulators and the infiltration score of the 28 immune cells. Interestingly, a stronger association was noted in cluster A, indicating that the lower m7G score suggested prominently elevated infiltration of immune cells in patients with OA (Figure 5B,C). Furthermore, the Pearson correlation coefficient between the expressions of a series of immune checkpoint-related genes in the above two clusters and the m7G score was calculated for more comprehensively exploring the immune signature between these two clusters. For the immune checkpoint inhibitors, the m7G score in m7G-related cluster A was positively associated with most of the inhibitors (Figure 5D). On the contrary, the m7G score in m7G-related cluster B was negatively correlated with most inhibitors (Figure 5E). ## 3.6. Exploration of Difference between the above Two Clusters and Construction of a Diagnostic Model To further emphasize the significance of the m7G-related clusters, differential expression analysis was performed with \|log2FC\| > 1 ($p \leq 0.05$) as the cutoff. Thus, 113 DEGs were identified, including 46 upregulated and 67 downregulated, which were visualized using the volcano and heat maps (Figure 6A,B). Moreover, the DEGs between NM and OA and between the above two clusters intersected. Finally, eight overlapped DEGs were obtained (CRYBB1, N6AMT1, SNORA21, HAUS2, P2RX3, RRN3P1, CC2D1A, and FKBP5), which were considered the candidate factors extracted for the LASSO Cox regression (Figure 6C,D). We regarded lambda-min: 0.0469 as the optimal value after running the 10-fold cross-validation. Thus, five factors were selected to construct our diagnostic model for OA: Diagnostic value = (−0.00806730919977491 × SNORA21) + (−0.00181179794976226 × HAUS2) + (−0.0134814157908044 × CC2D1A) + (−0.00123114088948981 × FKBP5) + (0.00983579963789009 × N6AMT1) (Figure 6E,F). Furthermore, all samples were randomly divided into two different subsets with a ratio of 1:1: a verification set and a training set. According to the above formula of the diagnostic score, the Wilcoxon signed-rank test was performed to explore the statistical difference between the NM and OA samples. Finally, the diagnostic score of both sets demonstrated a significant difference ($p \leq 0.05$) (Figure 6G,I). In addition, the area under the receiver operating characteristic curve of the diagnostic model in the verification and training set was 85.4701 and 93.0070, respectively, which further indicated the excellent effectiveness of our diagnostic model (Figure 6H,J). Moreover, to further verify the accuracy of the diagnostic model, another external dataset GSE12021 was selected, wherein a significant difference was observed in the diagnostic value between NM and OA samples ($p \leq 0.05$) (Figure 6K). Meanwhile, in GSE12021, the area under the receiver operating characteristic curve of the diagnostic model was 97.7778, demonstrating the accuracy of our model (Figure 6L). ## 3.7. Validation of the Four Significant m7G-Regulators in the Synovial Tissue of Patients with OA To explore the abovementioned four significant m7G-regulators (DCP2, EIF4E2, SNUPN, and LARP1) in OA, the synovial tissues of normal people and patients with OA were collected from the Second Affiliated Hospital of Nanchang University and the demographic data of the patients included in our study is listed in Table 2. Then, the expression difference in the RNA level was verified using qRT-PCR. Thus, except for LARP1, the other three regulators exhibited a significant high-expression level in OA, which was consistent with our above analysis (Figure 7A–D). ## 4. Discussion OA is usually assumed to be caused by non-inflammatory factors; that is, a series of mechanical stresses that destroys the cartilage. However, recently, some associated inflammatory factors have also been shown to contribute to OA development, allowing inflammatory cells to infiltrate the synovium [36]. Moreover, inflammatory cytokines play an essential role in the progression of OA by stimulating the production of matrix metalloproteinases and thus increasing matrix degradation [37]. An increasing number of studies have focused on the effects of nucleic acid site changes on the cell function and even body activities, wherein RNA modification plays a critical role. This can be seen in the methylation of m7G, which is located in the inner part of tRNA and rRNA and plays a significant role in coordinating numerous functions during the mRNA lifecycle. This is largely accredited to the protein factors that particularly bind to the cap structures, the cap-binding complex in the nucleus, and eIF4E in the cytoplasm. m7G is accountable for mRNA processing and nuclear export in the nucleus and is needed for effective pre-mRNA splicing. In vivo, mRNA interacts with protein factors throughout its life cycle and also plays a role in transcription termination and exosome degradation [38]. In the present study, we have briefly discussed the association between m7G and OA. Several research gaps still exist despite a growing interest in this field. Thus, we have attempted to create a novel prospect for the clinical diagnosis of OA. In this study, three datasets were downloaded from the Gene Expression Omnibus database, namely GSE55235, GSE55457, and GSE55584. After unified treatment, 18 regulatory factors associated with m7G were identified and analyzed from the data of 26 patients with OA and 20 normal people. Nine statistically significant m7G regulators were obtained; their association was strong enough. Second, a protein-protein interaction network was constructed to enrich and analyze these regulators using GO and KEGG. The results showed that these genes were mainly involved in no MTG-related dephosphorylation, RNA metabolism, RNA modification, and other processes and were enriched in hypoxia-related GFR signaling, insulin metabolism-related, and virus-related pathways. Next, two machine learning algorithms (RF and SVM-RFE) were used to further screen the nine genes, obtain the intersection, and further obtain four key genes, EIF4E2, DCP2, SNUPN, and LARP1. Based on these four genes, the diagnostic effectiveness of these four genes was further verified using the LASSO Cox regression. Furthermore, based on the expression levels of the abovementioned four hub genes, the 26 OA samples in the training set were divided into two different m7G-related clusters; the PCA algorithm was used to further calculate the m7G score to differentiate the two subtypes. Immune infiltration analysis demonstrated that cluster A was more closely associated with the immune system. Moreover, to better exhibit the characteristic of our m7G-related clusters, a diagnostic model was constructed for calculating the diagnostic score using the differential expression analysis and LASSO Cox regression. Finally, the differential expression of the abovementioned four genes in the synovial tissues of patients with OA was verified using an external validation set and RT-qPCR. Although unilaterally LARP1 exhibited a non-statistically significant difference between NM and OA in RT-qPCR, the m7G-related Score to distinguish two different m7G-related clusters of OA exhibited an accurate efficiency and the diagnostic model constructed via these four m7G-regulators played an extraordinary role in the diagnosis of OA. EIF4E2 is a protein-coding gene located in the p-body and belongs to part of the mRNA cap-binding active complex [39,40]. In the early stage of initiation, EIF4E2 recognizes and binds m7G mRNA cap, activates ubiquitin protein ligase-binding activity, and participates in miRNA-mediated translational inhibition. Contrary to EIF4E, EIF4E2 is unable to bind EIF4G (EIF4G1, EIF4G2, or EIF4G3), signifying that it assembles EIF4F by competing with EIF4E and blocking it in the cap region [41]. EEIF4E2-related diseases include casket-Siris syndrome 2, melanoma, cancer, viruses, and so on. The related pathways include the PI3K-Akt signaling and innate immune system pathways. RNA-binding and ubiquitin protein ligase binding are the GO annotations associated with this gene. Shaohong Chen et al. proposed that TNRC6 competes with EIF4E1 to recruit EIF4E2 for targeting mRNA, thus blocking translation initiation. Moreover, EIF4E2 mainly inhibits the expression of genes at the translational level but does not significantly affect the level of coding mRNA [42]. Moreover, Mir-29b is associated with EIF4E2. However, Mir-29b can silence premature AID expression in naive B cells, thus reducing the probability of inappropriate and potentially dangerous deamination activity [43]. Cadherin-22, a cell-cell adhesion molecule, is upregulated by promoter eIF4E2-mediated mTORC1-independent translational control during hypoxia; the novel function of cadherin-22 acts as a hypoxia-specific cell surface molecule and is involved in cancer cell migration, invasion, and adhesion [44]. However, no previous studies have noted and discussed the association between EIF4E2 and OA or the role of EIF4E2 in OA, which may become a potential topic for studies in the future. The protein encoded by DCP2 is a key component of the mRNA uncoating complex required for mRNA degradation. It removes the 7-methylguanine cap structure from mRNA and then degrades from the 5′ end. The involved pathways are an unfolded protein response and the regulation of activated PAK-2P34 by proteasome-mediated degradation. The GO annotations associated with this gene include RNA binding and manganese ion binding. Moreover, T cell intracellular antigen-1 (TIA-1)-induced transformation silencing promotes the decay of selected mRNAs, and TIA-1-mediated decay is inhibited by small interfering RNAs targeting the 5′-3′ (e.g., DCP2) or 3′-5′ (e.g., exosomal component Rrp46) decay pathway, suggesting that TIA-1 sensitizes mRNA to both major decay pathways [45]. Interestingly, there is no study on the association between DCP2 and OA. This is also the novelty of our study, and it is worthy of more detailed research. SNUPN is also a protein-coding gene, acts as a U snRNP-specific nuclear import adapter, and participates in the trimethylguanosine cap-dependent nuclear import of U snRNPs. SNUPN is associated with chronic lymphocytic leukemia, cancer, and so on. The related pathways include the translocation of pre-mRNA containing introns and SLBP-independent mature mRNA. The GO annotations associated with this gene include outdated protein transporter activity and RNA cap binding. Moreover, XPO1 binding to various proteins is mediated by the recognition of leucine-rich nuclear export signals at the N-terminus of SNUPN, thus transporting proteins. Overexpression or dysfunction of XPO1 has been reported in different cancers [46]. As the role of SNUPN in OA has not been explored by previous studies, our study is the first one to demonstrate a link between the two. To understand the role of SNUPN in OA in detail, further deeper research is necessary. LARP1 is a class of RNA-binding proteins that regulates the translation of specific target mRNAs downstream of the mTORC1 complex and plays a role in growth signaling and nutrient availability while regulating cell growth and proliferation [47]. The diseases associated with LARP1 are dengue virus and hepatocellular carcinoma. The pathways involved include disease and SARS-CoV-2 infection. The GO annotations associated with this gene include RNA binding and translation initiation factor binding. Furthermore, RNMT selectively regulates the LARP1 target (TOP mRNA in the terminal polypyrimidine tract) expression. Increased ribosome abundance leads to the upregulation of RNMT for coordinating mRNA capping and processing and increasing translational capacity during T-cell activation [48]. Meanwhile, the association between LARP1 and OA is still not studied. Our study is the first one to show that LARP1 may have an impact on the pathogenesis of OA, but it still needs further exploration. Furthermore, up to now, the diagnosis of OA has become clearer and accurate, which is based on X-rays and clinical symptoms [17,49,50]. However, with OA development, the joint pain caused by OA and its effect on the daily life and exercise capacity of patients becomes more and more severe. Thus, early and timely diagnosis of OA is urgent and necessary. In our study, the nomogram constructed based on the GSE32317 further revealed that our four regulators (DCP2, EIF4E2, SNUPN, and LARP1) also showed excellent accuracy to distinguish early- and end-stage OA, which provided effective insight to the early diagnosis of OA. In summary, in this study, the collected data sets were unified and merged into a whole for analysis to avoid homogeneity. The number of samples used was also significant. Simultaneously, we used two different machine learning methods to avoid the one-sidedness of screening methods and make the results more convincing. Furthermore, the scoring model of m7G was constructed for the first time, which is progressive. However, not enough synovial tissue samples were collected. Moreover, the role of our four m7G-regulators in the occurrence and development of OA was not explored with detailed experiments, which needs to be explored in the future. Moreover, the five-gene diagnostic model obtained via our m7G-related cluster was verified only via external and internal datasets without experiments; thus, its diagnostic accuracy needs to be further validated. In conclusion, we constructed an m7G-related scoring model, which can significantly differentiate patients with OA, and correlated it with different statuses of the immune microenvironment, based on which we constructed a diagnostic model to diagnose patients with OA. ## References 1. Loeser R.F., Goldring S.R., Scanzello C.R., Goldring M.B.. Osteoarthritis: A disease of the joint as an organ. Arthritis. Rheum. (2012) 64 1697-1707. DOI: 10.1002/art.34453 2. Geneen L.J., Moore R.A., Clarke C., Martin D., Colvin L.A., Smith B.H.. Physical activity and exercise for chronic pain in adults: An overview of Cochrane Reviews. Cochrane Database Syst. Rev. (2017) 4 CD011279. DOI: 10.1002/14651858.CD011279.pub3 3. Singh J.A., Noorbaloochi S., MacDonald R., Maxwell L.J.. Chondroitin for osteoarthritis. Cochrane Database Syst. Rev. (2015) 1 CD005614. DOI: 10.1002/14651858.CD005614.pub2 4. Yoon K.H., Tak D.H., Ko T.S., Park S.E., Nam J., Lee S.H.. Association of fibrosis in the infrapatellar fat pad and degenerative cartilage change of patellofemoral joint after anterior cruciate ligament reconstruction. Knee (2017) 24 310-318. DOI: 10.1016/j.knee.2016.10.018 5. Culvenor A.G., Cook J.L., Collins N.J., Crossley K.M.. Is patellofemoral joint osteoarthritis an under-recognised outcome of anterior cruciate ligament reconstruction? A narrative literature review. Br. J. Sports Med. (2013) 47 66-70. DOI: 10.1136/bjsports-2012-091490 6. Hugle T., Geurts J.. What drives osteoarthritis?—Synovial versus subchondral bone pathology. Rheumatology (2017) 56 1461-1471. DOI: 10.1093/rheumatology/kew389 7. Sellam J., Berenbaum F.. The role of synovitis in pathophysiology and clinical symptoms of osteoarthritis. Nat. Rev. Rheumatol. (2010) 6 625-635. DOI: 10.1038/nrrheum.2010.159 8. Zhang L., Li M., Li X., Liao T., Ma Z., Zhang L., Xing R., Wang P., Mao J.. Characteristics of sensory innervation in synovium of rats within different knee osteoarthritis models and the correlation between synovial fibrosis and hyperalgesia. J. Adv. Res. (2022) 35 141-151. DOI: 10.1016/j.jare.2021.06.007 9. Schulze-Tanzil G.. Intraarticular Ligament Degeneration Is Interrelated with Cartilage and Bone Destruction in Osteoarthritis. Cells (2019) 8. DOI: 10.3390/cells8090990 10. Wang Q., Zhu W., Cheng Y., Wu X.. Comparative study of female Chinese mitten crabs based on their sizes and weights. J. Food Sci. Technol. (2022) 59 2572-2583. DOI: 10.1007/s13197-021-05274-3 11. Zargar H., Zargar-Shoshtari K., Lotan Y., Shah J.B., van Rhijn B.W., Daneshmand S., Spiess P.E., Black P.. Final Pathological Stage after Neoadjuvant Chemotherapy and Radical Cystectomy for Bladder Cancer-Does pT0 Predict Better Survival than pTa/Tis/T1?. J. Urol. (2016) 195 886-893. DOI: 10.1016/j.juro.2015.10.133 12. Batushansky A., Zhu S., Komaravolu R.K., South S., Mehta-D’souza P., Griffin T.M.. Fundamentals of OA. An initiative of Osteoarthritis and Cartilage. Obesity and metabolic factors in OA. Osteoarthr. Cartil. (2022) 30 501-515. DOI: 10.1016/j.joca.2021.06.013 13. Glyn-Jones S., Palmer A.J., Agricola R., Price A.J., Vincent T.L., Weinans H., Carr A.J.. Osteoarthritis. Lancet (2015) 386 376-387. DOI: 10.1016/S0140-6736(14)60802-3 14. Mobasheri A.. The future of osteoarthritis therapeutics: Emerging biological therapy. Curr. Rheumatol. Rep. (2013) 15 385. DOI: 10.1007/s11926-013-0385-4 15. Hawker G.A.. Osteoarthritis is a serious disease. Clin. Exp. Rheumatol. (2019) 37 3-6. PMID: 31621562 16. Kiadaliri A.A., Lohmander L.S., Moradi-Lakeh M., Petersson I.F., Englund M.. High and rising burden of hip and knee osteoarthritis in the Nordic region, 1990–2015. Acta Orthop. (2018) 89 177-183. DOI: 10.1080/17453674.2017.1404791 17. Braun H.J., Gold G.E.. Diagnosis of osteoarthritis: Imaging. Bone (2012) 51 278-288. DOI: 10.1016/j.bone.2011.11.019 18. Jonkhout N., Tran J., Smith M.A., Schonrock N., Mattick J.S., Novoa E.M.. The RNA modification landscape in human disease. RNA (2017) 23 1754-1769. DOI: 10.1261/rna.063503.117 19. Yang D., Qiao J., Wang G., Lan Y., Li G., Guo X., Xi J., Ye D., Zhu S., Chen W.. N6-Methyladenosine modification of lincRNA 1281 is critically required for mESC differentiation potential. Nucleic Acids Res. (2018) 46 3906-3920. DOI: 10.1093/nar/gky130 20. Zhang H., Shi X., Huang T., Zhao X., Chen W., Gu N., Zhang R.. Dynamic landscape and evolution of m6A methylation in human. Nucleic Acids Res. (2020) 48 6251-6264. DOI: 10.1093/nar/gkaa347 21. Yang X., Yang Y., Sun B.F., Chen Y.S., Xu J.W., Lai W.Y., Li A., Wang X., Bhattarai D.P., Xiao W.. 5-methylcytosine promotes mRNA export-NSUN2 as the methyltransferase and ALYREF as an m(5)C reader. Cell Res. (2017) 27 606-625. DOI: 10.1038/cr.2017.55 22. Luo Y., Yao Y., Wu P., Zi X., Sun N., He J.. The potential role of N(7)-methylguanosine (m7G) in cancer. J. Hematol. Oncol. (2022) 15 63. DOI: 10.1186/s13045-022-01285-5 23. Chen X., Gong W., Shao X., Shi T., Zhang L., Dong J., Shi Y., Shen S., Qin J., Jiang Q.. METTL3-mediated m(6)A modification of ATG7 regulates autophagy-GATA4 axis to promote cellular senescence and osteoarthritis progression. Ann. Rheum. Dis. (2022) 81 87-99. DOI: 10.1136/annrheumdis-2021-221091 24. Yang J., Zhang M., Yang D., Ma Y., Tang Y., Xing M., Li L., Chen L., Jin Y., Ma C.. m(6)A-mediated upregulation of AC008 promotes osteoarthritis progression through the miR-328-3p–AQP1/ANKH axis. Exp. Mol. Med. (2021) 53 1723-1734. DOI: 10.1038/s12276-021-00696-7 25. Malbec L., Zhang T., Chen Y.S., Zhang Y., Sun B.F., Shi B.Y., Zhao Y.L., Yang Y., Yang Y.G.. Dynamic methylome of internal mRNA N(7)-methylguanosine and its regulatory role in translation. Cell Res. (2019) 29 927-941. DOI: 10.1038/s41422-019-0230-z 26. Edgar R., Domrachev M., Lash A.E.. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. (2002) 30 207-210. DOI: 10.1093/nar/30.1.207 27. Taminau J., Meganck S., Lazar C., Steenhoff D., Coletta A., Molter C., Duque R., de Schaetzen V., Weiss Solis D.Y., Bersini H.. Unlocking the potential of publicly available microarray data using inSilicoDb and inSilicoMerging R/Bioconductor packages. BMC Bioinform. (2012) 13. DOI: 10.1186/1471-2105-13-335 28. Ritchie M.E., Phipson B., Wu D., Hu Y., Law C.W., Shi W., Smyth G.K.. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. (2015) 43 e47. DOI: 10.1093/nar/gkv007 29. Tomikawa C.. 7-Methylguanosine Modifications in Transfer RNA (tRNA). Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19124080 30. Wilkerson M.D., Hayes D.N.. ConsensusClusterPlus: A class discovery tool with confidence assessments and item tracking. Bioinformatics (2010) 26 1572-1573. DOI: 10.1093/bioinformatics/btq170 31. Kanehisa M., Goto S.. KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res. (2000) 28 27-30. DOI: 10.1093/nar/28.1.27 32. Bindea G., Mlecnik B., Hackl H., Charoentong P., Tosolini M., Kirilovsky A., Fridman W.H., Pages F., Trajanoski Z., Galon J.. ClueGO: A Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics (2009) 25 1091-1093. DOI: 10.1093/bioinformatics/btp101 33. Zhao C., Roy S.. Special issue: Cilia and ciliopathies. J. Cell Physiol. (2022) 237 2611-2612. DOI: 10.1002/jcp.30816 34. Zhang G.Z., Wu Z.L., Li C.Y., Ren E.H., Yuan W.H., Deng Y.J., Xie Q.Q.. Development of a Machine Learning-Based Autophagy-Related lncRNA Signature to Improve Prognosis Prediction in Osteosarcoma Patients. Front. Mol. Biosci. (2021) 8 615084. DOI: 10.3389/fmolb.2021.615084 35. Barbie D.A., Tamayo P., Boehm J.S., Kim S.Y., Moody S.E., Dunn I.F., Schinzel A.C., Sandy P., Meylan E., Scholl C.. Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1. Nature (2009) 462 108-112. DOI: 10.1038/nature08460 36. Nedunchezhiyan U., Varughese I., Sun A.R., Wu X., Crawford R., Prasadam I.. Obesity, Inflammation, and Immune System in Osteoarthritis. Front. Immunol. (2022) 13 907750. DOI: 10.3389/fimmu.2022.907750 37. Woodell-May J.E., Sommerfeld S.D.. Role of Inflammation and the Immune System in the Progression of Osteoarthritis. J. Orthop. Res. (2020) 38 253-257. DOI: 10.1002/jor.24457 38. Ramanathan A., Robb G.B., Chan S.H.. mRNA capping: Biological functions and applications. Nucleic Acids Res. (2016) 44 7511-7526. DOI: 10.1093/nar/gkw551 39. Amaya Ramirez C.C., Hubbe P., Mandel N., Bethune J.. 4EHP-independent repression of endogenous mRNAs by the RNA-binding protein GIGYF2. Nucleic Acids Res. (2018) 46 5792-5808. DOI: 10.1093/nar/gky198 40. Peter D., Weber R., Sandmeir F., Wohlbold L., Helms S., Bawankar P., Valkov E., Igreja C., Izaurralde E.. GIGYF1/2 proteins use auxiliary sequences to selectively bind to 4EHP and repress target mRNA expression. Genes Dev. (2017) 31 1147-1161. DOI: 10.1101/gad.299420.117 41. Rosettani P., Knapp S., Vismara M.G., Rusconi L., Cameron A.D.. Structures of the human eIF4E homologous protein, h4EHP, in its m7GTP-bound and unliganded forms. J. Mol. Biol. (2007) 368 691-705. DOI: 10.1016/j.jmb.2007.02.019 42. Chen S., Gao G.. MicroRNAs recruit eIF4E2 to repress translation of target mRNAs. Protein Cell (2017) 8 750-761. DOI: 10.1007/s13238-017-0444-0 43. Recaldin T., Hobson P.S., Mann E.H., Ramadani F., Cousins D.J., Lavender P., Fear D.J.. miR-29b directly targets activation-induced cytidine deaminase in human B cells and can limit its inappropriate expression in naive B cells. Mol. Immunol. (2018) 101 419-428. DOI: 10.1016/j.molimm.2018.07.028 44. Kelly N.J., Varga J.F.A., Specker E.J., Romeo C.M., Coomber B.L., Uniacke J.. Hypoxia activates cadherin-22 synthesis via eIF4E2 to drive cancer cell migration, invasion and adhesion. Oncogene (2018) 37 651-662. DOI: 10.1038/onc.2017.372 45. Yamasaki S., Stoecklin G., Kedersha N., Simarro M., Anderson P.. T-cell intracellular antigen-1 (TIA-1)-induced translational silencing promotes the decay of selected mRNAs. J. Biol. Chem. (2007) 282 30070-30077. DOI: 10.1074/jbc.M706273200 46. Jain P., Kanagal-Shamanna R., Wierda W., Keating M., Sarwari N., Rozovski U., Thompson P., Burger J., Kantarjian H., Patel K.P.. Clinical and molecular characteristics of XPO1 mutations in patients with chronic lymphocytic leukemia. Am. J. Hematol. (2016) 91 E478-E479. DOI: 10.1002/ajh.24496 47. Burrows C., Abd Latip N., Lam S.J., Carpenter L., Sawicka K., Tzolovsky G., Gabra H., Bushell M., Glover D.M., Willis A.E.. The RNA binding protein Larp1 regulates cell division, apoptosis and cell migration. Nucleic Acids Res. (2010) 38 5542-5553. DOI: 10.1093/nar/gkq294 48. Galloway A., Kaskar A., Ditsova D., Atrih A., Yoshikawa H., Gomez-Moreira C., Suska O., Warminski M., Grzela R., Lamond A.I.. Upregulation of RNA cap methyltransferase RNMT drives ribosome biogenesis during T cell activation. Nucleic Acids Res. (2021) 49 6722-6738. DOI: 10.1093/nar/gkab465 49. Tack A., Preim B., Zachow S.. Fully automated Assessment of Knee Alignment from Full-Leg X-Rays employing a “YOLOv4 And Resnet Landmark regression Algorithm” (YARLA): Data from the Osteoarthritis Initiative. Comput. Methods Programs Biomed. (2021) 205 106080. DOI: 10.1016/j.cmpb.2021.106080 50. Prieto-Alhambra D., Judge A., Javaid M.K., Cooper C., Diez-Perez A., Arden N.K.. Incidence and risk factors for clinically diagnosed knee, hip and hand osteoarthritis: Influences of age, gender and osteoarthritis affecting other joints. Ann. Rheum. Dis. (2014) 73 1659-1664. DOI: 10.1136/annrheumdis-2013-203355
--- title: Oxidative Stress Biomarkers among Schizophrenia Inpatients authors: - Magdalena Więdłocha - Natalia Zborowska - Piotr Marcinowicz - Weronika Dębowska - Marta Dębowska - Anna Zalewska - Mateusz Maciejczyk - Napoleon Waszkiewicz - Agata Szulc journal: Brain Sciences year: 2023 pmcid: PMC10046541 doi: 10.3390/brainsci13030490 license: CC BY 4.0 --- # Oxidative Stress Biomarkers among Schizophrenia Inpatients ## Abstract Background. Finding the associations between schizophrenia symptoms and the biomarkers of inflammation, oxidative stress and the kynurenine pathway may lead to the individualization of treatment and increase its effectiveness. Methods. The study group included 82 schizophrenia inpatients. The Positive and Negative Symptoms Scale (PANSS), the Brief Assessment of Cognition in Schizophrenia (BACS) and the Calgary Depression in Schizophrenia Scale were used for symptom evaluation. Biochemical analyses included oxidative stress parameters and brain-derived neurotrophic factor (BDNF). Results. Linear models revealed the following: [1] malondiadehyde (MDA), N-formylkynurenine (N-formKYN), advanced oxidation protein products (AOPP), advanced glycation end-products of proteins (AGE) and total oxidative status (TOS) levels are related to the PANSS-total score; [2] MDA, reduced glutathione (GSH) and BDNF levels are related to the PANSS-negative score; [3] TOS and kynurenine (KYN) levels are related to the PANSS-positive score; [4] levels of total antioxidant status (TAS) and AOPP along with the CDSS score are related to the BACS-total score; [5] TAS and N-formKYN levels are related to the BACS-working memory score. Conclusions. Oxidative stress biomarkers may be associated with the severity of schizophrenia symptoms in positive, negative and cognitive dimensions. The identification of biochemical markers associated with the specific symptom clusters may increase the understanding of biochemical profiles in schizophrenia patients. ## 1. Introduction Despite advances in the knowledge of schizophrenia research, the efficacy of antipsychotic treatment is not satisfactory. It is estimated that approximately $30\%$ of patients respond to treatment by achieving complete remission of symptoms, $30\%$ manage to achieve partial remission while 20–$30\%$ do not respond to treatment [1]. Antipsychotics are much more effective against positive symptoms than against negative and cognitive symptoms. At the same time, it is the severity of the deficit symptoms and cognitive dysfunction that is the most important predictor of psychosocial functioning in schizophrenia [2]. The identification of sensitive and specific biomarkers that indicate susceptibility, which may help explain the clinical features, clinical response to treatment and improve therapy precision, remains one of the greatest challenges in schizophrenia research. Different profiles of gene expression abnormalities, epigenetic patterns, metabolic and inflammatory markers in the peripheral blood are found among patients with schizophrenia [3]. The involvement of inflammatory processes in the pathophysiology of schizophrenia is well documented. Many studies have focused on the search for disease biomarkers among proinflammatory and anti-inflammatory factors. It has been suggested that abnormalities of the immune system including cytokine profiles may be a biochemical endophenotype in some schizophrenia patient populations [4]. Abnormal cytokine levels are found in schizophrenic patients from the onset of the first episode psychosis (FEP), as well as in their first-degree relatives [5,6]. Pro-oxidant and antioxidant factors are also among the commonly assessed biomarkers in schizophrenia. Researchers report significant disease-related abnormalities suggesting increased oxidative stress (OS) and decreased activity of antioxidant mechanisms [7,8,9]. The increased production of reactive oxygen species (ROS) and reactive nitrogen species (RNS) as well as reduced antioxidant potential are considered risk factors for the development of schizophrenia. It was shown that different fractions of brain proteins exhibit signs of oxidative damage [10]. ROS stimulate the expression of genes encoding proinflammatory cytokines including tumor necrosis factor α (TNF-α), interleukin (IL) 1 and IL-6 and affect the activity of the nuclear factor (NF-kB) that determines proinflammatory activity [11]. There exist interrelationships between chronic OS and chronic inflammation that lead to impaired processes of apoptosis and neurogenesis and the occurrence of neurodegenerative changes which have a documented role in the etiopathogenesis of schizophrenia [12]. Craddock et al. found that in patients with schizophrenia, the intensity of the immune response was associated with the level of OS [13]. Moreover, antioxidant activity was shown to be decreased in schizophrenia patients [14,15,16]. Studies also indicated that metabolites of the kynurenine pathway (KP) may be a valuable biomarker of the disease process in schizophrenia [17,18]. The tryptophan metabolism activity within KP depends on the severity of inflammation [19]. KP metabolites directly or indirectly affect the oxidation-reduction status. This may be a pro-oxidant or antioxidant effect depending on the cell type, pH, exposure time and oxidation-reduction potential of the environment [20]. Under physiological conditions, the potentially adverse effects are balanced by oxidation-reduction balance mechanisms. The loss of this balance can lead to OS and cell destruction [17,18]. Another biomarker studied in patients with schizophrenia is brain-derived neurotrophic factor (BDNF). Its plasma and serum levels correlate with concentrations in the cerebrospinal fluid (CSF) and with concentrations of N-acetylaspartate, a marker of neuronal cortex integrity [21]. In schizophrenia, reduced BDNF levels have been found in the prefrontal cortex and hippocampus [22]. A study conducted by Belbasis et al. involving a large umbrella meta-analysis has shown that decreased BDNF levels are a significant factor associated with the increased risk of schizophrenia [23]. The influence of the above factors on specific dimensions of the schizophrenia features is being actively investigated, yet is not clearly defined [4,6,8,9]. The identification of biochemical markers associated with the specific symptom clusters may be the basis for the creation of a ‘biochemical profile’ to individualize treatment and increase its effectiveness. The assessment of peripheral biomarkers may also indicate the most appropriate adjuvant strategies to enhance the efficacy of schizophrenia treatment. The aim of this study was to determine the relationship between the levels of selected inflammatory, pro-oxidant and antioxidant markers; initial KP metabolites; BDNF; and symptom severity of the different dimensions of schizophrenia: positive, negative, cognitive and depressive. ## 2.1. Study Group The study group consisted of 82 patients with a diagnosis of paranoid schizophrenia hospitalized in the general psychiatric wards of the Mazovian Specialist Health Centre in Pruszków between 2017 and 2019. Each patient gave written informed consent to participate in the study. All subjects went through at least 2 psychotic episodes. During the study, patients were under antipsychotic treatment in monotherapy. Exclusion criteria included comorbid mental disorders, including organic disorders, intellectual disability, abuse of or addiction to alcohol and/or psychoactive substances (except nicotine and caffeine) or concomitant neurological, autoimmune, infectious or chronic somatic diseases. In addition, patients with present inflammatory markers in screening blood tests (C-reactive protein, CRP > 5mg/L and white blood count, WBC > 10,000/μL) were excluded from the study. The study received a positive opinion from the Bioethics Committee at the Warsaw Medical University. ## 2.2. Sociodemographic and Clinical Data Collection Data collected included age, gender, duration of illness (DI), number of psychotic episodes (NPE), body mass index (BMI), smoking and antipsychotic treatment used. To compare the doses of different antipsychotics (AP), the doses used were converted to chlorpromazine equivalents (CPZE) according to Woods et al. [ 24]. ## 2.3. Clinical Scales Used to Assess the Severity of Psychopathological Symptoms The Positive and Negative Symptoms Scale (PANSS) was used to assess the severity of positive (PANSS-P), negative (PANSS-N), general symptoms (PANSS-G) and the total score (PANSS-T) [25]. Cognitive functioning was assessed using the Brief Assessment of Cognition in Schizophrenia (BACS). It is a tool developed to assess cognitive functions in schizophrenia, validated with standardized batteries of cognitive function tests in patients with schizophrenia and in healthy individuals with respect to age, race and education. The BACS allows the examination of 6 domains that are particularly impaired in schizophrenia and have the greatest impact on functioning: verbal memory (BACS-VM), working memory (BACS-WM), motor speed (BACS-MS), verbal fluency (BACS-VF), attention and speed of information processing (BACS-ASP) and executive functions (BACS-EF) [25]. We also assessed the total BACS result (BACS-T). The results of each BACS subtest and BACS-T were standardized by calculating the deviation (Z-score) from the average result for healthy controls (0-score). The 0-score along with norms for age and gender were determined by the tool authors [26]. The Calgary Depression Scale for Schizophrenia (CDSS) was used to assess the severity of depressive symptoms in schizophrenia [27]. ## 2.4. Blood Sample Collection and Storage Venous blood samples were collected after overnight (12 h) fasting into an EDTA-containing tube. Subsequently, blood samples were centrifuged at 2000× g for 10 min at room temperature. The obtained serum was immediately frozen and stored at −80 °C until the biochemical analyses. All samples were analyzed in a single batch. ## 2.5.1. IL-6 IL-6 levels were determined spectrophotometrically using ELISA. The diagnostic kit used was the Human IL-6 Quantikine ELISA Kit according to the manufacturer’s recommendations. ## 2.5.2. Antioxidant Enzymes The catalase (CAT) activity was determined colorimetrically in triplicate samples using a method based on measuring the rate of decomposition of hydrogen peroxide in phosphate buffer at pH 7.0 at 240 nm. One unit of CAT activity was defined as the amount of enzymes that degraded 1 mmol H2O2 in 1 min [28]. The glutathione peroxidase (GPx) activity was determined in triplicate samples using a colorimetric method according to Mansson-Rahemtull et al. involving the reduction of 5,5′-dithio-bis-2-nitrobenzoic acid to thionitrobenzoic acid, which was then reacted with OSCN- ions (hypothiocyanates): a product of oxidation of potassium thiocyanate by GPx. The decrease in absorbance was measured at 412 nm, and 5 absorbance measurements were taken every 30 s [29]. ## 2.5.3. Reduced Glutathione (GSH) The GSH concentration was determined in duplicate samples using a colorimetric method based on the reduction of 5,5′-dithiobis-2-nitrobenzoic acid to 2-nitro-5-mercaptobenzoic acid under the influence of GSH contained in the test sample. The concentration of 2-nitro-5-mercaptobenzoic acid formed in the reaction was measured at 412 nm and calculated from a calibration curve determined for GSH solutions [30]. ## 2.5.4. Oxidation-Reduction Balance Parameters The total antioxidant status (TAS) was determined in triplicates using a colorimetric method measuring changes in absorbance of an ABTS+ (3-ethylbenzothiazoline-6-sulphonic acid radical cation) solution under the influence of antioxidants contained in the test sample at 660 nm. The TAS was calculated from the standard curve for Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) [31]. The total oxidative status (TOS) was determined in triplicate samples by a bichromatic method ($\frac{560}{800}$ nm) involving the oxidation of Fe ions2+ to Fe3+ in the presence of oxidants contained in the sample, followed by the detection of Fe3+ by xyleneol orange. The TOS concentration was presented as μmol H2O2 Equiv./L (micromolar hydrogen peroxide equivalent per liter) [32]. The oxidative stress index (OSI) was presented as the quotient of TOS to TAS and was expressed in %. ## 2.5.5. Products of Oxidative Damage of Proteins and Lipids The advanced oxidation protein products’ (AOPP) concentration was determined in duplicate samples by a colorimetric method measuring the oxidation capacity of iodide iodine at 340 nm. Serum samples were pre-diluted in phosphate-buffered saline at a ratio of 1:5 [33]. The advanced glycation end-products of proteins’ (AGE) concentration was determined by measuring the fluorescence characteristic of AGE derivatives (350 nm/440 nm) in duplicate samples. Serum samples were previously diluted in PBS solution in a ratio of 1:5 [33]. The malondialdehyde (MDA) concentration was determined in duplicate samples by the thiobarbituric acid colorimetric method. Absorbance was measured at 535 nm and 1,1′,3,3′-tetraethoxypropane was used as a standard [34]. ## 2.5.6. Protein Glyco-Oxidation Products To assess dityrosine (DITYR), kynurenine (KYN) and N-formylkynurenine (N-formKYN) concentrations, samples were diluted in 0.1 M sulfuric acid at a volume ratio of 1:10. Fluorescence was measured at wavelengths $\frac{330}{415}$ nm (DITYR), $\frac{365}{480}$ nm (KYN), $\frac{325}{434}$ nm (N-formKYN) and $\frac{95}{340}$ (Trp) [35]. ## 2.5.7. Nitrosative Stress Parameters The nitrogen oxide (NO) level was determined colorimetrically by measuring its stable metabolites NO3− and NO2− in a Griess reaction. Changes in optical density were measured at 543 nm [36]. The peroxynitrite (ONOO−) level was determined by a nitration reaction resulting in nitrophenols [37]. The 3-nitrotyrosine (3-NT) level was determined spectrophotometrically using ELISA. The Immundiagnostik AG diagnostic kit was used according to the manufacturer’s instructions. ## 2.5.8. BDNF BDNF levels were determined spectrophotometrically using an ELISA method. The Total BDNF Quantikine ELISA Kit was used according to the manufacturer’s recommendations. ## 2.6. Statistical Analysis A statistical analysis was performed using the Statistica package version 13.3, scripts implemented in R, version 4.1.2 and Excel spreadsheets. Quantitative variables were described by the arithmetic mean with standard deviation, median and range and binary variables by percentages. The Shapiro–Wilk W-test was used to determine whether a quantitative variable came from a population with a normal distribution. In the univariate analysis, relationships between quantitative variables were determined by identifying Spearman correlations along with r-correlation coefficient values. In order to identify factors that were predictors of greater severity of psychopathological symptoms from the individual dimensions of schizophrenia, generalized linear models (GLM) were used. For all statistical calculations, the p-value of 0.05 was taken as the limit of significance. For biochemical markers, as these were not normally distributed, analyses have been performed using natural logarithm transformed values. As the study population size was relatively small, Type I and II errors had an increased risk of occurring, which has been accounted for where applicable. ## 3.1. Characteristics of the Study Population Demographic and clinical data are shown in Table 1. The study group included 82 participants: $57.3\%$ males and $42.6\%$ females. The median of illness duration and number of psychotic episodes were, respectively, 89 months and 4.5. The BMI mean and median values were in the upper limit of healthy weight range. Most of the study participants were tobacco smokers. ## 3.2. Results of Clinical Scales and Measurement of Biochemical Parameters in Study Population Table 2 shows the means with SD, ranges, medians and $95\%$CI for the scores of the clinical scales and subscales in the study population. In the study population, all evaluated cognitive function scores were lower than the standards adopted for the healthy controls. Among all cognitive domains, working memory Z-scores were the highest. Table 3 shows the means with SD, ranges, medians and $95\%$CI for the values of each biochemical parameter in the study population. ## 3.3. Correlations between Assessed Parameters The univariate analysis examined correlations between biochemical parameters and PANSS, BACS and CDSS scores and demographic data. An analysis of the results was conducted taking multiple comparisons into account, and the significance level value was adjusted using the Holm–Bonferroni method to account for Type 1 errors. There was a high positive correlation of PANSS-P levels with TOS and OSI values and an average positive correlation with KYN levels (Table 4). There was also a high positive correlation of PANSS-N levels with MDA values and an average negative correlation with GSH levels (Table 5). We found a high positive correlation of PANSS-T levels with TOS and OSI values and an average positive correlation with IL-6 (Table 6). In the field of cognitive functions, there was an average positive correlation of BACS-WM scores with TAS values and an average negative correlation with N-formKYN levels (Table 7). Moreover, we found a high negative correlation of BACS-ASP and BACS-EF levels with ONOO- values and an average negative correlation with NO and AOPP levels (Table 8). An analysis of correlations between BACS, PANSS and CDSS scores revealed only an average negative correlation of BACS-T with CDSS scores (Table 9). Among the correlations between PANSS, BACS and CDSS scores with demographic data, no significant correlations were observed. ## 3.4. Linear Model Describing the Relationship between Biochemical and Clinical Parameters and the Severity of Psychopathological Symptoms as Measured by the PANSS-T The model was constructed to determine associations between the PANSS-T score and biochemical and clinical parameters. The model was built in a backward, stepwise manner optimizing the AIC value, and for the final model, an F-test was performed to assess model significance and R2 was calculated to indicate goodness of fit. The final model consisted of five predictors: MDA, N-formKYN, AOPP, AGE and TOS. These had a significant ($p \leq 0.05$) effect on the PANSS-T score (Table 10), with the model being significant and of satisfactory fit. Based on the constructed model, higher values of MDA, N-formKYN, AOPP, AGE and TOS were associated with a higher PANSS-T score. Additionally, DI, NPE, BMI, smoking, CPZE and AP were all included in the final model but had no significant impact on the PANSS-T score and did not influence other associations. Subsequent linear models describing the relationships between biochemical and clinical parameters and symptom severity of each dimension of schizophrenia were constructed in the same manner as the model above. ## 3.5. Linear Model Describing the Relationships between Biochemical and Clinical Parameters and the Severity of Negative Symptoms as Measured by the PANSS-N Subscale The model consisted of three predictors: MDA, BDNF and GSH. These had a significant ($p \leq 0.05$) effect on the PANSS-N score (Table 11). From the constructed model, higher values of MDA together with lower levels of GSH and BDNF were associated with a higher PANSS-N score. Additionally, DI, NPE, BMI, smoking, CPZE and AP were all included in the final model but had no significant impact on the PANSS-T score and did not influence other associations. ## 3.6. Linear Model Describing the Relationship between Biochemical and Clinical Parameters and the Severity of Positive Symptoms as Measured by the PANSS-P Subscale The model consisted of two predictors, TOS and KYN. These had a significant ($p \leq 0.05$) effect on the PANSS-P score (Table 12). From the constructed model, higher TOS and KYN values were associated with higher PANSS-P scores. Additionally, DI, NPE, BMI, smoking, CPZE and AP were all included in the final model but had no significant impact on the PANSS-T score and did not influence other associations. ## 3.7. Linear Model Describing the Relationship between Biochemical and Clinical Parameters and the BACS-WM Result The model consisted of two predictors, TAS and N-formKYN. These had a significant ($p \leq 0.05$) effect on the BACS-WM score (Table 13). From the constructed model, higher TAS and lower N-formKYN values were associated with a higher BACS-WM score. Additionally, DI, NPE, BMI, smoking, CPZE and AP were all included in the final model but had no significant impact on the PANSS-T score and did not influence other associations. ## 3.8. Linear Model Describing the Relationship between Biochemical and Clinical Parameters and the BACS-T The linear model consisted of three predictors: TAS, AOPP and CDSS. These had a significant ($p \leq 0.05$) effect on the BACS-T score (Table 14). From the constructed model, higher TAS values and lower AOPP and CDSS values were associated with a higher BACS-T score. Additionally, DI, NPE, BMI, smoking, CPZE and AP were all included in the final model but had no significant impact on the PANSS-T score and did not influence other associations. Statistical models constructed using the GLM to describe the relationship between biochemical and clinical parameters and PANSS-G subscale scores, BACS-ASP, BACS-EF, BACS-VF, BACS-VM, BACS-MS and CDSS were characterized by a coefficient of determination R2< 0.3. This indicates an unsatisfactory fit of the model to data; therefore, these models were not included in the results. ## 4.1. Inflammation and Oxidative-Nitrosative Stress and the Severity of Psychopathological Symptoms in Schizophrenia This study showed that a higher severity of OS expressed by OSI and higher concentrations of products of protein (AOPP, AGE, N-formKYN) and lipid (MDA) oxidation processes are associated with greater severity of psychopathological symptoms in schizophrenia. The OSI parameter identifies a shift in the oxidation-reduction balance towards oxidative processes. As reported by Juchnowicz et al. when compared with HC, schizophrenia patients, regardless of stage, exhibited several times higher TOS and OSI values, and these parameters were considered a risk marker for the development of the disease. However, in contrast to the results of the present study, no association was found between OSI levels and the severity of schizophrenia symptoms [38]. Other researchers found higher OSI levels in patients with marked deficit symptoms, in remission and with chronic illness compared to patients without deficit symptoms and those who did not achieve remission [39,40]. In the study by Sertan Copoglu et al. an association between oxidative DNA damage and the severity of schizophrenia symptoms was demonstrated. Patients not in remission showed higher levels of TOS, OSI and 8-hydroxydeoxyguanine (8-OHdG), a marker of oxidative DNA damage, and reduced levels of TAS. In contrast, patients in remission showed a positive correlation between TOS and OSI levels and 8-OH-dG [40]. AOPP is a sensitive marker of oxidative damage to proteins, particularly albumin, as well as fibrinogen and lipoproteins [41]. Oxidized proteins have an ability to activate the inflammatory response and can also induce a sudden release of large amounts of ROS by neutrophils and the production of chemotactic factors for inflammatory cells. They also stimulate the production of IL-8 and TNF-α by monocytes. Due to the above properties, AOPP is considered a marker and mediator of the proinflammatory effect of OS [42]. In relation to the results of the present study, it can therefore be speculated that inflammation may be a factor associated with the severity of psychopathological symptoms. This was also suggested by other reports. Zhang et al. stated that the cytokine system’s dysregulation and oxidative stress may induce clinical symptoms of schizophrenia [43]. Liemburg et al. found that CRP was associated with positive and negative symptom severity in a large sample of outpatients with chronic schizophrenia. Higher concentrations of the proinflammatory cytokines such as TNF-α and IL-6 were related to a deficit syndrome, while the TNF-α level was associated with negative symptom severity [44]. Guidara et al. also demonstrated an effect of AOPP levels on the severity of schizophrenia symptoms [45]. On the other hand, in a study by Juchnowicz et al. this parameter was related to the age of the patients and the duration of the illness, but no relationship was found with symptom severity [46]. AGEs, including methylglyoxal and 3-deoxyglucosone, are formed during the oxidation of lipids, glucose and amino acids. They are highly reactive and, as with AOPPs, tend to accumulate, generate ROS and induce inflammation [45]. The binding of AGEs to the membrane receptor of macrophages, myocytes, neurons and other cells results in the increased synthesis of proinflammatory cytokines and secondary production of ROS. AGEs also decrease the antioxidant potential by modifying and inactivating CAT, GPx and superoxide dismutase (SOD) [47]. Juchnowicz et al. found higher levels of AGE in patients with schizophrenia compared to the healthy controls [46]. A systematic review by Kouidrat et al. also found an accumulation of AGEs in patients with schizophrenia [48]. The level of MDA, a product of lipid peroxidation, along with the concentration of protein oxidation and glycation products, was associated in our study with greater severity of pathological symptoms in the study group of schizophrenia patients. The results from studies to date show an increase in lipid peroxidation in schizophrenia, and this is also supported by meta-analyses [8,49,50]. Arvindagshan et al. also showed that the severity of lipid peroxidation, as measured by the level of end-products of the process, was associated with symptom severity in schizophrenic patients [51]. Recent research indicated a positive correlation between the PANSS-P score and MDA as well as CRP levels in FEP drug-naïve patients. Dudzinska et al. suggested that MDA may be an early indicator of ongoing low-grade inflammation [52]. Lipid peroxidation causes structural and functional damage to cell membrane phospholipids and polysaturated fatty acids [53]. Guidara et al. suggested that the specific behavioral symptomatology of schizophrenia may be related to arachidonic acid oxidative damage and its consequences for central nervous system (CNS) neurochemistry [45]. The present study also found an association between the severity of psychopathological symptoms in PANSS and levels of N-formKYN and KYN: compounds that are products of tryptophan oxidation. Juchnowicz et al. also found an association between greater severity of psychopathological symptoms in PANSS and higher KYN and lower TAS levels in patients with schizophrenia [38]. Metabolism within KP provides neuroactive kynurenine derivatives that may significantly influence the pathophysiology of schizophrenia by modulating dopaminergic, glutamatergic and nicotinergic transmission and disrupting the oxidative-reduction balance. Kynurenic acid (KYNA), an N-methyl-D-aspartate receptor (NMDAR) antagonist, has a neuroprotective effect at normal concentrations, whereas at elevated concentrations it leads to excessive NMDAR blockade, contributing to psychotic symptoms and cognitive deficits [18]. Other metabolites of KP, such as 3-hydroxykynurenine and the NMDAR agonist quinolinic acid, have neurotoxic and neurodegenerative effects [19]. In our study, the levels of only the initial KP metabolites were assessed, so it is not possible to draw conclusions regarding a direct effect of all KP metabolites on the clinical picture in schizophrenia. However, it appears that the association between N-formKYN and KYN levels and the severity of psychopathological symptoms may be explained by increased tryptophan metabolism within KP and possibly increased levels of downstream, neurotoxic metabolites. ## 4.2. Biochemical Markers Associated with Increased Positive Symptoms in Schizophrenia A multivariate regression model analysis showed that higher OS as measured by TOS together with higher KYN concentrations were a predictor of higher positive symptom severity in the study group. As described above, KP metabolites may have pro-oxidant and neurotoxic effects. Previous studies indicate that there is a relationship between inflammatory markers, OS and KP metabolites and the severity of positive symptoms. It was shown that the severity of positive symptoms in schizophrenia is associated with higher levels of KYN and ferric reducing ability of plasma (FRAP) [38]. KYN was also postulated to be a biomarker in monitoring the progress of treatment [54]. SOD was also found to be negatively correlated with positive symptom severity, but this was not confirmed in the meta-analysis by Flatow et al. [ 7,55,56]. Dietrich-Muszalska et al. instead showed an association between the severity of positive symptoms and the severity of lipid peroxidation [57]. They also found a correlation between levels of the proinflammatory cytokines IL-1, IL-7 and IL-8 and the severity of delusions [58]. ## 4.3. Biochemical Markers Associated with the Severity of Negative Symptoms in Schizophrenia Statistical modeling of the obtained data showed that a shift in the oxidation-reduction balance towards oxidative processes expressed by increased lipid peroxidation along with lower GSH and BDNF levels is associated with greater severity of negative symptoms in schizophrenia. Previous studies confirmed the association between antioxidant potential and the severity of negative symptoms in schizophrenia. Li et al. showed a negative correlation between TAS levels and negative symptoms in patients with first episode psychosis (FEP). Moreover, they found that the presence of OS at the onset of psychosis influenced the subsequent course of the illness, especially the development of negative symptoms [59]. The results of the study by Albayrak et al. confirmed these relationships. Patients with persistent negative symptoms were found to have lower levels of TAS and higher levels of OSI compared to patients with non-deficit schizophrenia and healthy controls (HC). They also showed that higher CAT levels in patients with schizophrenia were associated with a lower risk of negative symptoms, shorter duration of illness and fewer episodes [39]. The results of a study by Juchnowicz et al. also showed an association of FRAP, CAT and dityrosine levels with the severity of negative symptoms [38]. The association of negative symptoms with GSH levels shown in this study is consistent with the results of previous studies and the meta-analysis by Flatow et al. [ 7,9]. In the study by Matsuzawa et al. greater negative symptom severity was associated with lower GSH levels in the posterior medial frontal cortex of patients with schizophrenia [60]. Maes et al. also found a greater reduction in GSH levels in schizophrenia patients with predominantly negative symptoms [61]. Decreased levels of GSH and the efficiency of the antioxidant system along with increased sensitivity to OS may influence various pathophysiological processes found in schizophrenia, including the impairment of dopaminergic neurotransmission and NMDAR responses to glutamate [62]. GSH has a direct effect on glutamatergic neurotransmission through interaction with NMDARs, and NMDAR activity enhances and regulates GSH metabolism [63]. GSH levels increase in response to the glutamate-dependent excitatory activity of parvalbumin-induced GABAergic interneurons (PVIs) in the prefrontal cortex, leading to its downregulation and protecting neurons from OS. Decreased NMDAR activity contributes to GSH deficits and increased OS in the CNS, and in turn, even transient GSH deficiency leads to decreased NMDAR activity [64]. As a consequence, the inhibitory activity of PVIs as well as their number is reduced, which result in an excitatory–inhibitory imbalance in the CNS [63,65]. GSH deficiency also contributes to myelination abnormalities, which may have a significant impact on the deterioration of cognitive function in schizophrenia [3]. A negative correlation was also found between the activity of GPx, an enzyme belonging to the glutathione system, and the severity of cerebral atrophy in patients with chronic schizophrenia [66]. Impaired antioxidant activity and its influence on structural and functional changes in the CNS may explain the association between lower GSH levels and greater severity of negative symptoms. The study also found a negative effect of MDA and a positive effect of BDNF on the severity of deficit symptoms in schizophrenia. Elevated levels of MDA and a decreased proportion of polyunsaturated fatty acids were reported in patients with negative symptoms, suggesting that oxidative damage may implicate this clinical dimension [67]. In addition, the association of reduced BDNF levels with the severity of negative symptoms in schizophrenia is supported by some studies [68,69]. Others do not show this relationship [22,70]. In patients with bipolar affective disorder, a negative correlation was found between plasma BDNF and lipid peroxidation product levels, which may indicate that BDNF protects neurons from damage resulting from OS [71]. The neuroprotective role of BDNF is documented in many studies. The antiapoptotic effect is associated with the activation of the intracellular signaling cascade by the receptor of the tyrosine kinase B family (TrkB), towards which BDNF has high affinity [72]. In an in vitro study, BDNF was found to protect cortical neurons from NMDA- and H2O2-induced apoptosis by inhibiting the mitogen-activated kinase (MAPK) cascade [73]. The use of exogenous BDNF significantly inhibited the loss of dopaminergic neurons in the black matter caused by oxidative damage to cells [74]. ## 4.4. Biochemical Markers Associated with the Severity of Cognitive Impairment in Schizophrenia We assessed six domains of cognitive function: verbal memory, working memory, verbal fluency, motor speed, attention and speed of information processing and executive functions. The results of our study suggest that higher levels of cognitive impairment in schizophrenia may be associated with more intense protein oxidation as measured by AOPP levels, lower plasma antioxidant potential as measured by TAS and greater severity of depressive symptoms. The predictive value of TAS as a biomarker of cognitive impairment in schizophrenia has been shown. Martinez-Cengotitabengoa et al. found a positive correlation of the TAS level with total cognitive performance both at FEP and after 2 years of illness [75]. In the study population, the levels of all assessed cognitive functions were below the norms adopted for healthy individuals. The largest deficit was in WM, the dysfunction of which is considered one of the primary disorders of the schizophrenic process [76]. A significant effect of lower TAS and higher N-formKYN concentrations on greater severity of WM impairment was found. Martinez-Cengotitabengoa et al. also showed a correlation between TAS levels and working memory performance in a group of patients with non-affective psychosis both at FEP and after a follow-up of 2 years [75]. The shift of tryptophan metabolism towards KP associated with the production of neurotoxic metabolites may have a significant impact on WM impairment in schizophrenia. Many studies have documented the adverse effects of KP metabolites on cognitive function in schizophrenia. Kindler et al. showed that higher values of the KYN/tryptophan ratio were associated with reduced volumes of the dorsolateral prefrontal cortex, a region crucial for normal WM function, and more severe attention disorders and increased levels of proinflammatory cytokines [77]. Koola et al. showed that peripheral levels of KYN and KYNA can be an indicator of the degree of cognitive deterioration and a useful marker for monitoring treatment effects [54]. EF as well as ASP also show significant deterioration in schizophrenia [70,76,78]. These manifest as difficulties with stimulus selection and a tendency to process irrelevant information, which can lead to misinterpretation of percepts. Some authors indicated a positive correlation of ASP and EF performance with BDNF levels [79,80]. The severity of cognitive dysfunction in schizophrenia, including attention and EF, has also been found to be associated with immune activation and cytokine levels. Perkins et al. documented the correlation of proinflammatory IL-1, IL-7 and IL-8 levels with the severity of attention deficits [58]. In contrast, analyses using statistical models showed an association of EF disorders with interactions between BDNF and IL-8, BDNF and TNF-α, BDNF and SOD and BDNF and MDA [70,81]. This study did not find any associations between ASP or EF impairment severity and IL-6 levels, duration of illness, BDNF levels or other symptom severity. In our results, the levels of NO, ONOO− and AOPP correlated negatively with ASP and EF scores, but statistical modeling did not include these parameters as predictors of EF or ASP scores. Nevertheless, the associations we found may imply a negative effect of OS and NS on these cognitive domains. As AOPP is considered a marker and a mediator of the proinflammatory effect of OS, it may confirm an important role of inflammation in ASP and EF impairment in schizophrenia [45]. The deleterious effects of NO are mainly related to the highly reactive products of its metabolism, including ONOO−, which reacts with downstream molecules, leading to increased levels of RNS, ROS and increased lipid and protein oxidation. The negative effect of NO on assessed cognitive functions may be a consequence of increased glutamatergic excitotoxicity. NO is a mediator of NMDAR activation, and its concentration reflects glutamatergic transmission in the CNS [82]. Wang et al. found a negative correlation of NO levels with information processing speed, working memory performance and verbal learning [83]. The results of this study, like ours, indicated a negative effect of NO and related NS on cognitive function in schizophrenia. ## 4.5. Biochemical Markers Associated with Severity of Depressive Symptoms in Schizophrenia The study found that higher BDNF levels were associated with lower depressive symptoms in schizophrenia. BDNF’s role in the pathophysiology of depression is well known, and lower serum levels of the factor were found in people with major depressive disorders compared to HCs [84]. Some researchers pointed to a beneficial effect of BDNF on the severity of depressive symptoms in schizophrenia, which they link to the neuroprotective effect of this factor [85]. Interestingly, there are studies showing a negative correlation between BDNF levels and the severity of depressive symptoms in schizophrenia [86,87]. This is explained as the result of a compensatory increase in BDNF synthesis in response to oxidative stress and increased proinflammatory cytokine activity [86]. ## 4.6. Limitations When interpreting the results presented in this study, it is important to take into account the limitations. The study did not include an HC group. Therefore, it cannot be concluded whether the measured parameters’ levels significantly differed between schizophrenia patients and HC. The second limitation is a relatively small size of the study group. Moreover, it was comprised of hospitalized patients only, which may imply a greater severity of symptoms and poorer functioning than in the general population of schizophrenic patients. All study subjects were on antipsychotic treatment, taking different antipsychotics and the duration of treatment was variable. Previous studies are inconsistent regarding the effects of antipsychotics on oxidative stress markers. Some researchers postulated that typical antipsychotics increase oxidants and decrease antioxidants, while atypical antipsychotics cause the opposite effect [88,89]. On the other hand, it was shown that atypical antipsychotics may increase OS and decrease antioxidant activity [90,91]. However, studies of previously untreated individuals with FEP show that abnormalities in the biochemical factors studied appear early in the course of schizophrenia, which suggests that they may be a part of disease pathophysiology independently from pharmacotherapy [59]. Our study showed no significant differences in the measured parameters between the groups of patients taking different antipsychotics. However, an influence of pharmacotherapy cannot be excluded. The majority of the study group were smokers. Although no significant differences in study factors were found between the smoking and non-smoking patients, a significant effect of nicotinism on the parameters of inflammation and oxidation-reduction balance cannot be excluded. Studies have shown an association between chronic smoking and levels of some OS and NS parameters, including AGE and NO [92,93]. The physiological process of aging also involves changes in factors related to the oxidation-reduction balance. With age, there is a decrease in the activity of antioxidant enzymes and an increase in the oxidative potential and the concentration of lipid peroxidation products [53]. It has also been shown that age-related decreases in GSH levels are more strongly expressed in schizophrenia. The study did not show any correlation between the age of the patients and the parameters assessed. However, the influence of aging-related processes on the results and the correlations found also cannot be excluded. The study assessed only biochemical parameters from peripheral blood. Based on the available data, it was assumed that the concentration/activity of the examined factors in peripheral blood correlated with values in the CNS [94,95]. In order to confirm the correlations shown and the validity of the conclusions drawn from them, it would be worth extending the methodology to neuroimaging studies. Moreover, the measured biochemical parameters are not specific to schizophrenia and may be involved in the pathophysiology of various psychiatric and somatic diseases. To minimize the influence of other medical conditions, subjects with comorbid psychiatric disorders, somatic diseases and clinical or laboratory signs of inflammation were not included in the study. In order to increase the sensitivity and specificity of the correlations found, in addition to the analysis of associations between individual factors, statistical modeling methods were applied to the data. The levels of studied biochemical parameters may be influenced by factors that were not included in the study, such as diet or physical activity [46]. ## 5. Conclusions 1. Increased oxidative stress, protein oxidation and glycation processes and lipid peroxidation are associated with greater severity of psychopathological symptoms in schizophrenia. 2. Higher levels of oxidative stress together with higher levels of kynurenine are related to greater severity of positive symptoms in schizophrenia. More intense lipid peroxidation together with lower levels of GSH and BDNF correspond with more expressed negative symptoms. 3. Greater cognitive impairment severity in schizophrenia is associated with lower antioxidant potential along with increased protein oxidation and more severe depressive symptoms. Greater working memory impairment in schizophrenia is linked with lower antioxidant potential and higher N-formylkynurenine levels. ## References 1. Steeds H., Carhart-Harris R.L., Stone J.M.. Drug models of schizophrenia. Adv. Psychopharmacol. (2015) 5 43-58. DOI: 10.1177/2045125314557797 2. Insel T.R.. Rethinking schizophrenia. Nature (2010) 468 187-193. DOI: 10.1038/nature09552 3. Luptak M., Michalickova D., Fisar Z., Kitzlerova E., Hroudova J.. Novel approaches in schizophrenia-from risk factors and hypotheses to novel drug targets. World J. Psychiatry (2021) 11 277-296. DOI: 10.5498/wjp.v11.i7.277 4. Feigenson K.A., Kusnecov A.W., Silverstein S.M.. Inflammation and the two-hit hypothesis of schizophrenia. Neurosci. Biobehav. Rev. (2014) 38 72-93. DOI: 10.1016/j.neubiorev.2013.11.006 5. Martinez-Gras I., Garcia-Sanchez F., Guaza C., Rodriguez-Jimenez R., Andres-Esteban E., Palomo T., Rubio G., Borrell J.. Altered immune function in unaffected first-degree biological relatives of schizophrenia patients. Psychiatry Res. (2012) 200 1022-1025. DOI: 10.1016/j.psychres.2012.05.036 6. Miller B.J., Buckley P., Seabolt W., Mellor A., Kirkpatrick B.. Meta-analysis of cytokine alterations in schizophrenia: Clinical status and antipsychotic effects. Biol. Psychiatry (2011) 70 663-671. DOI: 10.1016/j.biopsych.2011.04.013 7. Flatow J., Buckley P., Miller B.J.. Meta-analysis of oxidative stress in schizophrenia. Biol. Psychiatry (2013) 74 400-409. DOI: 10.1016/j.biopsych.2013.03.018 8. Buosi P., Borghi F.A., Lopes A.M., da Silva Facincani I., Fernandes-Ferreira R., Ferreira Oliveira-Brancati C.I., Silva do Carmo T., Rossi Silva Souza D., da Silva D.G.H., de Almeida E.A.. Oxidative stress biomarkers in treatment-responsive and treatment-resistant schizophrenia patients. Trends Psychiatry Psychother. (2021) 43 278-285. DOI: 10.47626/2237-6089-2020-0078 9. Pazvantoglu O., Selek S., Okay I.T., Sengul C., Karabekiroglu K., Dilbaz N., Erel O.. Oxidative mechanisms in schizophrenia and their relationship with illness subtype and symptom profile. Psychiatry Clin. Neurosci. (2009) 63 693-700. DOI: 10.1111/j.1440-1819.2009.02015.x 10. Patel S., Sharma D., Kalia K., Tiwari V.. Crosstalk between endoplasmic reticulum stress and oxidative stress in schizophrenia: The dawn of new therapeutic approaches. Neurosci. Biobehav. Rev. (2017) 83 589-603. DOI: 10.1016/j.neubiorev.2017.08.025 11. Bitanihirwe B.K., Woo T.U.. Oxidative stress in schizophrenia: An integrated approach. Neurosci. Biobehav. Rev. (2011) 35 878-893. DOI: 10.1016/j.neubiorev.2010.10.008 12. Upthegrove R., Khandaker G.M.. Cytokines, Oxidative Stress and Cellular Markers of Inflammation in Schizophrenia. Curr. Top. Behav. Neurosci. (2020) 44 49-66. PMID: 31115797 13. Craddock R.M., Lockstone H.E., Rider D.A., Wayland M.T., Harris L.J., McKenna P.J., Bahn S.. Altered T-cell function in schizophrenia: A cellular model to investigate molecular disease mechanisms. PLoS ONE (2007) 2. DOI: 10.1371/journal.pone.0000692 14. Yao J.K., Reddy R., McElhinny L.G., van Kammen D.P.. Reduced status of plasma total antioxidant capacity in schizophrenia. Schizophr. Res. (1998) 32 1-8. DOI: 10.1016/S0920-9964(98)00030-9 15. Ustundag B., Atmaca M., Kirtas O., Selek S., Metin K., Tezcan E.. Total antioxidant response in patients with schizophrenia. Psychiatry Clin. Neurosci. (2006) 60 458-464. DOI: 10.1111/j.1440-1819.2006.01532.x 16. Virit O., Altindag A., Yumru M., Dalkilic A., Savas H.A., Selek S., Erel O., Herken H.. A defect in the antioxidant defense system in schizophrenia. Neuropsychobiology (2009) 60 87-93. DOI: 10.1159/000239684 17. Chiappelli J., Postolache T.T., Kochunov P., Rowland L.M., Wijtenburg S.A., Shukla D.K., Tagamets M., Du X., Savransky A., Lowry C.A.. Tryptophan Metabolism and White Matter Integrity in Schizophrenia. Neuropsychopharmacology (2016) 41 2587-2595. DOI: 10.1038/npp.2016.66 18. Cao B., Chen Y., Ren Z., Pan Z., McIntyre R.S., Wang D.. Dysregulation of kynurenine pathway and potential dynamic changes of kynurenine in schizophrenia: A systematic review and meta-analysis. Neurosci. Biobehav. Rev. (2021) 123 203-214. DOI: 10.1016/j.neubiorev.2021.01.018 19. Schwarcz R., Stone T.W.. The kynurenine pathway and the brain: Challenges, controversies and promises. Neuropharmacology (2017) 112 237-247. DOI: 10.1016/j.neuropharm.2016.08.003 20. Esquivel D.G., Ramirez-Ortega D., Pineda B., Castro N., Rios C., Perez de la Cruz V.. Kynurenine pathway metabolites and enzymes involved in redox reactions. Neuropharmacology (2017) 112 331-345. DOI: 10.1016/j.neuropharm.2016.03.013 21. Lang U.E., Hellweg R., Seifert F., Schubert F., Gallinat J.. Correlation between serum brain-derived neurotrophic factor level and an in vivo marker of cortical integrity. Biol. Psychiatry (2007) 62 530-535. DOI: 10.1016/j.biopsych.2007.01.002 22. Fernandes B.S., Steiner J., Berk M., Molendijk M.L., Gonzalez-Pinto A., Turck C.W., Nardin P., Gonçalves C.-A.. Peripheral brain-derived neurotrophic factor in schizophrenia and the role of antipsychotics: Meta-analysis and implications. Mol. Psychiatry (2015) 20 1108-1119. DOI: 10.1038/mp.2014.117 23. Belbasis L., Kohler C.A., Stefanis N., Stubbs B., van Os J., Vieta E., Seeman M.V., Arango C., Carvalho A.F., Evangelou E.. Risk factors and peripheral biomarkers for schizophrenia spectrum disorders: An umbrella review of meta-analyses. Acta Psychiatr. Scand. (2018) 137 88-97. DOI: 10.1111/acps.12847 24. Woods S.W.. Chlorpromazine equivalent doses for the newer atypical antipsychotics. J. Clin. Psychiatry (2003) 64 663-667. DOI: 10.4088/JCP.v64n0607 25. Kay S.R., Fiszbein A., Opler L.A.. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. (1987) 13 261-276. DOI: 10.1093/schbul/13.2.261 26. Keefe R.S., Goldberg T.E., Harvey P.D., Gold J.M., Poe M.P., Coughenour L.. The Brief Assessment of Cognition in Schizophrenia: Reliability, sensitivity, and comparison with a standard neurocognitive battery. Schizophr. Res. (2004) 68 283-297. DOI: 10.1016/j.schres.2003.09.011 27. Addington D., Addington J., Schissel B.. A depression rating scale for schizophrenics. Schizophr. Res. (1990) 3 241-251. DOI: 10.1016/0920-9964(90)90005-R 28. Aebi H.. Catalase in vitro. Methods Enzymol. (1984) 105 121-126. PMID: 6727660 29. Mansson-Rahemtulla B.B., Baldone D.C., Pruitt K.M., Rahemtulla F.. specific assays for peroxidases in human saliva. Arch Oral Biol. (1986) 31 661-668. DOI: 10.1016/0003-9969(86)90095-6 30. Moron M.D.J., Mannervik B.. Levels of glutathione, glutathione reductase and glutathione S-transferase activities in rat lung and liver. Biophys. Acta Gen. Subj. (1979) 582 67-78. DOI: 10.1016/0304-4165(79)90289-7 31. Erel O.. A novel automated direct measurement method for total antioxidant capacity using a new generation, more stable ABTS radical cation. Clin. Biochem. (2004) 37 277-285. DOI: 10.1016/j.clinbiochem.2003.11.015 32. Erel O.. A new automated colorimetric method for measuring total oxidant status. Clin. Biochem. (2005) 38 1103-1111. DOI: 10.1016/j.clinbiochem.2005.08.008 33. Kalousová M.S.J., Zima T.. Advanced glycation end-products and advanced oxidation protein products in patients with diabetes mellitus. Physiol. Res. (2002) 51 597-604. PMID: 12511184 34. Buege J.A., Aust S.D.. Microsomal lipid peroxidation. Methods Enzymol. (1978) 52 302-310. PMID: 672633 35. Borys J., Maciejczyk M., Kretowski A.J., Antonowicz B., Ratajczak-Wrona W., Jabłońska E., Załęski P., Waszkiel D., Ładny J.R., Żukowski P.. The Redox Balance in Erythrocytes, Plasma, and Periosteum of Patients with Titanium Fixation of the Jaw. Front. Physiol. (2017) 8 386. DOI: 10.3389/fphys.2017.00386 36. Grisham M.B.J., Johnson G.G., Lancaster J.R.. Quantitation of nitrate and nitrite in extracellular fluids. Methods Enzymol. (1996) 268 237-246. PMID: 8782590 37. Beckman J.S.I.H., Zhu L., van der Woerd M., Smith C., Chen J., Harrison J., Martin J.C., Tsai M.. Kinetics of superoxide dismutase- and iron-catalyzed nitration of phenolics by peroxynitrite. Arch. Biochem. Biophys. (1992) 298 438-445. DOI: 10.1016/0003-9861(92)90432-V 38. Juchnowicz D., Dzikowski M., Rog J., Waszkiewicz N., Zalewska A., Maciejczyk M., Karakuła-Juchnowicz H.. Oxidative Stress Biomarkers as a Predictor of Stage Illness and Clinical Course of Schizophrenia. Front. Psychiatry (2021) 12 728986. DOI: 10.3389/fpsyt.2021.728986 39. Albayrak Y., Unsal C., Beyazyuz M., Unal A., Kuloglu M.. Reduced total antioxidant level and increased oxidative stress in patients with deficit schizophrenia: A preliminary study. Prog. Neuropsychopharmacol. Biol. Psychiatry (2013) 45 144-149. DOI: 10.1016/j.pnpbp.2013.04.020 40. Copoglu U.S., Virit O., Kokacya M.H., Orkmez M., Bulbul F., Erbagci A.B., Semiz M., Alpak G., Unal A., Ari M.. Increased oxidative stress and oxidative DNA damage in non-remission schizophrenia patients. Psychiatry Res. (2015) 229 200-205. DOI: 10.1016/j.psychres.2015.07.036 41. Witko-Sarsat V.F.M., Nguyen Khoa T., Capeillère-Blandin C., Nguyen A.T., Canteloup S., Dayer J.M., Jungers P., Drüeke T., Descamps-Latscha B.. Advanced oxidation protein products as novel mediators of inflammation and monocyte activation in chronic renal failure. J. Immunol. (1998) 161 2524-2532. DOI: 10.4049/jimmunol.161.5.2524 42. Kloska M.M.-W.D., Człapka-Matyasik M., Dobrowolska A., Grzymisławski M.. Oxidative stress in the etiopathogenesis of inflammatory bowel diseases. Proc. Biochem. (2020) 66 143-150 43. Zhang X.Y., Zhou D.F., Qi L.Y., Chen S., Cao L.Y., Chen D.C., Xiu M.H., Wang F., Wu G.Y., Lu L.. Superoxide dismutase and cytokines in chronic patients with schizophrenia: Association with psychopathology and response to antipsychotics. Psychopharmacology (2009) 204 177-184. DOI: 10.1007/s00213-008-1447-6 44. Goldsmith D.R., Haroon E., Miller A.H., Strauss G.P., Buckley P.F., Miller B.J.. TNF-alpha and IL-6 are associated with the deficit syndrome and negative symptoms in patients with chronic schizophrenia. Schizophr. Res. (2018) 199 281-284. DOI: 10.1016/j.schres.2018.02.048 45. Guidara W., Messedi M., Naifar M., Maalej M., Grayaa S., Omri S., Thabet J.B., Maalej M., Charfi N., Ayadi F.. Predictive value of oxidative stress biomarkers in drug-free patients with schizophrenia and schizo-affective disorder. Psychiatry Res. (2020) 293 113467. DOI: 10.1016/j.psychres.2020.113467 46. Juchnowicz D., Dzikowski M., Rog J., Waszkiewicz N., Karakuła K.H., Zalewska A., Maciejczyk M., Karakula-Juchnowicz H.. Pro/Antioxidant State as a Potential Biomarker of Schizophrenia. J. Clin. Med. (2021) 10. DOI: 10.3390/jcm10184156 47. Ramasamy R., Vannucci S.J., Yan S.S., Herold K., Yan S.F., Schmidt A.M.. Advanced glycation end products and RAGE: A common thread in aging, diabetes, neurodegeneration, and inflammation. Glycobiology. (2005) 15 16R-28R. DOI: 10.1093/glycob/cwi053 48. Kouidrat Y., Amad A., Arai M., Miyashita M., Lalau J.D., Loas G., Itokawa M.. Advanced glycation end products and schizophrenia: A systematic review. J. Psychiatr. Res. (2005) 66–67 112-117. DOI: 10.1016/j.jpsychires.2015.04.023 49. Zhang M., Zhao Z., He L., Wan C.. A meta-analysis of oxidative stress markers in schizophrenia. Sci. China Life Sci. (2010) 53 112-124. DOI: 10.1007/s11427-010-0013-8 50. Gubert C., Stertz L., Pfaffenseller B., Panizzutti B.S., Rezin G.T., Massuda R., Streck E.L., Gama C.S., Kapczinski F., Kunz M.. Mitochondrial activity and oxidative stress markers in peripheral blood mononuclear cells of patients with bipolar disorder, schizophrenia, and healthy subjects. J. Psychiatr. Res. (2013) 47 1396-1402. DOI: 10.1016/j.jpsychires.2013.06.018 51. Arvindakshan M., Sitasawad S., Debsikdar V., Ghate M., Evans D., Horrobin D.F., Bennett C., Ranjekar P.K., Mahadik S.P.. Essential polyunsaturated fatty acid and lipid peroxide levels in never-medicated and medicated schizophrenia patients. Biol. Psychiatry (2003) 53 56-64. DOI: 10.1016/S0006-3223(02)01443-9 52. Dudzińska E., Szymona K., Bogucki J., Koch W., Cholewińska E., Sitarz R., Ognik K.. Increased Markers of Oxidative Stress and Positive Correlation Low-Grade Inflammation with Positive Symptoms in the First Episode of Schizophrenia in Drug-Naïve Patients. J. Clin. Med. (2022) 11. DOI: 10.3390/jcm11092551 53. Całyniuk B.G.-N.E., Walkiewicz K., Kawecka S., Popiołek E., Fatyga E.. Malondialdehyde (MDA)—Product of lipid peroxidation as marker of homeostasis disorders and aging. Ann. Acad. Med. Siles. (2016) 70 224-228. DOI: 10.18794/aams/65697 54. Koola M.M.. Kynurenine pathway and cognitive impairments in schizophrenia: Pharmacogenetics of galantamine and memantine. Schizophr. Res. Cogn. (2016) 4 4-9. DOI: 10.1016/j.scog.2016.02.001 55. Wu Z., Zhang X.Y., Wang H., Tang W., Xia Y., Zhang F., Liu J., Fu Y., Hu J., Chen Y.. Elevated plasma superoxide dismutase in first-episode and drug naive patients with schizophrenia: Inverse association with positive symptoms. Prog. Neuropsychopharmacol. Biol. Psychiatry (2012) 36 34-38. DOI: 10.1016/j.pnpbp.2011.08.018 56. Al-Chalabi B.M., Thanoon I.A., Ahmed F.A.. Potential effect of olanzapine on total antioxidant status and lipid peroxidation in schizophrenic patients. Neuropsychobiology (2009) 59 8-11. DOI: 10.1159/000202823 57. Dietrich-Muszalska A., Kontek B.. Lipid peroxidation in patients with schizophrenia. Psychiatry Clin. Neurosci. (2010) 64 469-475. DOI: 10.1111/j.1440-1819.2010.02132.x 58. Perkins D.O., Jeffries C.D., Addington J., Bearden C.E., Cadenhead K.S., Cannon T.D., Cornblatt B.A., Mathalon D.H., McGlashan T.H., Seidman L.J.. Towards a psychosis risk blood diagnostic for persons experiencing high-risk symptoms: Preliminary results from the NAPLS project. Schizophr. Bull. (2015) 41 419-428. DOI: 10.1093/schbul/sbu099 59. Li X.F., Zheng Y.L., Xiu M.H., Chen D.C., Kosten T.R., Zhang X.Y.. Reduced plasma total antioxidant status in first-episode drug-naive patients with schizophrenia. Prog. Neuropsychopharmacol. Biol. Psychiatry (2011) 35 1064-1067. DOI: 10.1016/j.pnpbp.2011.03.001 60. Matsuzawa D., Obata T., Shirayama Y., Nonaka H., Kanazawa Y., Yoshitome E., Takanashi J., Matsuda T., Shimizu E., Ikehira H.. Negative correlation between brain glutathione level and negative symptoms in schizophrenia: A 3T 1H-MRS study. PLoS ONE (2008) 3. DOI: 10.1371/journal.pone.0001944 61. Maes M., Sirivichayakul S., Matsumoto A.K., Michelin A.P., de Oliveira Semeão L., de Lima Pedrão J.V., Moreira E.G., Barbosa D.S., Carvalho A.F., Solmi M.. Lowered Antioxidant Defenses and Increased Oxidative Toxicity Are Hallmarks of Deficit Schizophrenia: A Nomothetic Network Psychiatry Approach. Mol. Neurob. (2020) 57 4578-4597. DOI: 10.1007/s12035-020-02047-5 62. Steullet P., Cabungcal J.H., Monin A., Dwir D., O’Donnell P., Cuenod M., Do K.Q.. Redox dysregulation, neuroinflammation, and NMDA receptor hypofunction: A "central hub" in schizophrenia pathophysiology?. Schizophr. Res. (2016) 176 41-51. DOI: 10.1016/j.schres.2014.06.021 63. Barron H., Hafizi S., Andreazza A.C., Mizrahi R.. Neuroinflammation and Oxidative Stress in Psychosis and Psychosis Risk. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18030651 64. Hardingham G.E., Do K.Q.. Linking early-life NMDAR hypofunction and oxidative stress in schizophrenia pathogenesis. Nat. Rev. Neurosci. (2016) 17 125-134. DOI: 10.1038/nrn.2015.19 65. Ermakov E.A., Dmitrieva E.M., Parshukova D.A., Kazantseva D.V., Vasilieva A.R., Smirnova L.P.. Oxidative Stress-Related Mechanisms in Schizophrenia Pathogenesis and New Treatment Perspectives. Oxid. Med. Cell. Longev. (2021) 2021 8881770. DOI: 10.1155/2021/8881770 66. Goh X.X., Tang P.Y., Tee S.F.. Blood-based oxidation markers in medicated and unmedicated schizophrenia patients: A meta-analysis. Asian J. Psychiatr. (2022) 67 102932. DOI: 10.1016/j.ajp.2021.102932 67. Bošković M., Vovk T., Kores Plesničar B., Grabnar I.. Oxidative stress in schizophrenia. Curr. Neuropharmacol. (2011) 9 301-312. PMID: 22131939 68. Rizos E.R.I., Laskos E., Arsenis G., Michalopoulou P., Vasilopoulos D., Gournellis R., Lykouras L.. Decreased serum levels of BDNF in first episode patients with schizophrenia and its psychopathological correlations. Abstr. Schizophr. Res. (2008) 102 1-3. DOI: 10.1016/S0920-9964(08)70630-3 69. Zhang X.Y., Tan Y.L., Zhou D.F., Cao L.Y., Wu G.Y., Xu Q., Shen Y., Haile C.N., Kosten T.A., Kosten T.R.. Serum BDNF levels and weight gain in schizophrenic patients on long-term treatment with antipsychotics. J. Psychiatr. Res. (2007) 41 997-1004. DOI: 10.1016/j.jpsychires.2006.08.007 70. Wei C., Sun Y., Chen N., Chen S., Xiu M., Zhang X.. Interaction of oxidative stress and BDNF on executive dysfunction in patients with chronic schizophrenia. Psychoneuroendocrinology (2020) 111 104473. DOI: 10.1016/j.psyneuen.2019.104473 71. Newton D.F., Naiberg M.R., Andreazza A.C., Scola G., Dickstein D.P., Goldstein B.I.. Association of Lipid Peroxidation and Brain-Derived Neurotrophic Factor with Executive Function in Adolescent Bipolar Disorder. Psychopharmacology (Berl.) (2017) 234 647-656. DOI: 10.1007/s00213-016-4500-x 72. Numakawa T., Suzuki S., Kumamaru E., Adachi N., Richards M., Kunugi H.. BDNF function and intracellular signaling in neurons. Histol. Histopathol. (2010) 25 237-258. PMID: 20017110 73. Boutahar N., Reynaud E., Lassabliere F., Borg J.. Brain-derived neurotrophic factor inhibits cell cycle reentry but not endoplasmic reticulum stress in cultured neurons following oxidative or excitotoxic stress. J. Neurosci. Res. (2010) 88 2263-2271. DOI: 10.1002/jnr.22384 74. Altar C.A., Boylan C.B., Fritsche M., Jones B.E., Jackson C., Wiegand S.J., Lindsay R.M., Hyman C.. Efficacy of brain-derived neurotrophic factor and neurotrophin-3 on neurochemical and behavioral deficits associated with partial nigrostriatal dopamine lesions. J. Neurochem. (1994) 63 1021-1032. DOI: 10.1046/j.1471-4159.1994.63031021.x 75. Martínez-Cengotitabengoa M., Micó J.A., Arango C., Castro-Fornieles J., Graell M., Payá B., Leza J.C., Zorrilla I., Parellada M., López M.P.. Basal low antioxidant capacity correlates with cognitive deficits in early onset psychosis. A 2-year follow-up study. Schizophr. Res. (2014) 156 23-29. DOI: 10.1016/j.schres.2014.03.025 76. Moustafa A.A., Garami J.K., Mahlberg J., Golembieski J., Keri S., Misiak B., Frydecka D.. Cognitive function in schizophrenia: Conflicting findings and future directions. Rev. Neurosci. (2016) 27 435-448. DOI: 10.1515/revneuro-2015-0060 77. Kindler J., Lim C.K., Weickert C.S., Boerrigter D., Galletly C., Liu D., Jacobs K.R., Balzan R., Bruggemann J., O’Donnell M.. Dysregulation of kynurenine metabolism is related to proinflammatory cytokines, attention, and prefrontal cortex volume in schizophrenia. Mol. Psychiatry (2020) 25 2860-2872. DOI: 10.1038/s41380-019-0401-9 78. Keedy S.K., Reilly J.L., Bishop J.R., Weiden P.J., Sweeney J.A.. Impact of antipsychotic treatment on attention and motor learning systems in first-episode schizophrenia. Schizophr. Bull. (2015) 41 355-365. DOI: 10.1093/schbul/sbu071 79. Hori H., Yoshimura R., Katsuki A., Atake K., Igata R., Konishi Y., Beppu H., Tominaga H.. Blood Biomarkers Predict the Cognitive Effects of Aripiprazole in Patients with Acute Schizophrenia. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18030568 80. Yang Y., Liu Y., Wang G., Hei G., Wang X., Li R., Li L., Wu R., Zhao J.. Brain-derived neurotrophic factor is associated with cognitive impairments in first-episode and chronic schizophrenia. Psychiatry Res. (2019) 273 528-536. DOI: 10.1016/j.psychres.2019.01.051 81. Xiu M.H., Wang D.M., Du X.D., Chen N., Tan S.P., Tan Y.L., Yang F., Cho R.Y., Zhang X.Y.. Interaction of BDNF and cytokines in executive dysfunction in patients with chronic schizophrenia. Psychoneuroendocrinology (2019) 108 110-117. DOI: 10.1016/j.psyneuen.2019.06.006 82. Akyol O.Z., Zoroglu S.S., Armutcu F., Sahin S., Gurel A.. Nitric oxide as a physiopathological factor in neuropsychiatric disorders. In Vivo (2004) 18 377-390. PMID: 15341194 83. Wang Y.P., Zhang P.F., Yuan X.X., Liu Y.F., Li H.H., Tao Q., Li X., Pang L.J., Song X.Q.. Effects of oxidative stress on cognitive impairment in first episode schizophrenia. Zhonghua Yi Xue Za Zhi (2019) 99 9-13. PMID: 30641657 84. Eker S.S., Yavasci E.O., Cangur S., Kirli S., Sarandol E.. Can BDNF and IL-2 be indicators for the diagnosis in schizophrenic patients with depressive symptoms?. Acta Neuropsychiatr. (2014) 26 291-297. DOI: 10.1017/neu.2014.13 85. Wysokinski A.. Serum levels of brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3) in depressed patients with schizophrenia. Nord. J. Psychiatry (2016) 70 267-271. DOI: 10.3109/08039488.2015.1087592 86. Noto C.S., Gadelha A., Belangero S.I., Smith M.A., de Aguiar B.W., Panizzuti B., Mari Jde J., Gama C.S., Bressan R.A., Brietzke E.. Association of biomarkers and depressive symptoms in schizophrenia. Neurosci. Lett. (2011) 505 282-285. DOI: 10.1016/j.neulet.2011.10.042 87. Zhang X.Y., Chen D.C., Tan Y.L., Tan S.P., Wang Z.R., Yang F.D., Okusaga O.O., Zunta-Soares G.B., Soares J.C.. The interplay between BDNF and oxidative stress in chronic schizophrenia. Psychoneuroendocrinology (2015) 51 201-208. DOI: 10.1016/j.psyneuen.2014.09.029 88. Kropp S., Kern V., Lange K., Degner D., Hajak G., Kornhuber J., Rüther E., Emrich H.M., Schneider U., Bleich S.. Oxidative stress during treatment with first- and second-generation antipsychotics. J. Neuropsychiatry Clin. Neurosci. (2005) 17 227-231. DOI: 10.1176/jnp.17.2.227 89. Singh O.P., Chakraborty I., Dasgupta A., Datta S.. A comparative study of oxidative stress and interrelationship of important antioxidants in haloperidol and olanzapine treated patients suffering from schizophrenia. Indian J. Psychiatry (2008) 50 171-176. DOI: 10.4103/0019-5545.43627 90. Pillai A., Parikh V., Terry A.V., Mahadik S.P.. Long-term antipsychotic treatments and crossover studies in rats: Differential effects of typical and atypical agents on the expression of antioxidant enzymes and membrane lipid peroxidation in rat brain. J. Psychiatr. Res. (2007) 41 372-386. DOI: 10.1016/j.jpsychires.2006.01.011 91. Zhang X.Y., Zhou D.F., Cao L.Y., Zhang P.Y., Wu G.Y., Shen Y.C.. The effect of risperidone treatment on superoxide dismutase in schizophrenia. J. Clin. Psychopharmacol. (2003) 23 128-131. DOI: 10.1097/00004714-200304000-00004 92. Nasyrova R.F., Ivashchenko D.V., Ivanov M.V., Neznanov N.G.. Role of nitric oxide and related molecules in schizophrenia pathogenesis: Biochemical, genetic and clinical aspects. Front. Physiol. (2015) 6 139. DOI: 10.3389/fphys.2015.00139 93. Cerami C., Founds H., Nicholl I., Mitsuhashi T., Giordano D., Vanpatten S., Lee A., Al-Abed Y., Vlassara H., Bucala R.. Tobacco smoke is a source of toxic reactive glycation products. Proc. Natl. Acad. Sci. USA (1997) 94 13915-13920. DOI: 10.1073/pnas.94.25.13915 94. Dantzer R., Connor J.C., Freund G.G., Johnson R.W., Kelley K.W.. from inflammation to sickness and depression: When the immune system subjugates the brain. Nat. Rev. Neurosci. (2008) 9 46-56. DOI: 10.1038/nrn2297 95. Maciejczyk M., Zalewska A., Ladny J.R.. Salivary Antioxidant Barrier, Redox Status, and Oxidative Damage to Proteins and Lipids in Healthy Children, Adults, and the Elderly. Oxid. Med. Cell. Longev. (2019) 2019 4393460. DOI: 10.1155/2019/4393460
--- title: Subclinical Carotid Atherosclerosis in Patients with Rheumatoid Arthritis at Low Cardiovascular Risk authors: - Elena V. Gerasimova - Tatiana V. Popkova - Daria A. Gerasimova - Yuliya V. Markina - Tatiana V. Kirichenko journal: Biomedicines year: 2023 pmcid: PMC10046543 doi: 10.3390/biomedicines11030974 license: CC BY 4.0 --- # Subclinical Carotid Atherosclerosis in Patients with Rheumatoid Arthritis at Low Cardiovascular Risk ## Abstract Objective: To evaluate the rate of subclinical carotid atherosclerosis and clinical significance of immunoinflammatory markers in patients with rheumatoid arthritis (RA) at low cardiovascular risk. Materials and Methods: The study included 275 RA patients and a control group of 100 participants without autoimmune diseases. All study participants were at low cardiovascular risk, calculated by the QRISK3 scale (<$20\%$), and free of cardiovascular disease. Ultrasound examination of carotid arteries was performed to measure cIMT and to detect atherosclerotic plaques (ASP) in carotid arteries. sIСАМ-1, sVСАМ, and sCD40L levels were determined by enzyme immunoassay. Results: Carotid ASP was observed more frequently in RA patients ($27\%$) than in the control group ($17\%$), $$p \leq 0.03.$$ The frequency of ASP in RA patients did not depend on the disease’s stage or activity. There was a significant correlation between cIMT and age, cardiovascular risk determined by QRISK3, level of total cholesterol, LDL, and blood pressure in RA patients, $p \leq 0.05$ in all cases. No correlation between cIMT and blood levels of sCD40L, sVCAM, and sICAM was found. In RA patients, a higher concentration of sVCAM was detected in the carotid ASP group compared to the non-atherosclerotic group. sCD40L was associated with cIMT and total cholesterol in the ASP group and with total cholesterol and blood pressure in non-atherosclerotic patients. Conclusions: Subclinical atherosclerotic lesions of the carotid arteries were observed significantly more frequently in RA patients with low cardiovascular risk than in the control group. The results of the study demonstrate the association between cIMT, traditional cardiovascular risk factors, and immunoinflammatory markers in RA patients. ## 1. Introduction Cardiovascular disease (CVD) is the most prevalent and socially significant comorbidity and one of the main causes of premature mortality in rheumatoid arthritis (RA) [1,2,3,4]. Currently, RA is considered an independent cardiovascular risk factor that increases the risk of developing atherosclerotic CVD by about $50\%$, including in patients with subclinical or early-stage RA [5]. Timely diagnostics based on risk assessment is an extremely important aspect of the prevention of cardiovascular complications in patients with RA [6,7]. At the same time, the stratification of cardiovascular risk in RA is still a serious problem. Insufficient predictive value of traditional cardiovascular risk factors and risk scales, as well as the high frequency of subclinical atherosclerosis, leads to an underestimation of cardiovascular risk in patients with RA [6,8,9]. The use of ultrasound measurement of the thickness of the intima-media complex of carotid arteries (cIMT) in clinical practice has improved the stratification of cardiovascular risk significantly [10,11]. The accelerated development of atherosclerosis in RA is associated with both traditional cardiovascular risk factors and chronic inflammatory status, which determines the production of various inflammatory mediators. Despite the fact that the molecular mechanisms underlying relationships between RA and the development of the atherosclerotic plaques are not well understood, the pathogenesis of atherosclerosis progression in patients with autoimmune diseases is being actively studied for a better understanding of the inflammation–atherosclerosis axis in this group of patients, since it is one of the most actual questions of modern rheumatology [12]. It is known that the progression of the immunopathological process in RA is accompanied by the development of a wide range of immunoinflammatory reactions, which may have pathophysiological significance in the development of atherosclerosis. The imbalance of cytokines and the accumulation of inflammatory mediators contribute to the development of vascular disorders associated with atherogenesis—endothelial dysfunction, vasoconstriction, lipid and lipoprotein peroxidation, hypercoagulation—and later lead to the formation and destabilization of atherosclerotic plaques and the development of cardiovascular complications [13,14]. It was shown that in the assessment of early atherosclerotic vascular changes, the prognostic significance of soluble CD40 ligand (sCD40L) and adhesion molecules—intercellular adhesion molecule-1 (sICAM-1) and vascular cell adhesion molecule-1 (sVCAM-1)—was significantly higher compared to other biological mediators [15,16]. The crucial role of sCD40L and adhesion molecules in pathogenetic mechanisms of atherosclerosis development has been established in numerous studies [17,18]. Activation of the sCD40L is associated with the expression of adhesion molecules and pro-inflammatory cytokines by key cells of the vascular wall leading to the recruitment of immune cells to the arterial endothelium that is considered a main step in the pathogenesis of atherosclerosis. At the same time, it was demonstrated that RA patients have significantly increased levels of sCD40L, ICAM-1, and VCAM-1 than healthy subjects [19,20]. In this regard, these molecules may be considered important mediators in the development of atherosclerotic lesions in rheumatic patients. The aim of the present study was to examine the traditional cardiovascular risk factors, blood levels of sCD40L, ICAM-1, and VCAM-1, ultrasound measurements of carotid atherosclerosis, and to reveal their association in RA patients at low cardiovascular risk compared to healthy controls. ## 2. Materials and Methods The study was performed in accordance with the Declaration of Helsinki and approved by the local ethics committee. Informed consent to participate in the study was obtained from each participant. The study included 275 RA patients with a low cardiovascular risk free of CVD and 100 participants without autoimmune diseases of the control group matched by sex, age, traditional cardiovascular risk factors, and rate of cardiovascular risk. RA was diagnosed according to the American College of Rheumatology/European Alliance of Associations for Rheumatology (ACR/EULAR) 2010 criteria [21]. The mean age of study participants was 51 [47; 54] years old, $88\%$ of patients were female, and the RA duration was 124 [34; 225] months. The early stage of RA was determined in 38 ($14\%$) patients, advanced stage in 182 ($66\%$) patients, and late stage in 55 ($20\%$) patients. RA activity was evaluated using several scores: disease activity score 28 (DAS28), Clinical Disease Activity Index (CDAI), and Health assessment questionnaires (HAQ). Most patients ($69\%$) had moderate disease activity: DAS28—4.9 [3.7; 5.9] points, CDAI—18 [12; 29] points, and HAQ—1.25 [0.875; 1.75] points; $81\%$ of patients were seropositive for the rheumatoid factor (RF) and $77\%$ for the antibodies against cyclic citrullinated peptides (ACCP). When included in the study, $64\%$ of RA patients received methotrexate (mean dose 20 [15; 27] mg/week, the average duration 110 [21; 174] months; cumulative doses 8,8 [1,7; 13,9]g), $15\%$—leflunomide (20 mg/day), $12\%$—sulfasalazine (2000 mg/day), $42\%$—glucocorticoids (mean dose 4 [2; 6] mg/day, the average duration 75 [15; 175] months; cumulative doses 9,1 [1,8; 21,0] g), and $34\%$—non-steroid anti-inflammatory preparations. RA patients and participants of the control group have never been treated with statins. Clinical characteristics of patients with RA are presented in Table 1. The control group included 100 participants without autoimmune diseases with low cardiovascular risk, comparable in sex (female/male $\frac{88}{12}$) and age (mean age 47 [45; 51] years). The frequency of traditional cardiovascular risk factors in RA patients and participants of a control group was comparable as well: arterial hypertension in $48\%$ and $42\%$, dyslipidemia in $62\%$ and $58\%$, overweight in $45\%$ and $37\%$, family history for CVD in $43\%$ and $40\%$, and smoking in $25\%$ and $23\%$ of RA patients and controls, respectively. The QRISK calculator (QResearch Cardiovascular disease Risk Algorithm, QRISK®3-2018), developed for predicting the 10-year risk of coronary heart disease development and cardiovascular events in general medical practice, was used to calculate cardiovascular risk [22]. The QRISK3 algorithm includes almost all traditional cardiovascular risk factors and several additional factors (rheumatoid arthritis, systemic lupus erythematosus, diabetes mellitus, chronic kidney disease, ethnicity). The ultrasound examination of the carotid arteries was carried out on the ultrasound system Esaote MyLab Twice (Esaote S.p. A, Genoa, Italy). Atherosclerotic lesions of the carotid arteries were assessed by the detection of atherosclerotic plaques determined as a local increase of intima-media thickness ≥ 1.5 mm. The examination included scanning of the left and right common carotid arteries, the carotid bifurcation area, as well as external and internal carotid arteries with a focus on the far wall of the artery in three fixed projections—anterior, lateral, and posterior. Measurement of cIMT was performed at the far wall of the common carotid artery 10 mm opposite the top of the carotid bifurcation. cIMT was measured as the distance from the leading edge of the first echogenic area to the leading edge of the second echogenic area [23]. One researcher was responsible for all cIMT measurements throughout the study. The whole procedure was recorded on a digital scan medium for subsequent analysis using the dedicated software package M’Ath (Metris, SRL, Villers-Bretonneux, France). The average of three measurements (in the anterior, posterior, and lateral projections) was considered an integral indicator of cIMT. The concentrations of total cholesterol, high-density lipoproteins (HDL), and triglycerides (TG) were determined by standard enzymatic methods, and the level of low-density lipoproteins (LDL) was calculated using the Friedwald formula: LDL-C = cholesterol − TG/5 − HDL cholesterol. The levels of C-reactive protein (CRP) and IgM-RF in blood serum were measured by the immunonephelometric method using a BN Pro Spec analyzer (Siemens, Munich, Germany). The concentration of sVCAM, sICAM, and sCD40L in venous blood serum was determined by enzyme immunoassay using reagent kits and according to Bender MedSystems (Burlingame, CA, USA) protocols. Statistical data processing was carried out using Statistica 12 and SPSS 14.0 software. Results are presented as the median and interquartile range (Me [25th; 75th percentile]). The χ2 test was used to compare the frequencies of qualitative data in groups. The groups were compared using the nonparametric Mann–Whitney test; correlation analysis was performed by Spearman’s rank correlation method. Differences were considered significant at $p \leq 0.05.$ ## 3. Results Atherosclerotic plaques of the carotid arteries were observed more frequently in RA patients ($27\%$) with low cardiovascular risk than in the control group ($17\%$), $$p \leq 0.03.$$ Carotid atherosclerosis was detected in $50\%$ of men and $24\%$ of women, $p \leq 0.01.$ The frequency of carotid atherosclerotic plaques in RA patients with low cardiovascular risk did not depend on the stage and activity of the disease. In RA patients, a relationship between cIMT and age ($R = 0.48$), the value of cardiovascular risk assessed by QRISK3 scale ($R = 0.36$), the level of total cholesterol ($R = 0.28$), LDL ($R = 0.18$), systolic blood pressure (SBP) ($R = 0.37$), and diastolic blood pressure (DBP) ($R = 0.38$) was found, $p \leq 0.05$ in all cases. In the control group, a similar relationship with age ($R = 0.41$), concentrations of total cholesterol ($R = 0.22$), LDL ($R = 0.26$), levels of SBP ($R = 0.18$), and DBP ($R = 0.23$) was determined, $p \leq 0.05$ in all cases. Table 2 presents the comparison of study participants in basic traditional cardiovascular risk factors and carotid atherosclerosis parameters, as well as blood levels of adhesion molecules (sIСАМ-1, sVСАМ) and sCD40L. When comparing the blood levels of sCD40L, sVCAM, and sICAM of RA patients and control group participants, a higher concentration of sVCAM and a tendency to increase the blood level of sICAM in RA patients were revealed in comparison with the control group. No differences in blood levels of sCD40L were determined between groups. Correlation analysis did not reveal the association between the cIMT and blood concentrations of sICAM-1, sVCAM, and sCD40L both in patients with RA and in the control group. To search for risk factors for the development of carotid atherosclerotic plaques in RA patients with low cardiovascular risk, the method of multivariate logistic regression analysis was used with the calculation of OR and $95\%$ CI. The sensitivity of the model was $51.5\%$, the specificity was $96.2\%$, and the coefficient of multiple determination R2 was 0.396. The factors associated with the risk of carotid atherosclerosis are presented in Table 3. Thus, in RA patients with low cardiovascular risk, the presence of arterial hypertension increased the risk of developing carotid atherosclerotic plaques by two times and dyslipidemia by almost three times. For further analysis, patients with RA and low cardiovascular risk were divided into two groups depending on the presence of atherosclerotic plaques in carotid arteries. The first group consisted of 74 RA patients with carotid atherosclerosis; the second group included 201 RA patients without atherosclerosis. There were more men ($21\%$) and patients with dyslipidemia ($88\%$) in the atherosclerotic group of study participants in comparison with the non-atherosclerotic group ($p \leq 0.05$). It was found that RA patients with carotid atherosclerotic plaques were older and had higher body mass index (BMI), rate of cardiovascular risk calculated by the QRISK3 scale, as well as serum levels of cholesterol, LDL, triglycerides, and sVCAM concentrations compared to patients without carotid atherosclerotic plaques (Table 4). Then, we analyzed the correlation between immunoinflammatory markers and traditional cardiovascular risk factors in both groups. It was found that serum level of sCD40L correlated significantly with cIMT ($R = 0.40$, $$p \leq 0.04$$) and total cholesterol blood level ($R = 0.38$, $$p \leq 0.01$$) in a group of patients with atherosclerosis and with total cholesterol levels ($R = 0.41$, $$p \leq 0.02$$) and DBP ($R = 0.3$, $$p \leq 0.04$$) in the non-atherosclerosis group. It was also detected in patients without atherosclerosis that serum levels of sVCAM were associated with RF ($R = 0.52$, $$p \leq 0.01$$), while serum levels of sICAM were associated with DAS28 ($R = 0.44$, $$p \leq 0.04$$). ## 4. Discussion Stratification of cardiovascular risk in patients with RA remains a difficult problem, especially in cases with low cardiovascular risk. Cardiovascular risk calculated with the use of special scores does not always reflect the real risk of CVD in patients with RA [24,25]. The risk for CVD is 1.5 times higher in RA patients, and 10-year CVD risk scores underestimate the cardiovascular risk of rheumatic patients [24]. The choice of the QRISK3 calculator for risk calculation was based on its advantage in predicting the risk of CVD in patients with autoimmune rheumatic diseases [26,27,28]. Therefore, in the work of A. Corrales et al. [ 27], the best predictors of CVD and mortality in patients with RA, adjusted for age, sex, and disease duration, were the QRISK3 risk score and the detection of carotid atherosclerotic plaque. Given the low cardiovascular risk in RA patients, the role of surrogate markers increases, which primarily includes the visualization of subclinical atherosclerotic vascular lesions [27,29]. The high prevalence of subclinical atherosclerosis in patients with RA compared with subjects without autoimmune diseases is observed already in the early stages of RA and has a similar prevalence in patients with early and late stages of the disease [30,31,32]. In our previous study, carotid atherosclerosis (atherosclerotic plaques in carotid arteries) was diagnosed in $55.4\%$ of patients with early untreated RA mean aged 56 years old, as well as increased mean cIMT over 0.9 mm was detected in $51.4\%$ of RA patients [33]. According to other authors, subclinical carotid atherosclerosis was found in 21–$46\%$ of patients with RA [34,35,36]. The progression of atherosclerosis in carotid arteries was observed 3–5-fold more in patients with RA than in individuals without autoimmune diseases [35,36]. In the current study, carotid atherosclerosis was detected in $27\%$ of RA patients with low cardiovascular risk. The data obtained are consistent with the work of S. Hannawi et al. [ 35], who demonstrated that manifestations of asymptomatic carotid atherosclerosis were observed more often in RA patients with low cardiovascular risk than in subjects of a comparable control group ($21\%$ vs. $4\%$, $$p \leq 0.01$$). The association of immune disorder due to autoimmune processes and traditional cardiovascular risk factors, primarily dyslipidemia and hypertension, play an important role in the accelerated development of atherosclerosis in RA. The current study demonstrates the dependence of the cIMT on traditional risk factors for CVD, such as gender, age, serum levels of lipids, and blood pressure in patients with RA, which matches the results of other studies [33,35,36]. It has been shown that chronic systemic inflammation plays an important role in accelerated atherogenesis in autoimmune diseases. Inflammation can have a direct impact on the development of atherosclerosis or indirectly enhance the effect of traditional cardiovascular risk factors [14,31,32]. One of the mechanisms of the atherosclerotic process connecting inflammation and atherothrombosis is the activation of the CD40/CD40L signaling system. CD40L is a transmembrane glycoprotein belonging to the tumor necrosis factor (TNF) family. CD40 and CD40L are expressed by various cells, including the cells of atherosclerotic plaque: B-lymphocytes, monocytes/macrophages, and endothelial and smooth muscle cells. Increased expression of sCD40L is considered an important factor in the immunopathogenesis of RA. The mechanism of its action is realized through the polyclonal activation of B-lymphocytes, the formation of autoantibodies, the activation of endothelial cells, and the increased production of pro-inflammatory cytokines [37,38]. The diagnostic and prognostic value of sCD40L in patients with CVD and healthy individuals is discussed in the literature. The Ludwigshafen Risk and Cardiovascular Health study demonstrates that sCD40L independently predicted an increase in cardiovascular risk in the general population; a significant association between elevated sCD40L plasma levels and short-term all-cause and cardiovascular mortality was observed in 2759 study participants [39]. The association of elevated blood concentrations of sCD40L with several cardiovascular risk factors has been demonstrated in clinical trials. In particular, it was shown previously that there is a relationship between the CD40/CD40L system and cholesterol metabolism in patients with moderate hypercholesterolemia [40]. Our data indicate a similar relationship between sCD40L levels and cholesterol levels in RA patients with low cardiovascular risk. In addition, blood pressure can affect the level of sCD40L, which was confirmed in the current work and matches the results of other studies [41]. The increased blood level of sCD40L was previously demonstrated in RA patients compared to the control group, but in that study, no association of sCD40L with disease activity and markers of inflammation such as erythrocyte sedimentation rate and C-reactive protein was revealed. There are few works that assess the clinical significance of pCD40L as a marker of atherosclerotic vascular lesions in RA and SLE [42]. Our study found a positive correlation of this marker with traditional and non-traditional risk factors, as well as with atherosclerotic lesions of the carotid arteries in RA patients with low cardiovascular risk. The relationship between sCD40L level and cIMT in RA patients with atherosclerosis identified in our study may have clinical and diagnostic significance in the prediction of subclinical atherosclerosis in RA patients. Activation of the CD40/CD40L system is considered a key step in the development of atherosclerosis and its complications and is associated with the synthesis of adhesion molecules and chemokines by monocytes/macrophages, endotheliocytes, platelet-leukocyte adhesion, increased expression of tissue factor and matrix metalloproteinases, and activation and proliferation of vascular smooth muscle cells [15,43,44]. The binding of CD40L to the receptor on endotheliocytes and smooth muscle cells induces the expression of leukocyte adhesion molecules such as VCAM, E-selectin, and ICAM that causes the recruitment of immune-inflammatory cells with the subsequent development of local inflammation, leading to the formation of atherosclerotic lesions [45,46]. In RA, adhesion molecules VCAM and ICAM produced by synovial fibroblasts play an important role in supporting inflammation, promoting recruiting and adhesion of circulating inflammatory cells to endothelium and, consequently, their extravasation at the site of inflammation and tissue damage [47,48,49]. It has been shown in our study that patients with RA have a higher concentration of sVCAM compared to the individual without autoimmune diseases, and the level of sVCAM and sICAM in the peripheral blood is associated with immunological markers (RF) and RA activity indicators (DAS28), respectively. Similar data have been obtained in several other studies demonstrating elevated levels of adhesion molecules in the blood of RA patients compared to healthy controls and their association with DAS28 and RF [48,50]. sICAM-1 and sVCAM-1 are well-known molecules involved in the pathogenesis of atherosclerotic plaques [51]. In the work of JF.Varona et al., sICAM-1 and sVCAM-1 were strongly associated with early atherosclerotic disease in patients with low/intermediate cardiovascular risk [52]. sICAM-1 and sVCAM-1 contribute to the development of atherosclerosis through different mechanisms, and what mechanism is leading is not yet fully understood. On the one hand, the binding of these soluble adhesion molecules to circulating leukocyte receptors before they come into contact with the vessel wall has an anti-adhesive potential that may limit the immune-inflammatory response [53]. On the other hand, the response of macrophages to sICAM-1 lead to the production of MIP-2 and TNF-a through an NFkappaB-dependent mechanism, which in turn enhances inflammation [54]. ## 5. Conclusions Thus, cIMT is considered to be the most informative marker for the identification of the risk of CVD development in patients with RA. In our study, subclinical atherosclerotic lesions of the carotid arteries were detected much more often in RA patients with low cardiovascular risk than in the control group and were found in a quarter of patients. Therefore, ultrasound scanning of carotid arteries should be recommended for RA patients with low risk of CVD as a routine examination in general clinical practice. The relationship between the cIMT with traditional cardiovascular risk factors (age, sex, blood lipid levels, blood pressure) as well as with immunoinflammatory markers of cardiovascular risk (sCD40L, sVCAM-1) in patients with RA has been demonstrated. Serum levels of sCD40L and sVCAM-1 in RA patients appear to be useful biomarkers for detecting early subclinical atherosclerotic lesions in RA patients with low cardiovascular risk. ## References 1. Pappas D.A., Nyberg F., Kremer J.M., Lampl K., Reed G.W., Horne L., Ho M., Onofrei A., Malaviya A.N., Rillo O.L.. Prevalence of Cardiovascular Disease and Major Risk Factors in Patients with Rheumatoid Arthritis: A Multinational Cross-Sectional Study. Clin. Rheumatol. (2018.0) 37 2331-2340. DOI: 10.1007/s10067-018-4113-3 2. Balsa A., Lojo-Oliveira L., Alperi-López M., García-Manrique M., Ordóñez-Cañizares C., Pérez L., Ruiz-Esquide V., Corrales A., Narváez J., Rey-Rey J.. Prevalence of Comorbidities in Rheumatoid Arthritis and Evaluation of Their Monitoring in Clinical Practice: The Spanish Cohort of the COMORA Study. Reum. Clin. (2019.0) 15 102-108. DOI: 10.1016/j.reuma.2017.06.002 3. Wang H., Li X., Gong G.. Cardiovascular Outcomes in Patients with Co-Existing Coronary Artery Disease and Rheumatoid Arthritis: A Systematic Review and Meta-Analysis. Medicine (2020.0) 99 e19658. DOI: 10.1097/MD.0000000000019658 4. Lee Y.K., Ahn G.Y., Lee J., Shin J.M., Lee T.H., Park D.J., Song Y.J., Kim M.K., Bae S.C.. Excess Mortality Persists in Patients with Rheumatoid Arthritis. Int. J. Rheum. Dis. (2021.0) 24 364-372. DOI: 10.1111/1756-185X.14058 5. Hansildaar R., Vedder D., Baniaamam M., Tausche A.K., Gerritsen M., Nurmohamed M.T.. Cardiovascular Risk in Inflammatory Arthritis: Rheumatoid Arthritis and Gout. Lancet Rheumatol. (2021.0) 3 e58-e70. DOI: 10.1016/S2665-9913(20)30221-6 6. Conroy R.M., Pyörälä K., Fitzgerald A.P., Sans S., Menotti A., De Backer G., De Bacquer D., Ducimetière P., Jousilahti P., Keil U.. Estimation of Ten-Year Risk of Fatal Cardiovascular Disease in Europe: The SCORE Project. Eur. Heart J. (2003.0) 24 987-1003. DOI: 10.1016/S0195-668X(03)00114-3 7. Agca R., Heslinga S.C., Rollefstad S., Heslinga M., McInnes I.B., Peters M.J.L., Kvien T.K., Dougados M., Radner H., Atzeni F.. EULAR Recommendations for Cardiovascular Disease Risk Management in Patients with Rheumatoid Arthritis and Other Forms of Inflammatory Joint Disorders: 2015/2016 Update. Ann. Rheum. Dis. (2017.0) 76 17-28. DOI: 10.1136/annrheumdis-2016-209775 8. Dessein P.H., Semb A.G.. Could Cardiovascular Disease Risk Stratification and Management in Rheumatoid Arthritis Be Enhanced?. Ann. Rheum. Dis. (2013.0) 72 1743-1746. DOI: 10.1136/annrheumdis-2013-203911 9. Colaco K., Ocampo V., Ayala A.P., Harvey P., Gladman D.D., Piguet V., Eder L.. Predictive Utility of Cardiovascular Risk Prediction Algorithms in Inflammatory Rheumatic Diseases: A Systematic Review. J. Rheumatol. (2020.0) 47 928-938. DOI: 10.3899/jrheum.190261 10. O’Leary D.H., Bots M.L.. Imaging of Atherosclerosis: Carotid Intima-Media Thickness. Eur. Heart J. (2010.0) 31 1682-1689. DOI: 10.1093/eurheartj/ehq185 11. Khanna N.N., Jamthikar A.D., Gupta D., Nicolaides A., Araki T., Saba L., Cuadrado-Godia E., Sharma A., Omerzu T., Suri H.S.. Performance Evaluation of 10-Year Ultrasound Image-Based Stroke/Cardiovascular (CV) Risk Calculator by Comparing against Ten Conventional CV Risk Calculators: A Diabetic Study. Comput. Biol. Med. (2019.0) 105 125-143. DOI: 10.1016/j.compbiomed.2019.01.002 12. Popescu D., Rezus E., Codruta Badescu M., Dima N., Nicoleta P., Isac S., Dragoi I.-T., Rezus C.. Cardiovascular Risk Assessment in Rheumatoid Arthritis: Accelerated Atherosclerosis, New Biomarkers, and the Effects of Biological Therapy. Life (2023.0) 13. DOI: 10.3390/life13020319 13. Reiss A.B., Silverman A., Khalfan M., Vernice N.A., Kasselman L.J., Carsons S.E., De Leon J.. Accelerated Atherosclerosis in Rheumatoid Arthritis: Mechanisms and Treatment. Curr. Pharm. Des. (2019.0) 25 969-986. DOI: 10.2174/1381612825666190430113212 14. Fomicheva O.A., Popkova T.V., Krougly L.B., Gerasimova E.V., Novikova D.S., Pogorelova O.A.. Factors of Progression and Occurrence of Atherosclerosis in Rheumatoid Arthritis. Kardiologiia (2021.0) 61 12-21. DOI: 10.18087/cardio.2021.1.n1331 15. Lacy M., Bürger C., Shami A., Ahmadsei M., Winkels H., Nitz K., van Tiel C.M., Seijkens T.T.P., Kusters P.J.H., Karshovka E.. Cell-Specific and Divergent Roles of the CD40L-CD40 Axis in Atherosclerotic Vascular Disease. Nat. Commun. (2021.0) 12 3754. DOI: 10.1038/s41467-021-23909-z 16. Pereira-Da-silva T., Napoleão P., Pinheiro T., Selas M., Silva F., Ferreira R.C., Carmo M.M.. The Proinflammatory Soluble CD40 Ligand Is Associated with the Systemic Extent of Stable Atherosclerosis. Medicina (2021.0) 57. DOI: 10.3390/medicina57010039 17. Galkina E., Ley K.. Vascular Adhesion Molecules in Atherosclerosis. Arter. Thromb. Vasc. Biol. (2007.0) 27 2292-2301. DOI: 10.1161/ATVBAHA.107.149179 18. Pereira-da-Silva T., Ferreira V., Castelo A., Caldeira D., Napoleão P., Pinheiro T., Ferreira R.C., Carmo M.M.. Soluble CD40 Ligand Expression in Stable Atherosclerosis: A Systematic Review and Meta-Analysis. Atherosclerosis (2021.0) 319 86-100. DOI: 10.1016/j.atherosclerosis.2020.12.011 19. Södergren A., Karp K., Bengtsson C., Möller B., Rantapää-Dahlqvist S., Wållberg-Jonsson S.. Biomarkers Associated with Cardiovascular Disease in Patients with Early Rheumatoid Arthritis. PLoS ONE (2019.0) 14. DOI: 10.1371/journal.pone.0220531 20. Román-Fernández I.V., García-Chagollán M., Cerpa-Cruz S., Jave-Suárez L.F., Palafox-Sánchez C.A., García-Arellano S., Sánchez-Zuno G.A., Muñoz-Valle J.F.. Assessment of CD40 and CD40L Expression in Rheumatoid Arthritis Patients, Association with Clinical Features and DAS28. Clin. Exp. Med. (2019.0) 19 427-437. DOI: 10.1007/s10238-019-00568-5 21. Kay J., Upchurch K.S.. ACR/EULAR 2010 Rheumatoid Arthritis Classification Criteria. Rheumatology (2012.0) 51 vi5-vi9. DOI: 10.1093/rheumatology/kes279 22. Solomon D.H., Greenberg J., Curtis J.R., Liu M., Farkouh M.E., Tsao P., Kremer J.M., Etzel C.J.. Derivation and Internal Validation of an Expanded Cardiovascular Risk Prediction Score for Rheumatoid Arthritis: A Consortium of Rheumatology Researchers of North America Registry Study. Arthritis Rheumatol. (2015.0) 67 1995-2003. DOI: 10.1002/art.39195 23. Touboul P.J., Grobbee D.E., Den Ruijter H.. Assessment of Subclinical Atherosclerosis by Carotid Intima Media Thickness: Technical Issues. Eur. J. Prev. Cardiol. (2012.0) 19 18-24. DOI: 10.1177/2047487312448990 24. Mackey R.H., Kuller L.H., Moreland L.W.. Cardiovascular Disease Risk in Patients with Rheumatic Diseases. Clin. Geriatr. Med. (2017.0) 33 105-117. DOI: 10.1016/j.cger.2016.08.008 25. Gerasimova E.V., Popkova T.V., Gerasimova D.A., Glukhova S.I., Nasonov E.L., Lila A.M.. Application of Cardiovascular Risk Scales to Identify Carotid Atherosclerosis in Patients with Rheumatoid Arthritis. Ter. Arkh. (2021.0) 93 561-567. DOI: 10.26442/00403660.2021.05.200787 26. Edwards N., Langford-Smith A.W.W., Parker B.J., Bruce I.N., Reynolds J.A., Alexander M.Y., McCarthy E.M., Wilkinson F.L.. QRISK3 Improves Detection of Cardiovascular Disease Risk in Patients with Systemic Lupus Erythematosus. Lupus Sci. Med. (2018.0) 5 e000272. DOI: 10.1136/lupus-2018-000272 27. Corrales A., Vegas-Revenga N., Rueda-Gotor J., Portilla V., Atienza-Mateo B., Blanco R., Castañeda S., Ferraz-Amaro I., Llorca J., González-Gay M.A.. Carotid Plaques as Predictors of Cardiovascular Events in Patients with Rheumatoid Arthritis. Results from a 5-Year-Prospective Follow-up Study. Semin. Arthritis Rheum. (2020.0) 50 1333-1338. DOI: 10.1016/j.semarthrit.2020.03.011 28. Di Battista M., Barsotti S., Della Rossa A., Mosca M.. Cardiovascular Burden in Systemic Sclerosis: QRISK3 versus Framingham for Risk Estimation. Mod. Rheumatol. (2022.0) 32 584-588. DOI: 10.1093/mr/roab011 29. Kobayashi H., Giles J.T., Polak J.F., Blumenthal R.S., Leffell M.S., Szklo M., Petri M., Gelber A.C., Post W., Bathon J.M.. Increased Prevalence of Carotid Artery Atherosclerosis in Rheumatoid Arthritis Is Artery-Specific. J. Rheumatol. (2010.0) 37 730-739. DOI: 10.3899/jrheum.090670 30. Hannawi S., Haluska B., Marwick T.H., Thomas R.. Atherosclerotic Disease Is Increased in Recent-Onset Rheumatoid Arthritis: A Critical Role for Inflammation. Arthritis Res. (2007.0) 9 R116. DOI: 10.1186/ar2323 31. Giles J.T., Post W.S., Blumenthal R.S., Polak J., Petri M., Gelber A.C., Szklo M., Bathon J.M.. Longitudinal Predictors of Progression of Carotid Atherosclerosis in Rheumatoid Arthritis. Arthritis Rheum. (2011.0) 63 3216-3225. DOI: 10.1002/art.30542 32. Ambrosino P., Lupoli R., di Minno A., Tasso M., Peluso R., di Minno M.N.D.. Subclinical Atherosclerosis in Patients with Rheumatoid Arthritis. A Meta-Analysis of Literature Studies. Thromb. Haemost. (2015.0) 113 916-930. DOI: 10.1160/TH14-11-0921 33. Udachkina E.V., Novikova D.S., Popkova T.V., Kirillova I.G., Markelova E.I., Gorbunova Y.N., Karateev D.E., Luchikhina E.L., Demidova N.V., Borisova M.A.. Progression of Carotid Artery Atherosclerosis during Treatment to Target in Patients with Early Rheumatoid Arthritis. Rheumatol. Sci. Pract. (2018.0) 56 449-455. DOI: 10.14412/1995-4484-2018-449-455 34. Dehghan P., Rajaei A., Moeineddin R., Alizadeh A.M.. Prevalence of Atherosclerosis in Patients with Inactive Rheumatoid Arthritis. Clin. Rheumatol. (2015.0) 34 1363-1366. DOI: 10.1007/s10067-015-2996-9 35. Hannawi S.M.A., Hannawi H., Alokaily F., Salmi I.. Al Subclinical Atherosclerosis in Rheumatoid Arthritis Patients of the Gulf Cooperated Council. Saudi Med. J. (2020.0) 41 1022-1025. DOI: 10.15537/smj.2020.9.25319 36. Dalbeni A., Giollo A., Bevilacqua M., Cioffi G., Tagetti A., Cattazzo F., Orsolini G., Ognibeni F., Minuz P., Rossini M.. Traditional Cardiovascular Risk Factors and Residual Disease Activity Are Associated with Atherosclerosis Progression in Rheumatoid Arthritis Patients. Hypertens. Res. (2020.0) 43 922-928. DOI: 10.1038/s41440-020-0441-1 37. Liu M.F., Chao S.C., Wang C.R., Lei H.Y.. Expression of CD40 and CD40 Ligand among Cell Populations within Rheumatoid Synovial Compartment. Autoimmunity (2001.0) 34 107-113. DOI: 10.3109/08916930109001958 38. Alaaeddine N., Hassan G.S., Yacoub D., Mourad W.. CD154: An Immunoinflammatory Mediator in Systemic Lupus Erythematosus and Rheumatoid Arthritis. Clin. Dev. Immunol. (2012.0) 2012 490148. DOI: 10.1155/2012/490148 39. Gergei I., Kälsch T., Scharnagl H., Kleber M.E., Zirlik A., März W., Krämer B.K., Kälsch A.I.. Association of Soluble CD40L with Short-Term and Long-Term Cardiovascular and All-Cause Mortality: The Ludwigshafen Risk and Cardiovascular Health (LURIC) Study. Atherosclerosis (2019.0) 291 127-131. DOI: 10.1016/j.atherosclerosis.2019.09.004 40. Garlichs C.D., John S., Schmeier A., Eskafi S., Stumpf C., Karl M., Goppelt-Struebe M., Schmieder R., Daniel W.G.. Upregulation of CD40 and CD40 Ligand (CD154) in Patients with Moderate Hypercholesterolemia. Circulation (2001.0) 104 2395-2400. DOI: 10.1161/hc4501.099312 41. Guzel M., Dogru M.T., Simsek V., Demir V., Alp C., Kandemir H., Yildirim N., Kisa U.. Influence of Circadian Blood Pressure Alterations on Serum SCUBE-1 and Soluble CD40 Ligand Levels in Patients with Essential Hypertension. Am. J. Cardiovasc. Dis. (2019.0) 9 42. PMID: 31516762 42. Pamuk O.N., Tozkir H., Uyanik M.S., Gurkan H., Saritas F., Duymaz J., Donmez S., Yazar M., Pamuk G.E.. PECAM-1 Gene Polymorphisms and Soluble PECAM-1 Level in Rheumatoid Arthritis and Systemic Lupus Erythematosus Patients: Any Link with Clinical Atherosclerotic Events?. Clin. Rheumatol. (2014.0) 33 1737-1743. DOI: 10.1007/s10067-014-2771-3 43. Schönbeck U., Sukhova G.K., Shimizu K., Mach F., Libby P.. Inhibition of CD40 Signaling Limits Evolution of Established Atherosclerosis in Mice. Proc. Natl. Acad. Sci. USA (2000.0) 97 7458. DOI: 10.1073/pnas.97.13.7458 44. Gocmen A.Y., Ocak G.A., Ozbilim G., Delibas N., Gumuslu S.. Effect of Atorvastatin on Atherosclerotic Plaque Formation and Platelet Activation in Hypercholesterolemic Rats. Can. J. Physiol. Pharm. (2013.0) 91 680-685. DOI: 10.1139/cjpp-2012-0325 45. Hakonarson H., Kim C., Whelan R., Campbell D., Grunstein M.M.. Bi-Directional Activation between Human Airway Smooth Muscle Cells and T Lymphocytes: Role in Induction of Altered Airway Responsiveness. J. Immunol. (2001.0) 166 293-303. DOI: 10.4049/jimmunol.166.1.293 46. Antoniades C., Bakogiannis C., Tousoulis D., Antonopoulos A.S., Stefanadis C.. The CD40/CD40 Ligand System: Linking Inflammation with Atherothrombosis. J. Am. Coll. Cardiol. (2009.0) 54 669-677. DOI: 10.1016/j.jacc.2009.03.076 47. Vascular Cell Adhesion Molecule 1 (CD106): A Multifaceted Regulator of Joint Inflammation 48. Wang L., Ding Y., Guo X., Zhao Q.. Role and Mechanism of Vascular Cell Adhesion Molecule-1 in the Development of Rheumatoid Arthritis. Exp. Med. (2015.0) 10 1229-1233. DOI: 10.3892/etm.2015.2635 49. Liu Y., Wei W., Hong C., Wang Y., Sun X., Ma J., Zheng F.. Calreticulin Induced Endothelial ICAM-1 up-Regulation Associated with Tristetraprolin Expression Alteration through PI3K/Akt/ENOS/P38 MAPK Signaling Pathway in Rheumatoid Arthritis. Mol. Immunol. (2019.0) 107 10-20. DOI: 10.1016/j.molimm.2019.01.005 50. Navarro-Hernández R.E., Oregon-Romero E., Del Mercado M.V., Rangel-Villalobos H., Palafox-Sánchez C.A., Muoz-Valle J.F.. Expression of ICAM1 and VCAM1 Serum Levels in Rheumatoid Arthritis Clinical Activity. Association with Genetic Polymorphisms. Dis. Mrk. (2009.0) 26 119-126. DOI: 10.1155/2009/510104 51. Gómez Rosso L., Benítez M.B., Fornari M.C., Berardi V., Lynch S., Schreier L., Wikinski R., Cuniberti L., Brites F.. Alterations in Cell Adhesion Molecules and Other Biomarkers of Cardiovascular Disease in Patients with Metabolic Syndrome. Atherosclerosis (2008.0) 199 415-423. DOI: 10.1016/j.atherosclerosis.2007.11.007 52. Varona J.F., Ortiz-Regalón R., Sánchez-Vera I., López-Melgar B., García-Durango C., Castellano Vázquez J.M., Solís J., Fernández-Friera L., Vidal-Vanaclocha F.. Soluble ICAM 1 and VCAM 1 Blood Levels Alert on Subclinical Atherosclerosis in Non Smokers with Asymptomatic Metabolic Syndrome. Arch. Med. Res. (2019.0) 50 20-28. DOI: 10.1016/j.arcmed.2019.05.003 53. Videm V., Albrigtsen M.. Soluble ICAM-1 and VCAM-1 as Markers of Endothelial Activation. Scand. J. Immunol. (2008.0) 67 523-531. DOI: 10.1111/j.1365-3083.2008.02029.x 54. Soluble ICAM-1 Activates Lung Macrophages and Enhances Lung Injury—PubMed
--- title: 'Turning Back the Clock: A Retrospective Single-Blind Study on Brain Age Change in Response to Nutraceuticals Supplementation vs. Lifestyle Modifications' authors: - Andrew A. Fingelkurts - Alexander A. Fingelkurts journal: Brain Sciences year: 2023 pmcid: PMC10046544 doi: 10.3390/brainsci13030520 license: CC BY 4.0 --- # Turning Back the Clock: A Retrospective Single-Blind Study on Brain Age Change in Response to Nutraceuticals Supplementation vs. Lifestyle Modifications ## Abstract Background: There is a growing consensus that chronological age (CA) is not an accurate indicator of the aging process and that biological age (BA) instead is a better measure of an individual’s risk of age-related outcomes and a more accurate predictor of mortality than actual CA. In this context, BA measures the “true” age, which is an integrated result of an individual’s level of damage accumulation across all levels of biological organization, along with preserved resources. The BA is plastic and depends upon epigenetics. Brain state is an important factor contributing to health- and lifespan. Methods and Objective: Quantitative electroencephalography (qEEG)-derived brain BA (BBA) is a suitable and promising measure of brain aging. In the present study, we aimed to show that BBA can be decelerated or even reversed in humans ($$n = 89$$) by using customized programs of nutraceutical compounds or lifestyle changes (mean duration = 13 months). Results: We observed that BBA was younger than CA in both groups at the end of the intervention. Furthermore, the BBA of the participants in the nutraceuticals group was 2.83 years younger at the endpoint of the intervention compared with their BBA score at the beginning of the intervention, while the BBA of the participants in the lifestyle group was only 0.02 years younger at the end of the intervention. These results were accompanied by improvements in mental–physical health comorbidities in both groups. The pre-intervention BBA score and the sex of the participants were considered confounding factors and analyzed separately. Conclusions: Overall, the obtained results support the feasibility of the goal of this study and also provide the first robust evidence that halting and reversal of brain aging are possible in humans within a reasonable (practical) timeframe of approximately one year. ## 1. Introduction For millennia humans have been fascinated by the prospect of living forever. This aspiration has left noticeable marks in virtually every human culture reflecting on the possibility of transcending death [2,3]. While such an extreme wish to attain some form of immortality is still implicitly embedded in the so-called movement of “posthumanism” (posthumanism seeks to improve human nature by using technology to transcend the limitations of the body and mind [4,5,6]) (for a brief overview, see [7]), in biomedical science it has been transformed into a more practical aim of slowing down or potentially even reversing aging [8,9,10,11,12], progressively reaching the “age escape velocity” (such an approach presupposes that death could be interactively delayed by anticipating and fixing the damaging effects of aging across the lifespan [13]), which will open the prospect of extreme human life extension [14]. Over the past half century, life expectancy and the observed maximum age at death have increased dramatically [15], probably due to the successes of evidence-based medicine, which have been very effective at reducing mortality over the past few decades [16]. At the same time, it has become painfully evident that not all of the gained extra years are healthy: estimates have shown that the proportion of life characterized by good health has remained rather constant between 1990 and 2019 [17], implying that most of the life years gained are lived largely in poor health [12]. As pointed out by Olshansky [18], this leads to a situation where a significant portion of the lifespan is lived during a window of exponentially increasing risk of frailty and chronic disability (Figure 1), with the simultaneous manifestation of many chronic conditions as late life comorbidities [16,19,20]. Therefore, there is an increasing understanding of the importance of so-called “healthy aging” (healthy aging refers to the “healthspan”, which is a period of life free from serious chronic diseases and disability [21]. It has been proposed that by increasing the healthspan, one could achieve optimal longevity, when illness, disability, and their sequelae would be restricted to a very short period at the end of life—termed “compression of morbidity” [22]. Such optimal longevity would signify entering a fourth stage of epidemiological transition according to Omran [23]—the age of delayed degenerative diseases [24]) [21,25] and an unprecedented advance in research that focuses on the biology of aging [9,11,26,27]. Aging is commonly characterized as a progressive loss of physiological function due to the accumulation of molecular and cellular damage, leading to the development of chronic comorbidities that include metabolic, immune, cardiovascular, neoplastic, and neurodegenerative disorders, accompanied by geriatric symptoms, such as frailty and immobility [28,29,30]. Over the past few decades, some of the mechanistic pathways involved in aging have been elucidated; they are known as mechanisms [31], principles [32], biomarkers [33], hallmarks [34], pillars [35], or predictors [36] of aging. While the actual number of these hallmarks varies depending on the authors, the total is nine [34]: [1] genomic instability, [2] epigenetic alterations, [3] loss of proteostasis, [4] deregulated nutrient sensing, [5] mitochondrial dysfunction, [6] cellular senescence, [7] stem cell exhaustion, [8] altered intercellular communication, and [9] telomere attrition (recently, three additional hallmarks were added: chronic inflammation, disabled macroautophagy, and dysbiosis). Although the contribution of each of these hallmarks to the progression of aging is far from being completely understood (for a critical discussion of the hallmarks of aging, see Gems and de Magalhães [37]), it is nevertheless clear that they are interconnected and play a significant causal role in the process of aging [30]. When speaking about age, two concepts are sometimes used interchangeably, but they nonetheless have to be distinguished [29]: chronological age (CA) and biological age (BA) (Figure 1). Until recently, CA was a commonly used indicator of aging [38] as a universal feature shared by all living beings [16]; however, it only measures how much time has passed since birth, and it increases at the same rate for everyone [39,40]. CA has been shown to be a strong predictor of health status and mortality [19]. At the same time, life expectancy shows considerable variation among individuals with an equal or similar CA [38]. This means that if, for example, one chronological year has passed, it does not necessarily mean that an individual has also aged in biological terms the equivalent of one year [39,41]. It seems that the speed of aging processes varies both between different people [36,42], even in twins [41], and also within the same individual at different periods of the lifespan—the fluidity of ageotypes [42] (see also [43,44]). Therefore, CA is not an accurate indicator of the aging progress [45]. These inter- and intraindividual differences in aging can be captured by BA [35,46,47], which is thought to measure an individual’s risk of age-related outcomes and predict mortality better than actual CA [36,48]. In this context, BA (being a quantitative phenotype [29]) measures the “true” age (multiple longitudinal studies have shown that BA is the most convenient and reliable measure to determine the extent of age-related (i.e., biomarker) changes in an organism [29]. In this context, higher BA values are indicative of a higher intensity of age-related detrimental processes in comparison with CA, while lower BA values are proxy markers of a lower intensity of aging processes and overall higher resilience to them. Traditionally, BA metrics are designed to resemble the CA distribution within a cohort of healthy individuals, however, being more predictive of a person’s health status than CA itself [49]), which is an integrated result of an individual’s level of damage accumulation (i.e., price) at all levels of biological organization and preserved (a) capacity (i.e., maximal processing power), (b) efficiency (i.e., minimum number of operations and the energy expenditure per operation), (c) interprocess coordination, (d) functional integrity, and (e) resources (i.e., biocapital) which together determine resilience (i.e., compensatory and recovering mechanisms) and are associated with the risks of CA-related diseases, vigilance and cognitive decline, reduction in quality of life, and, ultimately, mortality [19,21,50,51,52,53,54,55] (Figure 1). Continuously growing data suggest that variability in the BA process is due to the diversity in genotypes (i.e., longevity or senescent mutations), family history (for example, having long-lived parents and grandparents is strongly correlated with a longer lifespan [56]), lifestyle habits (e.g., smoking, alcohol/drugs consumption, type of diet, physical and mental/intellectual exercise, duration and quality of sleep, medication use, occupational complexity, leisure activity, and social engagement), and environments that include (i) early-life development (i.e., utero characteristics and early stress/trauma), (ii) socioeconomic status, (iii) education level, and (iv) malnutrition, vitamins, and/or nutrient deficiencies or imbalances [39,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72]. Therefore, BA is plastic and hinges on the balance between the factors mentioned above [73,74]. This interaction of genotype with living habits and the environment is known as epigenetics [75]. Epigenetics is a complex biological mechanism that can switch genes ON or OFF, e.g., sleep, diet, and exercise can all cause chemical modifications around specific genes (DNA methylation is one of the major classes of epigenetic modifications in which a methyl group (one carbon) is covalently added to the C5 position of a cytosine base [76]. The degree of DNA methylation defines gene expression. The other classes include histone modifications and chromatin remodeling [77]) and histone proteins, hence, either promoting or silencing their expression over time and even leading to heritable changes to the genome without changes to the DNA sequence itself [77]. Accordingly, epigenome changes have consequences for the molecular pathways of cells, tissues, and organs [78]. Increasing empirical evidence demonstrates that certain changes in the epigenome during aging lead to genomic alterations and instability, contributing to the initiation of age-related diseases, such as cancer and neurodegenerative diseases (interest in epigenetic mechanisms is increasing due to the current evidence that epigenetic changes are capable of transmission across generations—so-called “epigenetic inheritance”—when several epigenetic marks are transferred to offspring, who inherit the phenotype in the absence of the external influence [79,80,81,82]. In humans, the transgenerational epigenetic effect has been shown in association with nutrition and food supply. For example, the experience of famine by women in early gestation is associated with glucose intolerance and chronic disease, as well as obesity and cardiovascular diseases, in her children and grandchildren [83,84]. Similarly, there are long-term consequences for the offspring’s later health induced by maternal obesity during pregnancy [85]) [86,87]. It has even been proposed that epigenetic modifications represent the primary driver or cause of aging (as a consequence, it has been proposed that an epigenetic assault on aging is a feasible way to reduce multimorbidities in an aging population and even potentially to reprogram the organism to a more youthful state [88,89,90]. The principle possibility of age reprogramming (reverting a differentiated cell back to an induced pluripotent stem cell) was demonstrated by Yamanaka over a decade ago [91]. Since then, reverse programming research has witnessed an explosion [92,93,94,95,96]) [9,10,34,86,97,98,99,100]. Indeed, older organisms have a different epigenome [101], while individuals with “slower” biological aging have a lower risk for morbidity, disability, and mortality (for example, it has been shown that each one-year increase in epigenetic age is associated with a 9 percent increase in all-cause mortality, a 10 percent increase in cardiovascular-related mortality, a 7 percent increase in cancer-related mortality, a 20 percent increase in diabetes-related mortality, and a 9 percent increase in chronic lower respiratory disease mortality, even after adjusting for chronological age [102]) [46,103,104], and in supercentenarians (supercentenarians are individuals who reach 110-year or longer lifespan [105]) the epigenetic age is younger than their CA, thus likely playing a significant role in their extremely long lifespan (curiously, such an association between the epigenetic clock “ticking” and longevity is also observed in other species. For example, the epigenetic clock ticks faster in chimpanzees than in humans [106], which is consistent with the fact that humans have approximately a four-fold greater maximum lifespan than chimpanzees [107]) [108,109]. However, most epigenetic research in the aging field has largely focused on the relationship of the epigenome with the overall organismal longevity and aging [102,110,111]. At the same time, growing research indicates that such primary causes of death, such as cardiovascular diseases and cancer, are progressively declining [112,113], while mortality due to the fact of neurodegenerative disorders, such as different dementias, Alzheimer’s disease, or Parkinson’s, has increased by $145\%$ over the last 20 years [114,115], thus implying that brain state is an important factor contributing to the overall health- and lifespan (Figure 1). Indeed, cognitive decline, neurodegeneration (neurodegeneration is one of the most fundamental pathological mechanisms shared by many brain disorders and different subtypes of dementia, including Alzheimer’s disease and Parkinson’s dementia [116]. Neurodegeneration is usually accompanied by impaired neurogenesis [117] and abnormal protein aggregations [118], which are products of dysfunctional autophagy [119], mitochondrial dysfunction, oxidative damage, and inflammation [120,121]), and many other brain disorders are “champions” of advanced age [28,122], so the brain’s link to the human lifespan is unmistakable, although understudied. All along, the brain contributes to the lifespan directly through a so-called circadian time-keeping system—the “central” circadian clock, which is located in the hypothalamic suprachiasmatic nucleus (SCN) [123,124,125]. This central clock dictates systemic and peripheral circadian behavior and rhythms by synchronizing the neuroendocrine system to the external light–dark cycle [126,127,128]. Disruptions in this central clock result in metabolic deregulation [129], cancer initiation [130,131], and accelerated aging and decreased longevity [132,133,134]. It has been further proposed that the brain also synchronizes the organismal epigenetic clock (including its rate in every tissue—tissue-specific epigenetic clocks [135]), suggesting the central role of the brain in the organismal health- and lifespan [136]. This may explain why an “older” brain may be hostile to a younger body [137] and is also in line with the finding that persons with an older brain age experienced at least two decades of accelerated age-related degradation of the body [138]. Indeed, many neurological and psychiatric diseases (such as schizophrenia, depression, epilepsy, HIV encephalopathy, Alzheimer’s, and traumatic brain injury) are associated with premature or accelerated aging (for an overview, see [139]; see also [140,141]). These observations have recently been supported by the estimation of the epigenetic clock rate: epigenetic aging is accelerated in schizophrenia [142,143], depression [144,145], post-traumatic stress disorder [146], HIV infection [147], Alzheimer’s disease [148], Huntington’s disease [149], and Parkinson’s disease [108]. There is, however, another important “product” of brain activity—subjectivity [150,151,152]—which has largely been ignored until recently in relation to aging and longevity [153,154] but which, nevertheless, stresses the importance of the brain for longevity. Indeed, the subjective perception of age may have profound effects on health and well-being, and it is connected to an individual’s lifespan [153,155]. For example, in a study using 2.253 adults, it was shown that an older subjective age was associated with accelerated epigenetic aging [156]. A link between subjective age and the probability of mortality has been established in three large samples [157]: a subjective age of approximately 8, 11, and 13 years older than CA in the three samples was correlated with an $18\%$, $29\%$, and $25\%$ higher risk of mortality, respectively. This link was confirmed in a meta-analysis of 19 longitudinal studies [155]. Recently, Zhavoronkov et al. [ 154] have shown that a subjective age that is +5 years more than the CA is associated with a more than two-fold increase in the mortality rate, and a subjective age that is –5 years less is clearly a major life protective factor (these findings have been corroborated by data obtained at the molecular level measuring the length of telomeres [158]. Telomeres are DNA–protein complexes that cap chromosomal ends, promoting chromosomal stability [159], and their length is a factor limiting the maximum number of cell divisions (i.e., the Hayflick limit) and the regenerative potential [160]. Telomeres shorten with age (i.e., the so-called “telomere attrition”) and, thus, telomere length often serves as a biomarker of cellular aging—senescence [161,162]. It was shown that an older subjective age is related to shorter telomeres, beyond what is expected as the CA effect [158]). Furthermore, a younger subjective age is associated with a lower risk of major depressive episodes [163], while an older perceived age predicts higher depressive symptoms or full depression in the future [164,165]. Additionally, a younger subjective age is associated with improved cognitive functioning 10 years later [166] and is associated with personality traits such as openness, conscientiousness, agreeableness, and extraversion [167] (see also [154]). Interestingly, elderly individuals that reported a subjective age similar to or younger than their actual CA have higher grey matter volume in several brain areas, and this subjective age was a reliable predictor of brain age [168]. Overall, people who feel subjectively younger have more resources, better mental and physical health, higher cognitive abilities, enhanced resilience to stress, a younger biological age (as measured by the epigenetic clock), and a longer lifespan [153,154] (see also [156,169]). Hence, we argue here that brain aging is the strongest risk factor for health- and lifespan, and it is a major contributor to quality of life and subjective well-being associated with the extension of lifespan and longevity (Figure 1). Thus, establishing effective biomarkers of brain aging is particularly important to better understand the aging process and contribute to a long healthspan by reducing neurodegenerative diseases of aging [170]. Furthermore, such brain age biomarkers may help guide the development of interventions to slow the aging process and extend the healthspan of the whole organism (not just the brain). Indeed, considering that the brain is a “chief” organ (in fact, contemporary neuroscience increasingly regards the health of the brain as being key to mental and general health, especially in light of new discoveries of the brain’s compensatory properties for the weak function of vital organs of the organism [171]) which controls, regulates, modifies, or modulates a multitude of physiological (and psychological), neuroendocrine, and immune processes [172,173,174], it contributes to multiple age-related comorbidities [139,175] (for example, cognitive decline and increased Alzheimer’s disease (AD) risk are associated with coronary heart disease, hypertension, and type 2 diabetes [59]). Thus, considering the “competing risks argument” [176], one may expect that reducing brain aging could also have a high impact on systemic/organismal life expectancy and healthspan, because the brain rejuvenation effect should be “felt” across multiple tissues and, hence, reflected in many age-related diseases. Indeed, it has been demonstrated recently that overexpressing sirtuins (sirtuins (SIRT1–7) are a family of nicotinamide adenine dinucleotide (NAD+)-dependent deacylases with many roles that prevent multiple diseases (control of energy metabolism, cell survival, DNA repair, tissue regeneration, inflammation, and neuronal signaling) and can even reverse aspects of aging, as well as prolong life [177]) exclusively in the mouse brain resulted in a longer mean lifespan of the whole organism, as well as a significant increase in the maximal longevity (importantly, sirtuin levels decline in the brain with age, and this relates to an overall health decline [178]. This process is associated with an age-dependent reduction in NAD+ levels in the brain of healthy individuals [179] and also with accelerated brain aging [180]) [181]. ## 1.1. Brain Biological Age Estimation What could be an appropriate measure of brain biological age (BBA)? Currently, there are several biological (epigenetic) “clocks” available that are based on DNA-methylation (DNAm) profiles (additionally, recent advances in artificial intelligence have allowed the development of other age biomarker measures based on (i) blood biochemistry [44,182], (ii) transcriptomics and proteomics [183], and (iii) the microbiome [184]); these are (i) the DNAm age clock [185], (ii) the DNAm age H [186], (iii) the DNAm PhenoAge [102], and (iv) the GimAge or DNAm age G [187]. Although it is well known that the aging process exhibits a tissue-specific signature [188,189] and that DNA methylation patterns are distinct between tissue and cell types [190], epigenetic clocks encompass pan-tissue aging changes, and all of them do not perform optimally in human brain tissue (this does not mean that a meaningful association between systemic DNAm age and neuropathology was not found. On the contrary, there is a robust association between DNAm and Alzheimer’s disease and Parkinson’s disease [191,192]. Moreover, accelerated DNAm age is associated with specific markers (e.g., neuritic plaques, diffuse plaques, and amyloid-b load) of Alzheimer’s disease and declining global cognitive functioning and deficits in episodic and working memory in persons with Alzheimer’s disease [102,193,194]) [190] (see also [195]), and brain aging also does not correlate with epigenetic aging ([196] and references within). Furthermore, almost all DNAm clock measures are invasive; they require either blood samples or samples derived from certain tissues of the organism, which impose multiple limitations on their usage in experimental settings and real-life applications [29]. As a consequence, these make DNAm clocks unsuitable for routine BBA estimation in living humans. Ideally, the BBA measure should be easily available, cheap, and noninvasive (Figure 1). Structural brain changes during normal aging comprise progressive decreases in grey and white matter (grey matter refers to the totality of neuronal cell bodies (also named soma), while white matter denotes the totality of myelinated axons, which are long relays that extend out from the soma (and which are whiteish in color due to the relatively high lipid content of the myelin protein that sheathes them) and form connections between neurons [197]) [198], which together are a major contributor to morbidity and loss of independence in older adults [199]. For example, postmortem brain studies indicate that myelin lipid loss (part of white matter) is progressive throughout adulthood, exceeding a $40\%$ decrease by 100 years of age [200]. Furthermore, long-distance connections show age-related reductions in both anatomical and functional connectivity [201]. These changes are associated with both general cognitive ability and processing speed decreases [202,203]. However, there is a significant interindividual variability in structural brain aging among older adults [204,205], which is uncoupled from CA, sex, education, or clinical markers such as body mass index (BMI) or uric acid [198,206,207,208]. Indeed, some older individuals experience strong and early manifestations of brain degeneration (i.e., accelerated brain aging), while others of comparable age do not experience the brain changes expected at that age (i.e., decelerated brain aging) [169,205,209,210]. Magnetic resonance imaging (MRI) of the brain can reliably detect subtle signs of brain structural aging decades before the onset of age-related disease [211,212]. These observations led to the emergence of the concept of brain age, which is a value estimated using a machine learning algorithm that is trained to predict CA from grey and white matter measures in several independent samples of individuals [53,213,214,215,216]. It was shown that age-related alterations in the brain structure that make the brain appear “older” are associated with Alzheimer’s disease, type 2 diabetes mellitus, a higher BMI, elevated cholesterol and fasting glucose levels, higher diastolic blood pressure, epilepsy, greater smoking and alcohol consumption, more severe depression, and mortality [54,55,141,207,215,217,218]. In summary, MRI-derived brain age reflects only structural brain aging—brain atrophy [169] (additionally, MRI is expensive, nonportable, and usually associated with high stress due to the loud noise and confined space [219]). However, a converging line of evidence suggests some level of decoupling between structure and function in the brain [220]. Indeed, observations in neurology demonstrated that (a) there is a relative disconnect between the clinical presentation and the underlying neuropathology or amount of brain damage—quite often patients that sustain severe, extensive, and irreversible bilateral physical brain damage have preserved functions or eventually recover in part or fully over time [221,222,223,224,225,226]; (b) different neuropsychological profiles are observed in patients with similar brain damage [223]; (c) in spite of a strong link between physiological and clinical health markers with structural brain aging, often no effects on cognitive scores are found [207]; (d) cognitively unimpaired elderly subjects are characterized by structural changes in the brain that reflect accelerated aging [207]; at the same time, (e) full pathologic criteria for Alzheimer’s disease have been observed postmortem in 25–$67\%$ of brains of elderly individuals with no indication of cognitive impairment prior to death [227,228]; and (f) one-sided injury or removal of any given cerebral cortex area does not abolish conscious thinking [229]; moreover, often, higher-order cognition in its core remains generally quite robust, even after extensive and bilateral focal brain damage [220]. Considering all of the above, it seems that a structurally based brain age measure cannot capture the full complexity of the BBA. In this respect, the quantitative electroencephalogram (qEEG)-based BBA could be a more suitable, rather simple, and promising measure of brain aging (an electroencephalogram (EEG) is a summation of the electrical activities along the scalp generated by the firing of nerve cells (i.e., neurons) in the brain [230]. The aggregate of these electric voltage fields creates an electrical reading, which electrodes on the scalp are able to detect and record [231]. qEEG (quantitative EEG) is a digitally recorded and mathematically/algorithmically/statistically analyzed EEG [232]). This is so because qEEG, in addition to being relatively cheap, portable, nonstressful, and noninvasive, has a number of useful and important characteristics or properties, most of which are age-related or age-dependent (Figure 1):(a)It constitutes a neural trait measure due to the fact of its high specificity (i.e., the extent to which an qEEG pattern is uniquely associated with a given person) and intra-individual high stability (test–retest reliability) [233,234,235,236,237];(b)qEEG is highly heritable and, thus, likely to be under strong genetic control [234,238,239,240];(c)It reflects both the brain’s structural characteristics (or “hardware”) such as the number of connections between neurons, fiber density, axonal diameter, degree of myelination and white matter integrity, as well as the integrity of the corticocortical and thalamocortical circuits, hippocampal volume (the hippocampus is a brain region central to both healthy memory function and also age-related memory decline [241]), number of active synapses in thalamic nuclei, brain hemodynamics and metabolism, and the number of potential neural pathways [231,242,243,244] and cognitive processes and functions (“neuropsychological competence” or “software”), such as memory performance, attention and processing speed, individual capacity for information processing (the capacity for storage, transfer, and retrieval of information) and cognitive preparedness (the brain’s capacity for higher-level cognitive functioning), network efficiency, and neural compensation at all ages, both in healthy individuals and in individuals with neurological conditions [245,246,247,248];(d)qEEG possesses age-related changes in both brain structural and functional integrity (in)dependently of pathology [245,249,250,251,252], thus directly reflecting an aging process;(e)It shows age-dependent changes that parallel neurological changes in typical aging [253]; indeed, it is known that, for example, atrophic brain regions detected in patients with dementia largely overlap with regions showing normal age-dependent decline in healthy individuals [254];(f)qEEG is associated with age-related conditions, such as cognitive decline, Alzheimer’s disease, mild cognitive impairment, vascular dementia, other dementias, multiple sclerosis, and cerebral tumors [244,255,256,257]. Capitalizing on these facts, we could conclude that the dualism of the brain’s anatomical (i.e., structural) and cognitive (i.e., functional) reserves can be unified within a single concept—brain resources (BR), which can be measured by qEEG. Thus, qEEG-based BBA can be considered a proxy for the BA of the brain. In this context, a person with high BR (brain reserve (“hardware”) + cognitive reserve (“software”)) (the brain reserve is a “passive” form of capacity that is dependent on the structural properties of the brain, such as a higher number of healthy synapses and neurons [258]. In this context, as brain volume or synaptic density decreases with age, individuals with more premorbid brain reserve will manifest symptoms later in life and less severely than individuals with less premorbid brain reserve—a compression of morbidity that improves quality of life [22]. On the contrary, cognitive reserve describes an “active” function of the brain that involves cognitive operations and representations [258] and refers to the ability to use alternative functions when a default function is rendered inoperable or to the robustness of a particular cognitive function against brain age-related pathologies (see also [259,260]). For example, it has been documented that elderly individuals with a lower cognitive reserve need to over-recruit neuronal networks (due to the lower efficiency and decreased structural properties of their neuronal networks), exhibiting less efficient brain functioning, to achieve the same level of cognitive performance as elderly individuals with a higher cognitive reserve [261]. Moreover, elderly individuals need higher activation of their neuronal networks than young individuals, for the same reason—lower efficiency and decreased structural properties of the elderly subjects’ neuronal networks [262] (see also [263])) has a younger brain phenotype (qEEG-based BBA) and is more likely to remain within normal (healthy) limits for a longer period of time [209,264]. Conversely, a person with fewer BR has an older brain phenotype (qEEG-based BBA). Indeed, it has been shown that an individual’s brain age can be reliably estimated from qEEG [137,249,250,265,266], and qEEG-derived increased BBA is associated with neurological and psychiatric diseases, diabetes, and hypertension [266], as well as reduced life expectancy and increased mortality risk in comorbidities, such as cardiovascular dysfunction, current smoking status, and increased body mass index [137]. Thus, qEEG-based BBA is a practical, simple, and compelling indication of the BA as opposed to the CA of the brain. It measures the full complexity of brain aging and age-related risks [137,266]. This justifies the use of such qEEG-based BBA to estimate the effectiveness of putative interventions aiming to ameliorate brain aging at a practical (i.e., limited) timescale. ## 1.2. Choosing a Brain Anti-Aging Intervention The most promising strategy to tackle aging as a whole is by targeting the epigenetic regulators associated with the aging process [34,86,267,268]. The same also applies to brain aging, since identical aging mechanisms are involved, and, as we discussed above, the brain is at the center of organismal processes and functions [172,173]. In this regard, there is growing evidence that the very same interventions that target epigenetic regulators across differently aged tissues have a concomitant anti-aging effect on the brain [170,267,269,270,271,272]. Currently, the most accessible anti-aging interventions that work through epigenetic regulation are physical exercise [90,272,273,274,275,276], diet strategies (for example, caloric restriction and intermittent fasting [90,271,275,276,277,278]) and nutritional supplementation (e.g., vitamins and macro- and micro-elements) [90,268,279,280,281,282,283,284]. Among these strategies, nutraceutical supplements, which are compounds of vitamins, minerals, and essential amino- and fatty acids, as well as plant extract isolates [21,282], may have further advantages (Figure 1): they (i) are widely available and commonly used; (ii) they affect a highly evolutionarily conserved nutrient-sensing pathway (this pathway regulates several key homeostatic processes, including autophagy, mRNA translation, and metabolism, each of which affects the hallmarks of aging [13,34] and, consequently, the lifespan [285,286]) linked to aging [287,288]; (iii) could prevent or slow the progression of a wide variety of illnesses [90,283,284], including neurodegeneration [289,290,291]; (iv) can affect the central circadian clock in the brain via sirtuins [134,292], which are linked to the regulation of aging [9,177,293,294]; and (v) do not require as much effort to comply with recommendations, for example, committing to regular physical exercise [295,296,297] or maintaining a rigorous diet [298,299,300,301]. Moreover, considering that many nutraceutical compounds are mimetics of calorie restriction [302] or physical exercise [303], manipulating the dosage of such compounds could achieve stronger and faster results. As a consequence, it is plausible to hypothesize that an individually tailored (the strategy of using personalized interventions to meet individual health needs as opposed to a “one-size-fits-all” approach has been recently proposed by Fahy et al. [ 268] and has shown encouraging results [268] (see also [304]). The need for the personalization of anti-aging interventions has also been recently reiterated [90,305]) program of nutraceutical compounds may delay or even reverse the BA of the brain, thus increasing the healthspan (the period spent free of chronic disease [306]) and lifespan (the period spent alive [307]) by targeting and manipulating multiple biological pathways that cause aging [34,308]. Furthermore, we expect this approach to be more efficient than lifestyle changes. ## 1.3. Aim of the Study Therefore, the aim of the present study was to examine whether an individually tailored program of multiple nutraceutical compounds can (a) increase BR (measured by qEEG), thus establishing a younger brain phenotype (younger qEEG-derived BBA), to return the normotonic older brain to a level more comparable to a younger brain (i.e., rejuvenation), and/or (b) slowdown the speed of aging of the brain (i.e., deceleration) in a cohort of “normal” adults. The lifestyle change group served as an active control. ## 2.1. Participants The participants’ EEG, clinical/medical, and demographic data were extracted for the retrospective analysis from the electronic record registry of BM-Science ($$n = 1$.010$ on the day of the study onset; the period for the data extraction was between 2013 and 2020). Subjects in this registry (initial cases) during this period were self-selected to receive well-being guidance (other cases in the registry are either participants from previous studies or were referred by doctors for neurophysiologic evaluations). The participants’ data were entered into the study in consecutive order as they met the inclusion criteria until a total of at least 40 individuals in each group (experimental and active control) was obtained in order to have sufficient statistical power ($80\%$) to detect the interventions’ effects. After the inclusion/exclusion criteria were met, the data of 42 (31 females; mean age: 54.1 ± 13 years) and 47 (25 females; mean age: 45.2 ± 7.3 years) participants (for the experimental and control groups, respectively) were included in the analysis. The inclusion criteria were male and female volunteers, aged 25 and above, self-selected to receive either a nutraceutical compounds program (experimental group) or lifestyle recommendation (active control group), able to follow the intervention for 6 to 18 months, availability of complete pre- and postintervention data, and signed informed consent. The exclusion criteria were: malignancies as suggested by personal medical history, treatment-resistant significant bradycardia (<55 bpm) or hypertension (systolic > 160 mmHg or diastolic > 90 mmHg), allergy/sensitivity to the studied nutraceutical compounds, alcoholism or drug addiction, a diagnosis of schizophrenia, epilepsy, Alzheimer’s disease or Parkinson disease, and no signed informed consent (the presence of various health complaints and different comorbidities was not qualified as an exclusion criterion for pragmatic reasons so that the study sample was more representative of the general population of “practically” healthy persons, where various health issues are commonly experienced). The demographic and clinical data, as well as baseline values of BBA and brain resources, are presented in Table 1. This retrospective study can be considered as single-blind because the participants were blinded to the interventions’ primary output related to the qEEG-derived BBA (the participants thought that the respected interventions aimed to improve their general well-being). This study was carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and the standards established by the Review Board of BM-Science—Brain and Mind Technologies Research Centre. Originally, prior to the EEG scanning and interventions, the experimental procedures were explained, and participants signed an informed consent form. The use of the data for scientific studies was authorized by the written informed consent of the subjects and approval by the Review Board of BM-Science—Brain and Mind Technologies Research Centre. ## 2.2. EEG Recording and Acquisition Ongoing EEG activity was recorded (using a digital EEG recording system—Mitsar) late in the morning to minimize drowsiness in a quiet and dimly lit room for at least 6 min while subjects were seated on a comfortable half-reclining armchair with their eyes closed. The subjects were asked to have a moderate breakfast and refrain from the consumption of psychoactive drugs (e.g., antidepressants and benzodiazepines) and other psychostimulants (e.g., coffee, tea, and alcohol) at the morning of the recording day. During the EEG recording, the subjects were requested to remain in a standard resting state condition (the resting-state qEEG manifests the baseline mechanics of self-organization that regulate multiple brain systems, adapting the brain and body to an ever-changing environment [309,310]. Thus, the resting-state qEEG reflects the intrinsic default activity that instantiates the maintenance of information for interpreting, responding to, and predicting environmental (internal and external) demands [247,311,312,313,314]). In this condition, they had to keep their muscles relaxed without any movements/talking and to stay awake, with their mind freely wandering without systematic goal-oriented mentalization. The following parameters of the EEG recording were enforced: (i) 19 scalp locations (i.e., O1, O2, P3, P4, Pz, C3, C4, Cz, T3, T4, T5, T6, Fz, F3, F4, F7, F8, Fp1, and Fp2) according to the International 10–20 System of the EEG electrode placement; (ii) 256 Hz sampling rate; (iii) monopolar montage with linked earlobes as a reference electrode; (iv) 0.5–30 Hz bandpass; (v) 50 Hz notch filter ON; (vi) electrooculogram (0.5–70 Hz bandpass); and (vii) impedance below 10 kΩ. Throughout the EEG recording, the experimenter monitored the participant’s state and ongoing EEG traces to assist the subject in maintaining an adequate level of vigilance (i.e., avoiding drowsiness and sleep onset). Artifact removal was performed by visual inspection of the raw EEG data, augmented by a computerized artifact detection and rejection algorithm (for details, see [315], p. 7). Artifact-free EEG data were subjected to a computerized analysis to estimate the BBA and BR. ## 2.3. Estimation of Cerebral Physiological Age as a Proxy of the Brain’s BA—BBA Briefly, the qEEG-based BBA was estimated using an established linear regression model that has previously been published and described in detail in [250]. The choice of regression as a method of analysis is defined by the continuous process of brain aging, which manifests itself in the gradual accumulation of age-related effects without clear leaps or stages due to the fact of various aging trajectories of the different functional and structural parameters [316,317,318,319]. The regression analysis resulted in a linear dependence between “age-specific” qEEG changes and CA (for details, see [250]). This linear regression model was used to estimate an individual’s BBA based on the qEEG data and calibrated to current data from the BM-Science registry. In short, the EEG time series were first divided into successive and overlapping 2-sec segments, which were windowed, Fourier transformed, and averaged to produce one power spectrum per recording site. Then, the age-dependent EEG feature based on alpha frequency (7–13 Hz) was extracted and averaged across selected EEG electrodes [249,250]. Since brain aging reflects gradual changes in the structure and function of the brain that occur over time and do not result from disease or other gross accidents, the brain’s aging can match the CA (i.e., normal healthy aging) or it can be delayed (i.e., deceleration—negative values of the BBA), facilitated (i.e., acceleration—positive values of the BBA), or reversed (i.e., rejuvenation) [169,205,209,210]. To capture all these conditions, the qEEG-based BBA was estimated at two time-points: the 1st visit—the baseline acquisition (pre-intervention) and the 2nd visit—the follow-up acquisition (postintervention) after 13 months (on average) of interventions. Comparing the 1st and 2nd visit BBAs, it was possible to evaluate the rate of aging (deceleration or acceleration) and direction (healthy aging vs. rejuvenation) in both groups (experimental and active control). The difference between the estimated BBA and CA normalized to the CA ranged between 16 and 100 years indicates the individual’s BR (the low boundary of 16 years was taken, because around this time-point, the maturation of the EEG characteristics (i.e., when the EEG patterns become very similar to the mature waveforms of the adult EEG [320,321]) and most brain areas [322] is completed; these are paralleled by the substitution of organismal growth and maturation with the beginning of biological aging on different levels of the organism [45,323]. A 100-year limit was taken as the potential maximum, which is actually rarely reached by humans). Values “around 0” indicate that the brain’s resources are in line with those typical for the individual’s CA (i.e., healthy aging); “negative values” indicate fewer brain resources for a given CA—the brain has “overspent” resources characteristic of healthy individuals of an older age—an older brain phenotype; “positive values” indicate more brain resources for a given CA—the brain has preserved resources characteristic of healthy individuals of a younger age—a younger brain phenotype. ## 2.4. Interventions The experimental group used an individually tailored program of nutraceutical compounds for 6–18 months (mean: 13 ± 1.13 months). Individual adjustment of the program was based on the qEEG characteristics that deviated from normative values [324,325,326], prenatal and postnatal data, medical history, personal complaints and existing symptoms, medication used, psychometrics (i.e., scores for depression [327], anxiety [328,329], neuroticism [330]), environmental conditions (stress presence), and life habits (alcohol consumption, smoking, and exercising). A tailored program of nutraceutical compounds with documented mechanistic activity on epigenetic pathways [283,284] included probiotics, vitamins, minerals, polyphenols, and omega-3 fatty acids grouped in sets (to maximize the synergetic effect and minimize the potential opposing effects of the compounds) that were timed throughout the day to align with the circadian rhythm and eating time [331], and the month to also be in keeping with the circannual rhythm—the annual variability of physiological processes [332]. While the exact number of compounds, the frequency of their intake during the day and also per month, as well as the dosages, varied for every participant (based on the criteria described above), the overlapping compounds included vitamin C, vitamin D, vitamin A, vitamin(s) B, omega-3, Mg, Zn, alpha-lipoic acid, CoQ-10, Bifidobacterium, and lactobacillus. All participants were asked to take the supplements on a daily basis in accordance with the program. The active control group used a tailored lifestyle recommendation program over 6–18 months (mean: 13.5 ± 1.10 months), since research also suggests that positive health habits may be able to offset earlier deleterious influences [61,62] and even reverse aging [90,278,333]. Individual adjustment of the lifestyle recommendations was conducted using the same criteria as for the experimental group. Tailored lifestyle interventions included dietary recommendations (plant- and fish-centered; low caloric intake; low carbohydrates; and fasting-mimicking), physical exercise (aerobic: cycling, walking, swimming, and jumping; resistance; sustained isometric nonmaximal voluntary contraction; up to 30 min per day and 3–7 days per week), and sleep of 7–8 h per night. The participants were requested to follow these recommendations daily. Table 1 \| Characteristics \| Nutraceuticals \| Lifestyle \| p-Value \| Test Type \| \| --- \| --- \| --- \| --- \| --- \| \| Sample size (N) \| 42 \| 47 \| Not applicable \| Not applicable \| \| Sex (% of females) \| 73.8 \| 53.2 \| 0.00204 \| Chi-square \| \| Chronological age—CA (mean/st.d) \| 54.1 (13) \| 45.2 (7.3) \| 0.00048 \| Mann–Whitney U test \| \| Brain biological age—BBA (mean/st.d) \| 46.3 (11) \| 37.7 (9.8) \| 0.00042 \| Mann–Whitney U test \| \| Brain resources—BR (mean %/st.d) \| 9.89 (20) \| 8.99 (13) \| Not significant \| Mann–Whitney U test \| \| Healthy lifestyle habits (% of those who have) \| 16.7 \| 12.8 \| Not significant \| Chi-square \| \| Current health symptoms (% of those who have) \| 33.3 \| 40.2 \| Not significant \| Chi-square \| \| Past health problems (% of those who had) \| 64.3 \| 57.4 \| Not significant \| Chi-square \| \| Relatives with mind/brain disorders (% of those who have) \| 16.7 \| 23.4 \| Not significant \| Chi-square \| \| Anxiety—Beck 1 (mean/st.d) \| 8.2 (7.1) \| 7.9 (6.5) \| Not significant \| Mann–Whitney U test \| \| Anxiety—Ham 2 (mean/st.d) \| 8.7 (6.6) \| 8.6 (5.5) \| Not significant \| Mann–Whitney U test \| \| Depression—Beck 3 (mean/st.d) \| 6.2 (6.7) \| 6.5 (4.8) \| Not significant \| Mann–Whitney U test \| \| Big-5—neuroticism 4 (mean/st.d) \| 2.8 (0.8) \| 2.9 (0.7) \| Not significant \| Mann–Whitney U test \| \| Handedness (% of right-handed) \| 83.3 \| 87.2 \| Not significant \| Chi-square \| \| Marital status (% of married) \| 73.8 \| 83 \| Not significant \| Chi-square \| \| Marital status (% of divorced) \| 9.5 \| 12.7 \| Not significant \| Chi-square \| \| Marital status (% of single) \| 16.7 \| 4.3 \| 0.002712 \| Chi-square \| \| Education (% of those who have a PhD) \| 14.3 \| 10.6 \| Not significant \| Chi-square \| \| Education (% of those who graduated from university or institute) \| 69 \| 74.4 \| Not significant \| Chi-square \| \| Education (% of those who completed high school (≥11–12 years)) \| 16.7 \| 15 \| Not significant \| Chi-square \| \| Job (% of directors or CEOs) \| 21.4 \| 17 \| Not significant \| Chi-square \| \| Job (% of senior managers) \| 38.1 \| 38.3 \| Not significant \| Chi-square \| \| Job (% of junior managers) \| 35.7 \| 38.3 \| Not significant \| Chi-square \| \| Job (% of students or trainees) \| 4.8 \| 6.4 \| Not significant \| Chi-square \| \| Number of interests or hobbies (mean/st.d) \| 4.3 (1.8) \| 3.6 (1.6) \| 0.0394 \| Mann–Whitney U test \| \| Smoking (% of those who smoke) \| 7.1 \| 2.1 \| Not significant \| Chi-square \| \| Alcohol consumption (1–2 drinks * per week; %) \| 40.5 \| 40.4 \| Not significant \| Chi-square \| \| Alcohol consumption (3–4 drinks per week; %) \| 47.6 \| 38.3 \| Not significant \| Chi-square \| \| Alcohol consumption (5–7 drinks per week; %) \| 7.1 \| 8.5 \| Not significant \| Chi-square \| \| Alcohol consumption (8–10 drinks per week; %) \| 4.8 \| 12.8 \| 0.04808 \| Chi-square \| Both interventions (nutraceutical compounds and lifestyle recommendations), as used in the present study, are generally considered safe, even when used for a long time [90]. Adherence to the interventions was verified by phone or email communication with the participants. We hypothesized that if the tailored program of nutraceutical compounds had a specific advantageous effect on BBA that went beyond the effects of the lifestyle changes, then (a) it should not only slowdown (i.e., deceleration) or reverse (i.e., rejuvenation) the brain’s aging, thus improving the BR, but (b) the magnitude of this effect should also be larger than in the control group that used lifestyle recommendations. ## 2.5. Statistical Analyses In order to compare the longitudinal changes in the BBA and BR scores between the pre- and postintervention endpoints within the same group, the Wilcoxon signed-rank test was employed. Comparisons between the experimental and control groups were performed using the Mann–Whitney U test and the chi-square test (for demographic characteristics). Additionally, we examined differences in the BBAs with respect to the interventions separately for (i) females and males, as well as for (ii) participants with a baseline (pre-intervention) BBA younger and older than their CA. The reported p-values were not corrected for multiple comparisons because all significant test results were highly correlated, making a Bonferroni correction overly conservative and, thus, inappropriate [334,335]. ## 3.1. Demographic Characteristics A group comparison of the demographic and psychometric characteristics is shown in Table 1. The experimental (nutraceuticals) and control (lifestyle) groups differed in respect to a number of demographic variables: sex, CA, BBA, marital status—% of singles, number of interests and hobbies, and alcohol consumption—% of those who have 8–10 drinks per week. There was no difference between the two groups for the remaining (majority) characteristics (Table 1). Despite the fact that CA and BBA differed between the groups, the BR was nearly identical—this is important for the purpose of the present study, since the qEEG-derived BR score, which is a proxy for the brain’s neurophenomenological condition, was on average identical at the baseline (pre-intervention) time-point, thus ensuring an equal starting point for the participants in both groups (Table 1). ## 3.2. Neurophysiological Findings: BBA and BR The findings of this study show that although, on average, the BBA was significantly younger than the CA at baseline (pre-intervention) for both groups (Wilcoxon signed-rank test: z = −2.72, $$p \leq 0.00652$$ for the nutraceuticals group; z = −3.98, $$p \leq 0.00006$$ for the lifestyle group), and both groups had increased BR (+$9.89\%$ for the nutraceuticals group; +$8.99\%$ for the lifestyle group); the BBA nevertheless significantly decreased and BR significantly increased (+$14.16\%$) as a result of the intervention (post-endpoint) only in the experimental/nutraceuticals group (Wilcoxon signed-rank test: z = −2.27, $$p \leq 0.0232$$ for BBA; z = −3.15, $$p \leq 0.00164$$ for BR) (Figure 2). On the contrary, in the control/lifestyle group, BBA and BR did not show a significant change postintervention (Wilcoxon signed-rank test: z = −0.42, $$p \leq 0.67448$$ for BBA; z = −1.48, $$p \leq 0.13622$$ for BR) (Figure 2). At the same time, on average, the BBA continued to be significantly younger in comparison with the CA at the postintervention endpoint of both groups (Wilcoxon signed-rank test: z = −4.12, $$p \leq 0.00001$$ for the nutraceuticals group; z = −4.07, $$p \leq 0.00001$$ for the lifestyle group) (Figure 2). On average, the decrease in the BBA in comparison to the CA (=BBA-CA) was −7.86 years for the experimental/nutraceuticals group and −7.49 years for the control/lifestyle group at the pre-intervention point and −11.8 years and −8.62 years, respectively, postintervention (Figure 3A). While there was no statistically significant difference between these values for the two groups at the pre-intervention point (Mann–Whitney U test: $z = 0.36$, $$p \leq 0.71884$$), postintervention the groups did differ significantly (Mann–Whitney U test: $z = 1.91$, $$p \leq 0.04961$$) due to the significant widening of the difference between BBA and CA in the experimental/nutraceuticals group (Wilcoxon signed-tank test: z = −3.43, $$p \leq 0.0006$$), and no significant difference between BBA and CA in the control/lifestyle group (Wilcoxon signed-rank test: z = −1.67, $$p \leq 0.09492$$) (Figure 3A). Furthermore, the BBA of the participants in the experimental/nutraceuticals group was, on average, 2.83 years younger at the endpoint of the intervention compared to the same individuals at the beginning. The BBA of the control/lifestyle participants was, on average, only 0.02 years younger compared to the baseline at the end of the intervention; this difference between the groups was statistically significant (Mann–Whitney U test: z = −3.98, $$p \leq 0.00006$$) (Figure 3B). As expected, the average CA values in both groups increased as a function of the follow-up time: approximately +1.1 years for both groups, without a statistical difference between the groups (Mann–Whitney U test: z = −0.89, $$p \leq 0.36812$$) (Figure 3B). Because the results above represent the average values for all participants in each group, they may not accurately capture the impact of the interventions on the different sexes or those whose BBA was either older or younger than their CA at the baseline (pre-intervention) point. Thus, sex, as well as baseline BBA, may be potential confounding covariates of the overall results. Therefore, we conducted separate stratification analyses based on “sex” and the “BBA pre-intervention score”. The stratification analyses revealed the following results (Figure 4). The BBA of females in the experimental/nutraceuticals group ($$n = 31$$) scored, on average, 2.98 years younger at the endpoint of intervention compared to the baseline. The BBA of females in the control/lifestyle group ($$n = 25$$) scored, on average, 0.19 years older at the end of the intervention compared with the baseline; this difference between the groups was statistically significant (Mann–Whitney U test: z = −2.02, $$p \leq 0.04338$$) (Figure 4). The CA became older in both groups at the endpoint of the interventions: on average, +1.08 years for the experimental/nutraceuticals group and +1.19 years for the control/lifestyle group, without a statistical difference between the groups (Mann–Whitney U test: $z = 0.35$, $$p \leq 0.72634$$) (Figure 4). For the male participants, the results were slightly different. The BBA of males in the experimental/nutraceuticals group ($$n = 10$$) was, on average, 2.31 years younger at the endpoint of the intervention compared to the baseline. The BBA of males in the control/lifestyle group ($$n = 22$$) was, on average, 0.26 years younger at the end of the intervention when compared with the baseline; this difference between the groups, however, did not reach statistical significance (Mann–Whitney U test: $z = 0.42$, $$p \leq 0.6672$$) (Figure 4). The CA became older in both groups at the endpoint of the interventions: on average, +1.28 years for the experimental/nutraceuticals group and +1.02 years for the control/lifestyle group, without a statistical difference between the groups (Mann–Whitney U test: z = −1.12, $$p \leq 0.26272$$) (Figure 4). For the participants whose pre-intervention BBA was older than their CA, the BBA in the experimental/nutraceuticals group ($$n = 15$$) was, on average, 6.77 years younger at the endpoint of the intervention compared to the baseline. For the control/lifestyle group ($$n = 13$$), the BBA was, on average. 0.25 years older at the end of the intervention when compared with the baseline; this difference between the groups was statistically significant (Mann–Whitney U test: z = −2.83, $$p \leq 0.00466$$) (Figure 5). The CA became older in both groups at the endpoint of the interventions: on average, +1.22 years for the experimental/nutraceuticals group and +0.86 years for the control/lifestyle group, without a statistical difference between the groups (Mann–Whitney U test: $z = 1.05$, $$p \leq 0.28914$$) (Figure 5). For the participants whose pre-intervention BBA was younger than their CA, the BBA in the experimental/nutraceuticals group ($$n = 27$$) was, on average, 0.64 years younger at the endpoint of the intervention compared to the baseline. For the control/lifestyle group ($$n = 34$$), the BBA was, on average, 0.13 years younger at the end of the intervention when compared with the baseline; this difference between the groups, however, did not reach statistical significance (Mann–Whitney U Test: z = −0.43, $$p \leq 0.65994$$) (Figure 5). The CA became older in both groups at the endpoint of the interventions: on average, +1.04 years for the experimental/nutraceuticals group and +1.21 years for the control/lifestyle group, without a statistical difference between the groups (Mann–Whitney U Test: z = −0.37, $$p \leq 0.70394$$) (Figure 5). In order to analyze the potential factors that may be associated with the pre-intervention BBA, we pooled together the demographic and clinical data from both groups and then stratified all participants into two subgroups: BBA > CA and BBA < CA at baseline. The result is presented in Table 2. Some differences between subgroups were expected because they themselves were the basis of the stratification (BBA and related to it BR), while in others they arose originally. The BBA < CA subgroup was characterized by a statistically significant smaller number of right-handed, single, and smoking participants with a total education of high school and a statistically significant higher number of participants who were married, had a PhD, had more hobbies and interests, and consumed more alcohol per week when compared to the BBA > CA subgroup (Table 2). Since the duration of the interventions varied between 6 and 18 months, it was interesting to see if the changes in the BBA scores were associated with the duration of the interventions. The correlation analysis did not reveal a significant correlation for either group: experimental/nutraceuticals: $r = 0.25$, $$p \leq 0.110319$$ (Pearson correlation test); control/lifestyle: $r = 0.21$, $$p \leq 0.156549$$ (Pearson correlation test). ## 3.3. Psychometrics and Health Symptoms While, on average, the experimental/nutraceuticals and control/lifestyle groups did not differ significantly in the scores for depression, anxiety, and neuroticism at the pre-intervention time-point (see Table 1), the postintervention scores for depression and anxiety decreased significantly in both groups as a function of the intervention (Figure 6; experimental/nutraceuticals—Wilcoxon signed-rank test: z = −3.19, $$p \leq 0.00138$$ (anxiety—Beck); Wilcoxon signed-rank test: z = −4.29, $$p \leq 0.00001$$ (anxiety—Ham); Wilcoxon signed-rank test: z = −2.92, $$p \leq 0.0035$$ (depression—Beck). Control/lifestyle—Wilcoxon signed-rank test: z = −2.13, $$p \leq 0.03318$$ (anxiety—Beck); Wilcoxon signed-rank test: z = −2.68, $$p \leq 0.00736$$ (anxiety—Ham); Wilcoxon signed-rank test: z = −2.95, $$p \leq 0.00318$$ (depression—Beck)). Compared to the control/lifestyle group, the magnitude of the significance was larger in the experimental/nutraceuticals group for the anxiety scores measured by both the Beck and Ham tests (Figure 6). At the same time, the postintervention scores for depression and anxiety did not differ significantly between the groups (anxiety—Beck: Mann–Whitney U test: $z = 0.83$, $$p \leq 0.4009$$; anxiety—Ham: Mann–Whitney U test: $z = 0.83$, $$p \leq 0.4009$$; depression—Beck: Mann–Whitney U test: $z = 1.38$, $$p \leq 0.1645$$). The estimation of the current health symptoms (Table 3) revealed a comparable percentage of participants who experienced them in both groups at the pre-intervention time-point (Chi-square statistic = 1.0571, $$p \leq 0.303887$$). However, postintervention, only the experimental/nutraceuticals group had a significant decrease in the percentage of the participants who experienced current health symptoms when compared with the baseline (Chi-square statistic = 25.4711, $$p \leq 0.00001$$). In the control/lifestyle group, the decrease was small and nonsignificant (Table 2; Chi-square statistic = 1.3889, $$p \leq 0.238593$$). ## 4. Discussion The goal of the present study was to demonstrate the slowing down or even reversal of the brain BA by means of safe and accessible interventions (nutraceutical supplementation vs. lifestyle changes) in order to ameliorate brain aging at a practical (limited) timescale (Figure 1). The obtained results, while limited, support the feasibility of this goal and also provide the first robust evidence that the regression of brain aging is indeed possible in humans. Compared to lifestyle changes, the intervention involving nutraceutical supplementation was efficient in significantly reducing (i.e., reversing) BBA and enhancing BR at the end of the 13-month (on average, the minimum was 6 months and the maximum was 18 months) program (Figure 2). In contrast, the lifestyle intervention was able to only slow down the BBA and stabilize the BR, keeping them at the same rate as before the intervention (Figure 2) despite the increase in CA. The BBA was 11.8 years younger than the CA in the nutraceuticals group at the end of the intervention (such a difference between biological and chronological ages is comparable with the differences reported in previous studies: 12 years [336], 12.6 years for Hannum’s epiclock, and 17.5 years for Levine’s epiclock [268]; 15.3 years for females and 16.7 years for males [39]). This difference was significantly larger than at the beginning of the intervention (Figure 3A). For the lifestyle group, the BBA was 8.62 years younger than the CA at the end of the study, although this difference was not significantly different from the beginning of the study (Figure 3A). The BBA of the participants in the nutraceuticals group was 2.83 years younger at the endpoint of the intervention compared with the baseline BBA (again, such a rate of reversal in the BA is comparable with reported rates in previous studies: 2.5 years [268] and 1.96 years [90]), while the BBA of the lifestyle participants was essentially unchanged, measuring only a few days younger compared to the baseline (Figure 3B). Together, these findings provide substantial evidence that nutraceutical compounds (vitamins, minerals, and essential amino- and fatty acids, as well as plant extract isolates, such as polyphenols [21,282])—when used in specific combinations and adjusted individually—may reverse BBA and increase BR. While the exact mechanisms involved are not clear, one may speculate that different nutraceutical compounds probably have unique and often small effects that are in opposition to brain aging, and when combined in an individually adjusted fashion, these compounds activate a broad enough range of synergistically interacting metabolic pathways that then restore brain resources and reverse brain biological aging. This suggestion is consistent with the known ability of nutraceuticals to affect a highly evolutionarily conserved nutrient-sensing pathway linked to aging [287,288,337] and lifespan [285,286]; prevent or slow the progression of a wide variety of illnesses [90,283,284], including neurodegeneration [289,290,291,337]; improve cerebral blood flow and antioxidant capacity [338,339]; and, additionally, affect the central circadian clock in the brain via sirtuins [134,292], which are also linked to the regulation of aging [9,177,293,294]. In this regard, as has been proposed by Nur et al. [ 284], nutraceuticals could even be considered “epidrugs”. Indeed, for example, in addition to its role as a cellular antioxidant [339], vitamin C is a critical epigenome remodeler that ameliorates epigenome dysregulation (by enhancing the activity of Jumonji-C domain-containing histone demethylases (JHDMs) and ten-eleven translocation (TET), which drive histone and DNA demethylation) and restores the youthful state of cells (additionally, vitamin C can also target α-ketoglutarate-dependent dioxygenases (α-KGDDs), which are essential in regulating metabolism, DNA repair, and DNA/RNA demethylation and plays an important role in fine-tuning the reprogramming stages of youthful states of cells [340]) [341]. TETs are highly expressed in the brain [342,343], with TET1 and TET3 involved in proper brain and cognitive function [103,344,345], while TET2 is associated with neurogenic processes by restoring adult neurogenesis to youthful levels and, thus, enhancing cognitive function [267] (neurogenesis is a process of generating new functional neurons in the brain [346]. For a long time, it was thought that the loss of neurons was irreversible in the adult brain because dying neurons cannot be replaced; however, later it was demonstrated that life-long continuous neurogenesis takes place in almost all mammals, including humans [347]). Vitamin A works synergistically with vitamin C by stimulating TET expression [280] (for the role of other vitamins in epigenetic modification, see Nur et al. [ 284], and for the effects of vitamins, polyphenols, and minerals on the cells’ homeostasis, senescence, telomere length, and counteraction of DNA damage, see Proshkina et al. [ 283]). Another vitamin (vitamin D) may stimulate the production of neurotrophic, antioxidative, and anti-inflammatory factors; reduce risk of cerebrovascular (as well as cardiovascular) diseases; and even influence amyloid phagocytosis and clearance (it is known that the aging brain is vulnerable to inflammation, where the circulating proinflammatory factors can promote cognitive decline and are responsible for the loss of macrophages’ and microglia’s ability to clear misfolded proteins in the brain, which are associated with neurodegeneration, dementia, and Alzheimer’s disease [348]) [349]. Furthermore, a high level of vitamin D is associated with the reduced degeneration of major brain white matter tracts, even in cognitively healthy elderly individuals [349]. Additionally, vitamin D happens to upregulate αKlotho (KL) transcription [350]. KL is a protein that is mainly expressed in the brain and also the kidneys [351]; it has strong anti-inflammatory and neuroprotective properties, making this protein a key factor for health and longevity [78]. Interestingly, some polyphenols have a synergetic effect, making it easier for vitamin D to upregulate KL gene expression [352]. Furthermore, the mammalian target of the rapamycin (mTOR) pathway, which detects high amino acid concentrations, is one of the hallmarks of aging [34]. Its overactivation promotes aging and decreases lifespan (for a review, see [353]), while its suppression is associated with an increase in lifespan (importantly, lifespan extension is comparable if the anti-aging intervention is initiated at a young age, middle age, or in late life [354]) [355]. In the brain, upregulated mTOR signaling has been associated with amyloid accumulation and, conversely, downregulated mTOR signaling is associated with reduced amyloid levels [356]. In addition, higher levels of mTOR activation—alongside its downstream effectors—were found in brain regions that were affected by Alzheimer’s disease or mild cognitive impairment [357,358]. Therefore, the inhibition of mTOR is desirable. A number of nutraceutical compounds can do this: vitamin D [359], curcumin [360], EGCG—green tea component [361], omega-3 [362], and alpha-lipoic acid [363]. Another important regulator of aging is adenosine monophosphate-activated protein kinase (AMPK), the increased activity of which is related to an extended lifespan [364]. Studies indicate that the responsiveness of AMPK signaling steadily declines with age [365,366]. AMPK activation in the brain is responsible for neuroprotection through the induction of autophagy, angiogenesis, and neurogenesis [337,367]. It has been demonstrated that some polyphenols with antioxidant and anti-inflammatory properties [368] can activate silent information regulator 1 (SIRT1), which belongs to the Sirtuin family [369] and the activation of which can stimulate the activation of AMPK (interestingly, AMPK activation may restimulate the functional activity of SIRT1 [370], thus resulting in a positive feedback loop between SIRT1 and AMPK, which, in turn, can potentiate the function of the other AMPK-activated signaling pathways important for healthspan in general [364] and the brain in particular [78]) [371], thus providing anti-aging effects in the brain (polyphenols such as resveratrol easily cross the blood–brain barrier (BBB) to express their effects in the brain [372,373,374]) [78,375,376]. SIRT1 also has another path to affect brain aging: regulation of the central circadian clock [292,377]. Apparently, the loss of SIRT1 in the brain not only dysregulates the circadian clock but also accelerates the aging process [134,294] (such acceleration is most likely mediated by NAD+ [134]. Indeed, an age-dependent reduction in the levels of NAD+ in the brain was reported in healthy individuals [179], as well as in accelerated brain aging [180]. Furthermore, considering that the circadian clock regulates the oscillatory dynamics of NAD+ levels [378] and that this clock is dysregulated in the aging brain [377], a decline in NAD+ levels over a person’s lifespan may be attributed to the loss of circadian clock function [134]. A deficiency in NAD+ can be restored by vitamin B3 (and its derivatives) supplementation [379]). On the contrary, the upregulation of SIRT1 in the brain results in an increase in lifespan [181]. Moreover, the antioxidant carotenoid astaxanthin, especially when combined with folic acid, selenium, zinc, and omega-3, can reduce the degree of hypermethylation [282], which normally shows a robust and progressive rise during CA in the brain [380], as well as in the organism as a whole [9], and it is accelerated in neurodegeneration [148]. Additionally, zinc contributes to genomic stability [381], which tends to destabilize with age [9], and together with selenium, it might prevent or delay Alzheimer’s disease in the elderly with mild cognitive impairment [382]. Higher omega-3 levels are associated with greater total grey matter, total brain volume, and lower white matter lesion volume [383]. Omega-3 has been shown to display a decreased concentration in patients with dementia or predementia syndrome [384], while supplementation with omega-3 improved cognitive function in elderly patients with mild cognitive impairment [385] and Alzheimer’s disease [386]. Taking these observations together, one may conclude that there are multiple ways in which an individually tailored combination of nutraceutical compounds may contribute to BBA reversal, as well as BR enhancement, by modulating the epigenome [280], thus safeguarding physical and mental health during CA, and hypothetically even reducing mortality [281]. In contrast to the nutraceutical supplementation intervention, the lifestyle change intervention was quite effective in slowing down the brain BA and maintaining BR, thus stabilizing them against the natural and inevitable pressure of CA (Figure 2 and Figure 3). This is in agreement with previous research that suggests the beneficial effects of healthy habits over life [61,62]. Indeed, a healthy lifestyle that incorporates regular physical activity and a balanced diet promotes multiple anti-aging processes in the organism and the brain [205] and may even reverse the epigenetic age [90]. For example, the beneficial effects of physical exercise (through a mediation of glycosylphosphatidylinositol-specific phospholipase D1, which increases after exercise) on neurogenesis in the aged brain and to improve cognition have been recently demonstrated [272]. Neurogenesis progressively declines with age [387]; its decline is exacerbated in Alzheimer’s disease [388], correlates with cognitive dysfunction [389], and contributes to lifespan duration [390]. Thus, maintaining higher levels of brain neurogenesis is proposed to be neuroprotective and responsible for a rejuvenating/regenerative capacity in the aging brain [387], as it is linked to enhanced cognition and slower disease progression in the context of Alzheimer’s disease [388]. Generally, regular physical exercise plays an essential role in maintaining healthy neurocognitive function (especially in chronologically older individuals) [391], preservation of brain grey matter [392] and hippocampus volume [347], upregulation of neurotrophic factors, including brain-derived neurotrophic factors [393], and maintaining a healthy central nervous system immunometabolism during aging [394]. Similarly, a calorie restriction diet has been systematically demonstrated to extend both the life- and healthspan and to delay many aspects of aging (for example, the well-documented good health and high number of centenarians among the population of the Japanese of Okinawa island have been attributed to calorie restriction [395]) [396,397,398]. When it comes to the brain, diet, and specifically a fasting-mimicking diet, has been shown to be able to enhance remyelination (myelination refers to the process of creating myelin on the neuron axons (the nervous system’s “wires”), whereas myelin is a lipid-rich (fatty) substance that surrounds axons to insulate them and increase the rate at which electrical impulses (called action potentials) are passed along the axon [399]. In the central nervous system, axons carry electrical signals from one nerve cell body to another) in the aging brain by affecting the oligodendrocyte precursor cells [271] (oligodendrocyte precursor cells (OPCs) differentiate into mature oligodendrocytes, which myelinate axons in the mammalian brain, allowing for the rapid propagation of action potentials and metabolic support of axons [271]. While most myelination occurs during early postnatal development, OPCs persist in the adult brain [400]). The deeper mechanism at play is that the fasting-mimicking diet upregulates AMPK activity, which, in turn, inhibits mTOR activity in the oligodendrocyte precursor cells, leading to a markedly increased differentiation capacity of such cells, reminiscent of the young brain [400]. Furthermore, a fast-mimicking diet also leads to SIRT1 activation [369] and increased expression of mesencephalic astrocyte-derived neurotrophic factor (MANF) (MANF is an evolutionarily conserved protein [401] that is expressed by most tissues in the body [402] and is cytoprotective in multiple systems [270]) in the brain [403]. It is known that MANF levels progressively and significantly decline with age; however, its overexpression prevents age-related inflammation, deregulates metabolic function, and results in significant mean and maximum lifespan extension in animal models [404]. Thus, existing evidence highlights the benefits of lifestyle management as an effective intervention capable of slowing down brain aging. However, one has to follow such a rigorous program rather precisely on an everyday basis to achieve results beyond just the deceleration of aging, namely, the reversal of brain age, which is not easy in real life, where slowly accruing benefits may not be reaped or noticed (especially in the healthy/young) for decades to come. The difficulty of long-term compliance (which is well documented for the lifestyle changes [296,297,300,302]) was probably responsible for the fact that in our study only a slowdown (i.e., deceleration) of brain BA was achieved with the lifestyle intervention and not actual brain age reversal (this is in contrast to a much shorter (eight-week) lifestyle intervention (that included diet recommendations, physical exercise, and sleep advice) study, where the systemic/organismal BA was reversed by the end of the study [90]. The duration of the trial may, in fact, contribute to this discrepancy, because it might be easier for participants to follow the intervention program accurately for a much shorter time (by comparison, our study’s intervention duration was, on average, 13 months). Furthermore, while the study by Fitzgerald et al. [ 90] did not involve any nutraceutical compounds, it nevertheless allowed participants to continue using some nutraceuticals that they had used before enrolling in the study, thus creating a synergistic effect, where nutraceuticals worked alongside the lifestyle recommendations. Furthermore, our study estimated the qEEG-derived brain BA, while the study by Fitzgerald et al. [ 90] measured epigenetic systemic BA, which may have contributed to the difference in the results). Moreover, many nutraceutical compounds are, in fact, exercise or calorie restriction mimetics [302,303] (mimetics are compounds that activate (mimic) the same metabolic, biochemical, and physiological response pathways induced by calorie restriction (or fasting) or physical exercise without lowering food intake or practicing exercise [405,406]). Thus, with a proper dosing regimen and using combinations that reinforce the effects of separate compounds, one could amplify the beneficial effects of physical activity and diet and, thus, achieve stronger effects. This may explain why not only brain age deceleration but also brain BA reversal and an increase in BR were achieved with nutraceutical supplementation in the present study. Despite this difference (brain BA deceleration for the lifestyle intervention vs. BA reversal for the nutraceutical supplementation intervention), both results of the studied interventions are, in fact, important, as one may expect that each chronologically passing year (CA) produces less damage and smaller deteriorations in brain health (BA), thus resulting in a slower brain aging and, as a consequence, a greater gap between the biological and chronological age of the brain (Figure 3A). However, the intervention involving nutraceutical supplementation had an additional advantage: the BBA reversal was also accompanied by a dramatic decrease in the number of individuals who had ongoing health complaints (Table 3). This result is significant, especially in light of the current understanding that interventions that target aging have a greater impact on life expectancy and healthspan when the incidence of multiple diseases is reduced—compressed morbidity [12] (see also [16,25,176]). The analogous decrease in the lifestyle intervention was small and nonsignificant (Table 3). At the same time, our results show that both interventions effectively and similarly decreased the scores for depression and anxiety (Figure 6), thus having a comparable effect on mental health. Considering the known correlation between mental health and subjective age [153], we hypothesize that both interventions resulted in a decreased subjective age (importantly, it has been shown that subjective feeling regarding personal age is associated with brain BA [169]: persons who had an older brain BA reported that they felt less healthy and older than their CA; additionally, they also reported that they looked older than their CA and did not feel likely to live past 75 years. Such individuals had a thinner and smaller cortex, reduced hippocampal volume, and displayed early signs of white matter deterioration, as well as cognitive decline [169]). Since personal attitude towards aging is strongly associated with the incidence of age-related diseases, epigenetic aging, and mortality [154,156], modifying it by means of such interventions could be a simple and accessible way to increase human healthspan and improve well-being. One may consider that the reported BBA decrease of 2.8 years (after nutraceutical supplementation for approximately 1 year) is rather modest; however, such a decrease, if sustained, is likely to have a significant impact on personal health risks and well-being, as well as broad economic and societal benefits [8,12,407,408]. Indeed, it has been documented that slowed brain aging is associated with an increase in compensatory and neuroprotective mechanisms and an increased ability to maintain focus, adapt flexibly and quickly to new circumstances, integrate across multiple sensory modalities, and learn efficiently, while accelerated brain aging is associated with an increased risk of Alzheimer’s disease and other diseases that are typically accompanied by cognitive decline, as well as increased mortality [54,55]. Furthermore, the postintervention difference between brain BA and CA was very large in our study—BBA was 11.8 years younger than CA. In and of itself, this is remarkably significant; for example, for the organismal BA, it has been shown that for every 1-year increase in the calculated difference between the BA and CA (when the BA was older than the CA), the hazard ratio for mortality significantly increased by $1.6\%$ ($1.5\%$ in males and $2.0\%$ in females), as well as for hypertension ($2.5\%$), diabetes mellitus ($4.2\%$), heart disease ($1.3\%$), stroke ($1.6\%$), and cancer incidence ($0.4\%$) [47]. So, a younger age is associated with better prognoses for a variety of leading sources of human mortality, including, of course, the ongoing SARS-CoV-2 pandemic [409,410,411]. All these have relatively straightforward benefits to individual health- and lifespan; however, where society, as a collection of many individuals, is concerned, the economic benefits begin to emerge as well [21]. While some see health- and lifespan extension as a problem for society (for a review, see [412]), others show that there are, in fact, serious overall economic and societal gains to be had. It has been calculated that a slowdown in aging that increases life expectancy by one year is worth USD 38 trillion, and an increase of ten years is worth USD 367 trillion [12] (see also [407,408]). This is because biologically younger brains correlate with a longer life- and healthspan [178,181], where more people are alive at older (chronological) ages in better health (biological age), thus compressing morbidity [16,21,22,25]. So, when reaching older ages in good health, individuals also tend to (re)allocate more consumption, leisure, and productivity to these years, as they become more valuable [12]—people want to live long but with an ever-stronger interest in remaining healthy and living well [21]. In the words of Scott and colleagues [12], this situation creates a virtuous circle, such that the more successful a society is at improving how people age, the greater the economic and also individual value of further age improvements (however, not everyone is so optimistic. For example, Davis [413] asked if radical life extension would have value, meaning that such a life would have the unity or coherence to be recognizably human or whether a very long life must invariably become tedious. He also raises moral and political issues, for example, fairness, by asking who would be able to afford the life-extension interventions and whether such interventions would be accessible to everyone). Moreover, in our study we did not find any association between BBA and the duration of the interventions. One explanation could be that the beneficial effects of the interventions on brain BA (either its reversal by nutraceutical compounds or its stabilization by lifestyle) were effectively achieved during the first 6 months and then remained relatively stable. This interpretation is consistent with mathematical projections from a large-scale study, according to which the effects of a given longevity intervention in a “practically” healthy population will saturate in a relatively short period of time [414], but somehow this is in contrast with the observation of the systemic BA (estimated by four different epigenetic “clocks”), where there was a marked acceleration of BA reversal after 9 months of intervention that included recombinant human growth hormone, dehydroepiandrosterone, and metformin [268]. This discrepancy remains to be explored; however, it might be that hormonal and medication usage require more time to “kickoff”, or it could be that the brain is a faster responder than the whole organism. Another compelling result of the present study was that the baseline (pre-intervention) BBA was generally younger than the CA in both groups (Figure 2). This result is consistent with the estimation of organismal BA, which has repeatedly been shown to be lower than CA [111,190,195,268,415,416]. The same dependency was also found for psychological age, where people have a tendency to perceive themselves as younger than their calendar age [417,418], and curiously, this difference increases with CA [419,420]. Taken together, these observations (including the present study) may signify the existence of some deep mechanism that keeps BA and BBA systematically younger than the CA in the human population, thus uncoupling aging from the fixed progression of chronological time [74]. This may explain why humans are generally rather resilient [220] and the longest living among their closest ape “relatives” [107]. It must be noted, though, that this result reflects the average for the groups and that both groups had participants whose BBA at the pre-intervention time-point was either younger or, on the contrary, older than their CA. To analyze these participants, we pooled together data from both groups and then stratified the whole sample into two subgroups: pre-intervention BBA > CA and BBA < CA (see Table 2). The demographic data revealed that the participants whose pre-intervention BBA was younger than their CA had more BR, were more likely to be left-handed, were predominantly married, were more likely to have a PhD, enjoyed more hobbies, smoked less, and consumed more alcoholic drinks per week (Table 2). Largely, these findings are consistent with previous observations: the degree of education, marriage and socialization, diverse leisure activity/hobbies, and increased cognitive reserve were all associated with higher cognitive performance, neuroprotection, and resilience to neurodegeneration and Alzheimer’s disease [57,205,421,422,423], as well as with younger systemic (organismic) epigenetic age [102]. For example, it was found that superagers (or “high-performing older adults”)—individuals aged 80 years or older who retain exceptional cognitive and memory performance equal to or greater than that of individuals aged in their 50s or 60s [424]—had a higher level of education [425], a significantly thicker brain cortex [426], and a greater anterior cortex volume [427] compared with their age-matched peers with average-for-age memory and cognition (this is particularly important since the anterior cortex is linked with the phenomenal first-person perspective and the sense of agency or being a self [315]). Furthermore, long-term smoking has been associated with brain aging and degeneration [205,428,429]. The findings on alcohol consumption are rather mixed: while it was shown that heavy drinking is associated with a greater loss of grey and white matter in the brain [430] and with brain aging [431], moderate alcohol consumption (in particular wine) may be beneficial for the cardiovascular system, which is related to brain health and is associated with a reduced risk of dementia and better cognitive function [432]. Such positive effects might be mediated by polyphenols, flavonoids, and organic acids present in wine, which have antioxidant, anti-inflammatory, and neuroprotective mechanisms [433]. Interestingly, it has been documented recently that wine consumption is associated with a decelerated epigenetic aging [276]. The larger proportion of left-handed individuals in the BBA < CA subgroup (see Table 2) is peculiar and requires further study; however, some clues in the literature may already be established. Left-handed individuals usually experience a very quick reversal of pathological states, including brain functions after trauma or disorders [434]; left-handedness may be associated with a longer lifespan, especially if one reaches middle age [435,436], and there are disproportionately fewer left-handed patients with Alzheimer’s disease [437,438]. All this may point to some potential neuroprotective mechanisms present in left-handed individuals. Pre-intervention BBA score could be a covariate that may contribute differently to the overall results of the present study and, hence, we examined the effects of both interventions on the BBA separately after splitting the whole sample based on pre-intervention BBA scores (Figure 5). We found that in the experimental/nutraceuticals group, for the participants whose pre-intervention BBA was older than their CA, the BBA scored, on average, 6.77 years younger at the endpoint of the intervention compared to the beginning. For the participants whose pre-intervention BBA was younger than their CA, the BBA scored, on average, only 0.64 years younger at the endpoint of the intervention compared to the beginning (Figure 5). These results indicate that the BBA reversal after nutraceuticals supplementation was stronger for participants whose pre-intervention BBA was older than their CA (a result that is consistent with a recent finding that supplementation with alpha-ketoglutarate and vitamins resulted in a stronger decrease in systemic biological age in biologically older individuals [416]). A straightforward explanation could be that individuals with initially younger brains (and thus high brain and cognitive reserves) are already functioning at an optimal level (see also [260]). Consequently, additional interventions do not further optimize the functional brain patterns (contributing to the BBA) because of a ceiling effect: both capacity and efficiency in their brains have already reached the limit and “topped out”. As for the control/lifestyle group, the BBA scored, on average, 0.25 years older (for those whose pre-intervention BBA was older than their CA) and 0.13 years younger (for those whose pre-intervention BBA was younger than their CA) at the end of the intervention when compared with the baseline (Figure 5). Both changes were small and nonsignificant; therefore, one may conclude that the lifestyle intervention was not effective for reversing BBA but rather stabilized it despite the pressure of CA, thus achieving age deceleration. It is known that there are sex-related differences in brain structure (thickness of the cortex and proportion of grey matter) [439], morphology [440], functional organization [441], and aging trajectories [207] in humans. Therefore, we considered sex as a covariate that may affect the overall results and, hence, examined sex-specific differences in BBA for both interventions. For both females and males, the BBA scored younger at the end of the nutraceutical supplementation when compared with the baseline; however, the decrease in the BBA scores (i.e., age reversal) was stronger for females (Figure 4). This discrepancy between sexes may relate to the persistent observation that age-related brain atrophy (or metabolic brain age) is more extensive in males than in females [442,443] and, thus, initially, the BBA could be older in males when compared with females. Indeed, the BBA in males at the beginning of the study was, on average, 52.58 years old, while in females it was 44.31 years old, thus signifying an older brain BA in males at the pre-intervention time-point when compared with females. By comparison, for the lifestyle intervention, the BBA decreased slightly only in males, while it increased insignificantly in females at the end of the intervention (Figure 4). Both changes were very small, and knowing that lifestyle changes mostly stabilize the BBA (i.e., deceleration of aging; see above), one may conclude that the lifestyle intervention did not cause any significant changes in the BBA of both sexes, it only kept the BBA rate slowed down despite the pressure of CA. Additionally, other influences, such as genetic variations in females and males, may have had a further impact on the BBA [444,445]. ## 5. Conclusions, Significance, Limitations, and Future Research The present study demonstrated that brain BA deceleration, and even reversal, with accompanying improvements in mental–physical health comorbidities is possible in humans using accessible interventions, such as lifestyle changes or nutraceutical supplementation, within a practical time frame (~13 months). Although much more remains to be investigated, from a translational perspective, these findings are noteworthy given that lifestyle changes such as calorie restriction and intermittent fasting, physical exercise and proper sleep, and vitamins and nutrients are the most commonly used practices worldwide. While currently there is an unprecedented explosion of breakthroughs in many areas of basic science and even translational medicine related to aging [96,308], the new putative interventions are unlikely to be available to the average person any time soon. Therefore, such commonly available and relatively affordable interventions as a healthy lifestyle and nutraceutical supplementation are extremely important. Indeed, if these interventions can be made practical and scalable, we may find ourselves in a future in which we have “no time to age” [74]. Despite the remarkably promising results, the present study has several limitations. While this study was not small, larger prospective trials are warranted to confirm the initial observations of the present retrospective study. Furthermore, both interventions (experimental and active control) contained several components individually adjusted for every participant, and this was considered an advantage. At the same time, this makes it impossible to attribute the overall beneficial result of the intervention to any individual element of the intervention. As for nutraceutical compounds, currently there are new substances emerging that have a putatively strong anti-aging and geroprotective potential by targeting multiple hallmarks of aging [9] (see also [49,416]). Such new substances may be considered in future studies of BBA. Another limitation of the present study relates to the absence of an evaluation of the sustainability of BBA changes following some temporal interval after the discontinuation of the interventions to see if the decrease in BBA remains stable or rebounds back towards the pre-intervention levels. However, hints are already possible to derive; for example, we had one participant who had three assessments instead of two, roughly 6 months apart. At the first assessment, the person received the nutraceutical supplementation program, which she followed for 6 months until the second assessment, when the program was discontinued; the third assessment was conducted after another 6 months, without any intervention (Figure 7). For this person, although their pre-intervention BBA was lower than their CA (assessment 1), the BBA noticeably decreased postintervention after 6 months (assessment 2) and then had a tendency to rebound back towards the pre-intervention level 6 months after discontinuing the intervention (assessment 3). At the same time, the BBA rebound was not complete, still having an improvement of 11.6 years in comparison with their CA at the third assessment. If future durability studies do establish a gradual loss of BBA reversal compared to baseline, it will be interesting to determine whether periodic repetition of the intervention might restore the initially achieved level of BBA reversal. It also remains to be seen whether further adjustments of the combination of nutraceutical compounds and/or their dosages will further augment BBA reversal. Additionally, the combination of nutraceutical supplementation and lifestyle changes within the same intervention program may be even more impactful, and this remains to be studied in future trials (for example, see [446]). Furthermore, since this was the first study to show that nutraceutical supplementation and lifestyle could affect brain aging, only standard statistical analyses were performed, resulting in a large spectrum of results. However, secondary analysis of these data will be needed to dissect the causal relationships between the BBA rates and nutraceuticals/lifestyle factors and to estimate the potential correlations between the covariates. Additionally, we applied a previously developed method to quantify the qEEG-derived BBA. At the same time, there may be other methods, and they may produce different results. Yet another limitation relates to the potential role of hormones which may contribute to changes in BBA, as the levels of hormones are naturally different in males and females, as well as in young and older persons. The hormonal status was not controlled in the present study and, therefore, future research should consider it. CA, itself, could also be a confounding factor, and future larger prospective studies should specifically address this issue, though it has been documented previously that lifespan extension is comparable if the anti-aging intervention is initiated at a young age, middle age, or in late life [354]. Finally, the participants were self-selected with respect to the intervention type (nutraceutical compounds vs. lifestyle), and the blinding of interventions was not possible due to the different nature of the selected interventions. At the same time, the participants were blinded to the interventions’ primary output related to the qEEG-derived BBA (participants thought that the interventions aimed to improve their general well-being); therefore, a potential placebo effect related to BBA in both interventions could be excluded. In this respect, this retrospective study can be considered single-blind. In spite of these limitations, some of the strengths of our study include a relatively large sample size (42 participants in the experimental/nutraceuticals group and 47 in the control/lifestyle group), a wide range of CAs in the sample spanning from 25 to 77 years old, and the use of the qEEG-derived BBA as a simple and reliable biomarker of brain aging. Furthermore, the present study had a longitudinal design, which allowed for conclusions regarding the directionality of the anti-aging intervention effects. ## References 1. Weissman A.. Essays Upon Heredity and Kindred Biological Problems (1891) 2. Cave S.. Immortality: The Quest to Live Forever and How It Drives Civilization (2012) 3. Cholbi M.J.. The Science of Immortality (2018) 4. Bostrom N.. Are you living in a computer simulation?. Philos. Q. (2003) 53 243-255. DOI: 10.1111/1467-9213.00309 5. Bostrom N., Berg Olsen J.-K., Selinger E., Aldershot S.R.. The Future of Humanity. New Waves in Philosophy of Technology (2009) 186-216 6. Hughes J.J.. The compatibility of religious and transhumanist views of metaphysics, suffering, virtue and transcendence in an enhanced future. Glob. Spiral (2007) 8 1-40 7. Fingelkurts A.A., Fingelkurts A.A.. After human. Futura (2018) 4 60-74 8. Zhavoronkov A.. The Ageless Generation: How Advances in Biomedicine will Transform the Global Economy (2013) 9. Sinclair D.A.. Lifespan: Why We Age—And Why We Don’t Have To (2019) 10. West M.D., Sternberg H., Labat I., Janus J., Chapman K.B., Malik N.N., De Grey A.D., Larocca D.. Toward a unified theory of aging and regeneration. Regen. Med. (2019) 14 867-886. DOI: 10.2217/rme-2019-0062 11. Barzilai N.. Age Later: Health Span, Life Span, and the New Science of Longevity (2020) 12. Scott A.J., Ellison M., Sinclair D.A.. The economic value of targeting aging. Nat. Aging (2021) 1 616-623. DOI: 10.1038/s43587-021-00080-0 13. de Grey A.D.N.J.. A thousand years young. Futurist (2012) 46 18-23 14. de Grey A.D.N.J.. Escape Velocity: Why the Prospect of Extreme Human Life Extension Matters Now. PLoS Biol. (2004) 2. DOI: 10.1371/journal.pbio.0020187 15. Oeppen J., Vaupel J.W.. Broken limits to life expectancy. Science (2002) 296 1029-1031. DOI: 10.1126/science.1069675 16. Zhavoronkov A., Bischof E., Lee K.-F.. Artificial intelligence in longevity medicine. Nat. Aging (2021) 1 5-7. DOI: 10.1038/s43587-020-00020-4 17. Global age-sex-specific fertility, mortality, healthy life expectancy (HALE), and population estimates in 204 countries and territories, 1950–2019: A comprehensive demographic analysis for the Global Burden of Disease Study 2019. Lancet (2020) 396 1160-1203. DOI: 10.1016/S0140-6736(20)30977-6 18. Olshansky S.J.. From Lifespan to Healthspan. JAMA (2018) 320 1323-1324. DOI: 10.1001/jama.2018.12621 19. Cutler R.G., Mattson M.P.. The adversities of ageing. Ageing Res. Rev. (2006) 5 221-238. DOI: 10.1016/j.arr.2006.05.002 20. de Grey A.D.. Life Span Extension Research and Public Debate: Societal Considerations. Stud. Ethic Law, Technol. (2007) 1. DOI: 10.2202/1941-6008.1011 21. Seals D.R., Melov S.. Translational Geroscience: Emphasizing function to achieve optimal longevity. Aging (2014) 6 718-730. DOI: 10.18632/aging.100694 22. Fries J.F.. Aging, natural death, and the compression of morbidity. N. Engl. J. Med. (1980) 303 130-135. DOI: 10.1056/NEJM198007173030304 23. Omran A.R.. The Epidemiologic Transition: A Theory of the Epidemiology of Population Change. Milbank Mem. Fund Q. (1971) 49 509. DOI: 10.2307/3349375 24. Olshansky S.J., Ault A.B.. The Fourth Stage of the Epidemiologic Transition: The Age of Delayed Degenerative Diseases. Milbank Q. (1986) 64 355. DOI: 10.2307/3350025 25. Campisi J., Kapahi P., Lithgow G.J., Melov S., Newman J.C., Verdin E.. From discoveries in ageing research to therapeutics for healthy ageing. Nature (2019) 571 183-192. DOI: 10.1038/s41586-019-1365-2 26. Skulachev V.P.. What is “phenoptosis” and how to fight it?. Biochemistry (2012) 77 689-706. DOI: 10.1134/S0006297912070012 27. Moskalev A.A., Shaposhnikov M.V., Plyusnina E.N., Zhavoronkov A., Budovsky A., Yanai H., Fraifeld V.E.. The role of DNA damage and repair in aging through the prism of Koch-like criteria. Ageing Res. Rev. (2013) 12 661-684. DOI: 10.1016/j.arr.2012.02.001 28. Sierra F.. Geroscience and the challenges of aging societies. Aging Med. (2019) 2 132-134. DOI: 10.1002/agm2.12082 29. Galkin F., Mamoshina P., Aliper A., de Magalhães J.P., Gladyshev V.N., Zhavoronkov A.. Biohorology and biomarkers of aging: Current state-of-the-art, challenges and opportunities. Ageing Res. Rev. (2020) 60 101050. DOI: 10.1016/j.arr.2020.101050 30. Kulkarni A.S., Gubbi S., Barzilai N.. Benefits of Metformin in Attenuating the Hallmarks of Aging. Cell Metab. (2020) 32 15-30. DOI: 10.1016/j.cmet.2020.04.001 31. Skulachev V.P.. Aging is a specific biological function rather than the result of a disorder in complex living systems: Biochemical evidence in support of Weismann’s hypothesis. Biochemistry (1997) 62 1191-1195. PMID: 9467841 32. de Grey A.D.N.J.. Challenging but essential targets for genuine antiageing drugs. Expert Opin. Ther. Targets (2003) 7 1-5. DOI: 10.1517/14728222.7.1.1 33. Butler R.N., Sprott R., Warner H., Bland J., Feuers R., Forster M., Fillit H., Harman S.M., Hewitt M., Hyman M.. Aging: The Reality: Biomarkers of Aging: From Primitive Organisms to Humans. J. Gerontol. Ser. A (2004) 59 B560-B567. DOI: 10.1093/gerona/59.6.B560 34. López-Otín C., Blasco M.A., Partridge L., Serrano M., Kroemer G.. The hallmarks of aging. Cell (2013) 153 1194-1217. DOI: 10.1016/j.cell.2013.05.039 35. Kennedy B.K., Berger S.L., Brunet A., Campisi J., Cuervo A.M., Epel E.S., Franceschi C., Lithgow G.J., Morimoto R.I., Pessin J.E.. Geroscience: Linking Aging to Chronic Disease. Cell (2014) 159 709-713. DOI: 10.1016/j.cell.2014.10.039 36. Jylhävä J., Pedersen N.L., Hägg S.. Biological Age Predictors. Ebiomedicine (2017) 21 29-36. DOI: 10.1016/j.ebiom.2017.03.046 37. Gems D., de Magalhães J.P.. The hoverfly and the wasp: A critique of the hallmarks of aging as a paradigm. Ageing Res. Rev. (2021) 70 101407. DOI: 10.1016/j.arr.2021.101407 38. Jia L., Zhang W., Chen X.. Common methods of biological age estimation. Clin. Interv. Aging (2017) 12 759-772. DOI: 10.2147/CIA.S134921 39. Finkel D., Sternäng O., Wahlin Å.. Genetic and Environmental Influences on Longitudinal Trajectories of Functional Biological Age: Comparisons Across Gender. Behav. Genet. (2017) 47 375-382. DOI: 10.1007/s10519-017-9851-5 40. Gott A., Andrews C., Larriva-Hormigos M., Spencer K., Bateson M., Nettle D.. Chronological age, biological age, and individual variation in the stress response in the European starling: A follow-up study. PeerJ (2018) 6 e5842. DOI: 10.7717/peerj.5842 41. Gunn D.A., Rexbye H., Griffiths C., Murray P.G., Fereday A., Catt S.D., Tomlin C.C., Strongitharm B.H., Perrett D., Catt M.. Why Some Women Look Young for Their Age. PLoS ONE (2009) 4. DOI: 10.1371/journal.pone.0008021 42. Ahadi S., Zhou W., Rose S.M.S.-F., Sailani M.R., Contrepois K., Avina M., Ashland M., Brunet A., Snyder M.. Personal aging markers and ageotypes revealed by deep longitudinal profiling. Nat. Med. (2020) 26 83-90. DOI: 10.1038/s41591-019-0719-5 43. Lehallier B., Gate D., Schaum N., Nanasi T., Lee S.E., Yousef H., Losada P.M., Berdnik D., Keller A., Verghese J.. Undulating changes in human plasma proteome profiles across the lifespan. Nat. Med. (2019) 25 1843-1850. DOI: 10.1038/s41591-019-0673-2 44. Mamoshina P., Kochetov K., Cortese F., Kovalchuk A., Aliper A., Putin E., Scheibye-Knudsen M., Cantor C.R., Skjodt N.M., Kovalchuk O.. Blood Biochemistry Analysis to Detect Smoking Status and Quantify Accelerated Aging in Smokers. Sci. Rep. (2019) 9 142. DOI: 10.1038/s41598-018-35704-w 45. Jagust W.J.. Youthfulness begins in youth. Nat. Aging (2021) 1 239-240. DOI: 10.1038/s43587-021-00048-0 46. Levine M.E.. Modeling the rate of senescence: Can estimated biological age predict mortality more accurately than chronological age?. J. Gerontol. A Biol. Sci. Med. Sci. (2013) 68 667-674. DOI: 10.1093/gerona/gls233 47. Kang Y.G., Suh E., Lee J.-W., Kim D.W., Cho K.H., Bae C.-Y.. Biological age as a health index for mortality and major age-related disease incidence in Koreans: National Health Insurance Service—Health screening 11-year follow-up study. Clin. Interv. Aging (2018) 13 429-436. DOI: 10.2147/CIA.S157014 48. Zhavoronkov A., Mamoshina P.. Deep Aging Clocks: The Emergence of AI-Based Biomarkers of Aging and Longevity. Trends Pharmacol. Sci. (2019) 40 546-549. DOI: 10.1016/j.tips.2019.05.004 49. Zhavoronkov A., Li R., Ma C., Mamoshina P.. Deep biomarkers of aging and longevity: From research to applications. Aging (2019) 11 10771-10780. DOI: 10.18632/aging.102475 50. Borkan G.A., Norris A.H.. Assessment of Biological Age Using a Profile of Physical Parameters. J. Gerontol. (1980) 35 177-184. DOI: 10.1093/geronj/35.2.177 51. Jackson S.H., Weale M.R., Weale R.A.. Biological age-what is it and can it be measured?. Arch. Gerontol. Geriatr. (2003) 36 103-115. DOI: 10.1016/S0167-4943(02)00060-2 52. Fried L.P., Xue Q.-L., Cappola A.R., Ferrucci L., Chaves P., Varadhan R., Guralnik J.M., Leng S.X., Semba R.D., Walston J.D.. Nonlinear Multisystem Physiological Dysregulation Associated With Frailty in Older Women: Implications for Etiology and Treatment. J. Gerontol. Ser. A (2009) 64A 1049-1057. DOI: 10.1093/gerona/glp076 53. Franke K., Gaser C.. Longitudinal changes in individual BrainAGE in healthy ageing, mild cognitive impairment, and Alzheimer’s Disease. GeroPsych (2012) 25 235-245. DOI: 10.1024/1662-9647/a000074 54. Cole J.H., Annus T., Wilson L.R., Remtulla R., Hong Y.T., Fryer T.D., Acosta-Cabronero J., Cardenas-Blanco A., Smith R., Menon D.K.. Brain predicted age in Down syndrome is associated with beta amyloid deposition and cognitive decline. Neurobiol. Ageing (2017) 56 41-49. DOI: 10.1016/j.neurobiolaging.2017.04.006 55. Cole J.H., Ritchie S.J., Bastin M.E., Hernández M.C.V., Maniega S.M., Royle N., Corley J., Pattie A., Harris S.E., Zhang Q.. Brain age predicts mortality. Mol. Psychiatry (2018) 23 1385-1392. DOI: 10.1038/mp.2017.62 56. Christensen K., Wojczynski M.K., Pedersen J.K., Larsen L.A., Kløjgaard S., Skytthe A., McGue M., Vaupel J.W., Province M.A.. Mechanisms underlying familial aggregation of exceptional health and survival: A three-generation cohort study. Aging Cell (2020) 19 13228. DOI: 10.1111/acel.13228 57. Stern Y., Gurland B., Tatemichi T.K., Tang M.X., Wilder D., Mayeux R.. Influence of education and occupation on the incidence of Alzheimer’s disease. JAMA (1994) 271 1004-1010. DOI: 10.1001/jama.1994.03510370056032 58. Sayer A.A., Cooper C., Evans J.R., Rauf A., Wormald R.P., Osmond C., Barker D.J.. Are rates of ageing determined in utero?. Age Ageing (1998) 27 579-583. DOI: 10.1093/ageing/27.5.579 59. Syddall H.E., Sayer A.A., Simmonds S.J., Osmond C., Cox V., Dennison E.M., Barker D.J., Cooper C.. Birth weight, infant weight gain, and cause-specific mortality: The Hertfordshire Cohort Study. Am. J. Epidemiol. (2005) 161 1074-1080. DOI: 10.1093/aje/kwi137 60. Glatt S.J., Chayavichitsilp P., Depp C., Schork N.J., Jeste D.V.. Successful Aging: From Phenotype to Genotype. Biol. Psychiatry (2007) 62 282-293. DOI: 10.1016/j.biopsych.2006.09.015 61. Pruchno R.A., Wilson-Genderson M., Rose M., Cartwright F.. Successful Aging: Early Influences and Contemporary Characteristics. Gerontologist (2010) 50 821-833. DOI: 10.1093/geront/gnq041 62. Schafer M.H., Ferraro K.F.. Childhood Misfortune as a Threat to Successful Aging: Avoiding Disease. Gerontologist (2012) 52 111-120. DOI: 10.1093/geront/gnr071 63. Wang H.-X., Gustafson D.R., Kivipelto M., Pedersen N.L., Skoog I., Windblad B., Fratiglioni L.. Education halves the risk of dementia due to apolipoprotein ε4 allele: A collaborative study from the Swedish Brain Power initiative. Neurobiol. Aging (2012) 33 1007.e1-1007.e7. DOI: 10.1016/j.neurobiolaging.2011.10.003 64. Brown B.M., Peiffer J.J., Taddei K., Lui J.K., Laws S.M., Gupta V.B., Taddei T., Ward V.K., Rodrigues M.A., Burnham S.. Physical activity and amyloid-beta plasma and brain levels: Results from the Australian imageing, biomarkers and lifestyle study of ageing. Mol. Psychiatry (2013) 18 875-881. DOI: 10.1038/mp.2012.107 65. Sexton C.E., Storsve A.B., Walhovd K.B., Johansen-Berg H., Fjell A.M.. Poor sleep quality is associated with increased cortical atrophy in community-dwelling adults. Neurology (2014) 83 967-973. DOI: 10.1212/WNL.0000000000000774 66. Arenaza-Urquijo E.M., Gonneaud J., Fouquet M., Perrotin A., Mezenge F., Landeau B., Egret S., De la Sayette V., Desgranges B., Chételat G.. Interaction between years of education and APOE epsilon4 status on frontal and temporal metabolism. Neurology (2015) 85 1392-1399. DOI: 10.1212/WNL.0000000000002034 67. Gardener S.L., Rainey-Smith S.R., Barnes M.B., Sohrabi H.R., Weinborn M., Lim Y.Y., Harrington K., Taddei K., Gu Y., Rembach A.. Dietary patterns and cognitive decline in an Australian study of ageing. Mol. Psychiatry (2015) 20 860-866. DOI: 10.1038/mp.2014.79 68. Branger P., Arenaza-Urquijo E.M., Tomadesso C., Mézenge F., André C., de Flores R., Mutlu J., de La Sayette V., Eustache F., Chételat G.. Relationships between sleep quality and brain volume, metabolism, and amyloid deposition in late adulthood. Neurobiol. Aging (2016) 41 107-114. DOI: 10.1016/j.neurobiolaging.2016.02.009 69. Löwe L.C., Gaser C., Franke K.. The Effect of the APOE Genotype on Individual BrainAGE in Normal Aging, Mild Cognitive Impairment, and Alzheimer’s Disease. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0157514 70. Steffener J., Habeck C., O’Shea D., Razlighi Q., Bherer L., Stern Y.. Differences between chronological and brain age are related to education and self-reported physical activity. Neurobiol. Aging (2016) 40 138-144. DOI: 10.1016/j.neurobiolaging.2016.01.014 71. Vemuri P., Lesnick T.G., Przybelski S.A., Knopman D.S., Machulda M., Lowe V.J., Mielke M.M., Roberts R.O., Gunter J.L., Senjem M.L.. Effect of intellectual enrichment on AD biomarker trajectories: Longitudinal imageing study. Neurology (2016) 86 1128-1135. DOI: 10.1212/WNL.0000000000002490 72. Reutrakul S., Van Cauter E.. Sleep influences on obesity, insulin resistance, and risk of type 2 diabetes. Metab. Clin. Exp. (2018) 84 56-66. DOI: 10.1016/j.metabol.2018.02.010 73. Hayflick L.. Biological Aging Is No Longer an Unsolved Problem. Ann. N. Y. Acad. Sci. (2007) 1100 1-13. DOI: 10.1196/annals.1395.001 74. Larocca D., Lee J., West M., Labat I., Sternberg H.. No Time to Age: Uncoupling Aging from Chronological Time. Genes (2021) 12. DOI: 10.3390/genes12050611 75. Bernstein B.E., Meissner A., Lander E.S.. The Mammalian Epigenome. Cell (2007) 128 669-681. DOI: 10.1016/j.cell.2007.01.033 76. Bird A.. DNA methylation patterns and epigenetic memory. Genes Dev. (2002) 16 6-21. DOI: 10.1101/gad.947102 77. Feinberg A.P.. Phenotypic plasticity and the epigenetics of human disease. Nature (2007) 447 433-440. DOI: 10.1038/nature05919 78. Griñán-Ferré C., Bellver-Sanchis A., Izquierdo V., Corpas R., Roig-Soriano J., Chillón M., Andres-Lacueva C., Somogyvári M., Sőti C., Sanfeliu C.. The pleiotropic neuroprotective effects of resveratrol in cognitive decline and Alzheimer’s disease pathology: From antioxidant to epigenetic therapy. Ageing Res. Rev. (2021) 67 101271. DOI: 10.1016/j.arr.2021.101271 79. van Otterdijk S.D., Michels K.B.. Transgenerational epigenetic inheritance in mammals: How good is the evidence?. FASEB J. (2016) 30 2457-2465. DOI: 10.1096/fj.201500083 80. Trerotola M., Relli V., Simeone P., Alberti S.. Epigenetic inheritance and the missing heritability. Hum. Genom. (2015) 9 17. DOI: 10.1186/s40246-015-0041-3 81. Tian S., Lin X.-H., Xiong Y.-M., Liu M.-E., Yu T.-T., Lv M., Zhao W., Xu G.-F., Ding G.-L., Xu C.-M.. Prevalence of Prediabetes Risk in Offspring Born to Mothers with Hyperandrogenism. Ebiomedicine (2017) 16 275-283. DOI: 10.1016/j.ebiom.2017.01.011 82. Izquierdo V., Palomera-Ávalos V., López-Ruiz S., Canudas A.-M., Pallàs M., Griñán-Ferré C.. Maternal Resveratrol Supplementation Prevents Cognitive Decline in Senescent Mice Offspring. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20051134 83. Lumey L.. Decreased birthweights in infants after maternal in utero exposure to the Dutch famine of 1944–1945. Paediatr. Périnat. Epidemiology (1992) 6 240-253. DOI: 10.1111/j.1365-3016.1992.tb00764.x 84. Lumey L.H., Stein A., Kahn H., Bruin K.M.V.D.P.-D., Blauw G.J., Zybert P.A., Susser E.S.. Cohort Profile: The Dutch Hunger Winter Families Study. Leuk. Res. (2007) 36 1196-1204. DOI: 10.1093/ije/dym126 85. Eriksson J., Sandboge S., Salonen M.K., Kajantie E., Osmond C.. Long-term consequences of maternal overweight in pregnancy on offspring later health: Findings from the Helsinki Birth Cohort Study. Ann. Med. (2014) 46 434-438. DOI: 10.3109/07853890.2014.919728 86. Sen P., Shah P.P., Nativio R., Berger S.L.. Epigenetic Mechanisms of Longevity and Aging. Cell (2016) 166 822-839. DOI: 10.1016/j.cell.2016.07.050 87. Guillaumet-Adkins A., Yañez Y., Peris-Diaz M.D., Calabria I., Palanca-Ballester C., Sandoval J.. Epigenetics and Oxidative Stress in Aging. Oxidative Med. Cell. Longev. (2017) 2017 9175806. DOI: 10.1155/2017/9175806 88. Jin K., Simpkins J.W., Ji X., Leis M., Stambler I.. The Critical Need to Promote Research of Aging and Aging-related Diseases to Improve Health and Longevity of the Elderly Population. Aging Dis. (2014) 6 1-5. DOI: 10.14336/ad.2014.1210 89. Kane A.E., Sinclair D.A.. Epigenetic changes during aging and their reprogramming potential. Crit. Rev. Biochem. Mol. Biol. (2019) 54 61-83. DOI: 10.1080/10409238.2019.1570075 90. Fitzgerald K.N., Hodges R., Hanes D., Stack E., Cheishvili D., Szyf M., Henkel J., Twedt M.W., Giannopoulou D., Herdell J.. Potential reversal of epigenetic age using a diet and lifestyle intervention: A pilot randomized clinical trial. Aging (2021) 13 9419-9432. DOI: 10.18632/aging.202913 91. Takahashi K., Tanabe K., Ohnuki M., Narita M., Ichisaka T., Tomoda K., Yamanaka S.. Induction of Pluripotent Stem Cells from Adult Human Fibroblasts by Defined Factors. Cell (2007) 131 861-872. DOI: 10.1016/j.cell.2007.11.019 92. Vaziri H., Chapman K., Guigova A., Teichroeb J., Lacher M.D., Sternberg H., Singec I., Briggs L., Wheeler J., Sampathkumar J.. Spontaneous reversal of the developmental aging of normal human cells following transcriptional reprogramming. Regen. Med. (2010) 5 345-363. DOI: 10.2217/rme.10.21 93. Lapasset L., Milhavet O., Prieur A., Besnard E., Babled A., Aït-Hamou N., Leschik J., Pellestor F., Ramirez J.-M., De Vos J.. Rejuvenating senescent and centenarian human cells by reprogramming through the pluripotent state. Genes Dev. (2011) 25 2248-2253. DOI: 10.1101/gad.173922.111 94. Wahlestedt M., Erlandsson E., Kristiansen T., Lu R., Brakebusch C., Weissman I.L., Yuan J., Martin-Gonzalez J., Bryder D.. Clonal reversal of ageing-associated stem cell lineage bias via a pluripotent intermediate. Nat. Commun. (2017) 8 14533. DOI: 10.1038/ncomms14533 95. Lee J., Bignone P.A., Coles L.S., Liu Y., Snyder E., Larocca D.. Induced pluripotency and spontaneous reversal of cellular aging in supercentenarian donor cells. Biochem. Biophys. Res. Commun. (2020) 525 563-569. DOI: 10.1016/j.bbrc.2020.02.092 96. Lu Y., Brommer B., Tian X., Krishnan A., Meer M., Wang C., Vera D.L., Zeng Q., Yu D., Bonkowski M.S.. Reprogramming to recover youthful epigenetic information and restore vision. Nature (2020) 588 124-129. DOI: 10.1038/s41586-020-2975-4 97. Johnson A.A., Akman K., Calimport S.R.G., Wuttke D., Stolzing A., de Magalhães J.P.. The Role of DNA Methylation in Aging, Rejuvenation, and Age-Related Disease. Rejuv. Res. (2012) 15 483-494. DOI: 10.1089/rej.2012.1324 98. Mitteldorf J.J.. How does the body know how old it is? Introducing the epigenetic clock hypothesis. Biochemistry (2013) 78 1048-1053. DOI: 10.1134/s0006297913090113 99. Field A.E., Robertson N.A., Wang T., Havas A., Ideker T., Adams P.D.. DNA Methylation Clocks in Aging: Categories, Causes, and Consequences. Mol. Cell (2018) 71 882-895. DOI: 10.1016/j.molcel.2018.08.008 100. Horvath S., Raj K.. DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nat. Rev. Genet. (2018) 19 371-384. DOI: 10.1038/s41576-018-0004-3 101. Pal S., Tyler J.K.. Epigenetics and aging. Sci. Adv. (2016) 2 e1600584. DOI: 10.1126/sciadv.1600584 102. Levine M.E., Lu A.T., Quach A., Chen B.H., Assimes T.L., Bandinelli S., Hou L., Baccarelli A.A., Stewart J.D., Li Y.. An epigenetic biomarker of aging for lifespan and healthspan. Aging (2018) 10 573-591. DOI: 10.18632/aging.101414 103. Li X., Wei W., Zhao Q.-Y., Widagdo J., Baker-Andresen D., Flavell C.R., D’Alessio A., Zhang Y., Bredy T.W.. Neocortical Tet3-mediated accumulation of 5-hydroxymethylcytosine promotes rapid behavioral adaptation. Proc. Natl. Acad. Sci. USA (2014) 111 7120-7125. DOI: 10.1073/pnas.1318906111 104. Marioni R.E., Shah S., McRae A.F., Ritchie S.J., Muniz-Terrera G., Harris S.E., Gibson J., Redmond P., Cox S.R., Pattie A.. The epigenetic clock is correlated with physical and cognitive fitness in the Lothian Birth Cohort 1936. Int. J. Epidemiol. (2015) 44 1388-1396. DOI: 10.1093/ije/dyu277 105. Maier H., Gampe J., Jeune B., Robine J.-M., Vaupel J.W.. Supercentenarians (2010) 106. Guevara E.E., Lawler R.R., Staes N., White C.M., Sherwood C.C., Ely J.J., Hopkins W.D., Bradley B.J.. Age-associated epigenetic change in chimpanzees and humans. Philos. Trans. R. Soc. B Biol. Sci. (2020) 375. DOI: 10.1098/rstb.2019.0616 107. Skulachev V.P., Holtze S., Vyssokikh M.Y., Bakeeva L.E., Skulachev M.V., Markov A.V., Hildebrandt T.B., Sadovnichii V.A.. Neoteny, prolongation of youth: From naked mole rats to “naked apes” (Humans). Physiol. Rev. (2017) 97 699-720. DOI: 10.1152/physrev.00040.2015 108. Horvath S., Pirazzini C., Bacalini M.G., Gentilini D., Di Blasio A.M., Delledonne M., Mari D., Arosio B., Monti D., Passarino G.. Decreased epigenetic age of PBMCs from Italian semi-supercentenarians and their offspring. Aging (2015) 7 1159-1170. DOI: 10.18632/aging.100861 109. Gutman D., Rivkin E., Fadida A., Sharvit L., Hermush V., Rubin E., Kirshner D., Sabin I., Dwolatzky T., Atzmon G.. Exceptionally Long-Lived Individuals (ELLI) Demonstrate Slower Aging Rate Calculated by DNA Methylation Clocks as Possible Modulators for Healthy Longevity. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21020615 110. Fransquet P.D., Wrigglesworth J., Woods R.L., Ernst M.E., Ryan J.. The epigenetic clock as a predictor of disease and mortality risk: A systematic review and meta-analysis. Clin. Epigenetics (2019) 11 62. DOI: 10.1186/s13148-019-0656-7 111. Grodstein F., Lemos B., Yu L., Iatrou A., De Jager P.L., Bennett D.A.. Characteristics of Epigenetic Clocks Across Blood and Brain Tissue in Older Women and Men. Front. Neurosci. (2021) 14 555307. DOI: 10.3389/fnins.2020.555307 112. Siegel R.L., Miller K.D., Jemal A.. Cancer statistics, 2018. CA Cancer J. Clin. (2018) 68 7-30. DOI: 10.3322/caac.21442 113. Shah N.S., Lloyd-Jones D.M., O’Flaherty M., Capewell S., Kershaw K.N., Carnethon M., Khan S.. Trends in Cardiometabolic Mortality in the United States, 1999-2017. JAMA (2019) 322 780-782. DOI: 10.1001/jama.2019.9161 114. Marras C., Beck J.C., Bower J.H., Roberts E., Ritz B., Ross G.W., Abbott R.D., Savica R., Van Den Eeden S.K., Willis A.W.. Prevalence of Parkinson’s disease across North America. NPJ Parkinson’s Dis. (2018) 4 21. DOI: 10.1038/s41531-018-0058-0 115. 2019 Alzheimer’s disease facts and figures. Alzheimer’s Dement. (2019) 15 321-387. DOI: 10.1016/j.jalz.2019.01.010 116. Moya-Alvarado G., Gershoni-Emek N., Perlson E., Bronfman F.C.. Neurodegeneration and alzheimer’s disease (AD). What can proteomics tell us about the alzheimer’s brain?. Mol. Cell Proteom. (2016) 15 409-425. DOI: 10.1074/mcp.R115.053330 117. Gao C., Wang Q., Chung S.K., Shen J.. Crosstalk of metabolic factors and neurogenic signaling in adult neurogenesis: Implication of metabolic regulation for mental and neurological diseases. Neurochem. Int. (2017) 106 24-36. DOI: 10.1016/j.neuint.2017.02.001 118. Dugger B.N., Dickson D.W.. Pathology of Neurodegenerative Diseases. Cold Spring Harb. Perspect. Biol. (2017) 9 a028035. DOI: 10.1101/cshperspect.a028035 119. Wong E., Cuervo A.M.. Autophagy gone awry in neurodegenerative diseases. Nat. Neurosci. (2010) 13 805-811. DOI: 10.1038/nn.2575 120. Berry B.J., Kaeberlein M.. An energetics perspective on geroscience: Mitochondrial protonmotive force and aging. Geroscience (2021) 43 1591-1604. DOI: 10.1007/s11357-021-00365-7 121. Misrani A., Tabassum S., Yang L.. Mitochondrial Dysfunction and Oxidative Stress in Alzheimer’s Disease. Front. Aging Neurosci. (2021) 13 617588. DOI: 10.3389/fnagi.2021.617588 122. Harper S.. Economic and social implications of aging societies. Science (2014) 346 587-591. DOI: 10.1126/science.1254405 123. Patton A.P., Hastings M.H.. The suprachiasmatic nucleus. Curr. Biol. (2018) 28 R816-R822. DOI: 10.1016/j.cub.2018.06.052 124. Hastings M., Maywood E., Brancaccio M.. The Mammalian Circadian Timing System and the Suprachiasmatic Nucleus as Its Pacemaker. Biology (2019) 8. DOI: 10.3390/biology8010013 125. Sinturel F., Gos P., Petrenko V., Hagedorn C., Kreppel F., Storch K.-F., Knutti D., Liani A., Weitz C., Emmenegger Y.. Circadian hepatocyte clocks keep synchrony in the absence of a master pacemaker in the suprachiasmatic nucleus or other extrahepatic clocks. Genes Dev. (2021) 35 329-334. DOI: 10.1101/gad.346460.120 126. Berger J.. Regulation of circadian rhythms. J. Appl. Biomed. (2004) 2 131-140. DOI: 10.32725/jab.2004.016 127. Mohawk J.A., Green C.B., Takahashi J.S.. Central and peripheral circadian clocks in mammals. Annu. Rev. Neurosci. (2012) 35 445-462. DOI: 10.1146/annurev-neuro-060909-153128 128. Astiz M., Heyde I., Oster H.. Mechanisms of Communication in the Mammalian Circadian Timing System. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20020343 129. Bass J.. Circadian topology of metabolism. Nature (2012) 491 348-356. DOI: 10.1038/nature11704 130. Fu L., Lee C.C.. The circadian clock: Pacemaker and tumour suppressor. Nat. Rev. Cancer (2003) 3 350-361. DOI: 10.1038/nrc1072 131. Sahar S., Sassone-Corsi P.. Metabolism and cancer: The circadian clock connection. Nat. Rev. Cancer (2009) 9 886-896. DOI: 10.1038/nrc2747 132. Froy O.. Circadian Rhythms, Aging, and Life Span in Mammals. Physiology (2011) 26 225-235. DOI: 10.1152/physiol.00012.2011 133. Kondratova A.A., Kondratov R.V.. The circadian clock and pathology of the ageing brain. Nat. Rev. Neurosci. (2012) 13 325-335. DOI: 10.1038/nrn3208 134. Masri S.. Sirtuin-dependent clock control: New advances in metabolism, aging and cancer. Curr. Opin. Clin. Nutr. Metab. Care (2015) 18 521-527. DOI: 10.1097/MCO.0000000000000219 135. Samoilova E.M., Belopasov V.V., Ekusheva E.V., Zhang C., Troitskiy A.V., Baklaushev V.P.. Epigenetic Clock and Circadian Rhythms in Stem Cell Aging and Rejuvenation. J. Pers. Med. (2021) 11. DOI: 10.3390/jpm11111050 136. Lehmann M., Canatelli-Mallat M., Chiavellini P., Goya R.G.. A hierarchical model for the control of epigenetic aging in mammals. Ageing Res. Rev. (2020) 62 101134. DOI: 10.1016/j.arr.2020.101134 137. Paixao L., Sikka P., Sun H., Jain A., Hogan J., Thomas R., Westover M.B.. Excess brain age in the sleep electroencephalogram predicts reduced life expectancy. Neurobiol. Aging (2020) 88 150-155. DOI: 10.1016/j.neurobiolaging.2019.12.015 138. Elliott M.L., Belsky D.W., Knodt A.R., Ireland D., Melzer T.R., Poulton R., Ramrakha S., Caspi A., Moffitt T.E., Hariri A.R.. Brain-age in midlife is associated with accelerated biological aging and cognitive decline in a longitudinal birth cohort. Mol. Psychiatry (2021) 26 3829-3838. DOI: 10.1038/s41380-019-0626-7 139. Cole J.H., Marioni R.E., Harris S.E., Deary I.J.. Brain age and other bodily ‘ages’: Implications for neuropsychiatry. Mol. Psychiatry (2019) 24 266-281. DOI: 10.1038/s41380-018-0098-1 140. Nenadić I., Dietzek M., Langbein K., Sauer H., Gaser C.. BrainAGE score indicates accelerated brain aging in schizophrenia, but not bipolar disorder. Psychiatry Res. Neuroimaging (2017) 266 86-89. DOI: 10.1016/j.pscychresns.2017.05.006 141. Pardoe H.R., Cole J.H., Blackmon K., Thesen T., Kuzniecky R.. Structural brain changes in medically refractory focal epilepsy resemble premature brain aging. Epilepsy Res. (2017) 133 28-32. DOI: 10.1016/j.eplepsyres.2017.03.007 142. Higgins-Chen A.T., Boks M.P., Vinkers C.H., Kahn R.S., Levine M.E.. Schizophrenia and Epigenetic Aging Biomarkers: Increased Mortality, Reduced Cancer Risk, and Unique Clozapine Effects. Biol. Psychiatry (2020) 88 224-235. DOI: 10.1016/j.biopsych.2020.01.025 143. Ori A.P.S., Olde Loohuis L.M., Guintivano J., Hannon E., Dempster E., St. Clair D., Bass N.J., McQuillin A., Mill J., Sullivan P.F.. Epigenetic age is accelerated in schizophrenia with age- and sex-specific effects and associated with polygenic disease risk. bioRxiv (2021) 727859. DOI: 10.1101/727859 144. Han L., Aghajani M., Clark S.L., Chan R., Hattab M.W., Shabalin A.A., Zhao M., Kumar G., Xie L.Y., Jansen R.. Epigenetic Aging in Major Depressive Disorder. Am. J. Psychiatry (2018) 175 774-782. DOI: 10.1176/appi.ajp.2018.17060595 145. Protsenko E., Yang R., Nier B., Reus V., Hammamieh R., Rampersaud R., Wu G.W.Y., Hough C.M., Epel E., Prather A.A.. “GrimAge,” an epigenetic predictor of mortality, is accelerated in major depressive disorder. Transl. Psychiatry (2021) 11 193. DOI: 10.1038/s41398-021-01302-0 146. Yang R., Wu G.W.Y., Verhoeven J.E., Gautam A., Reus V.I., Kang J.I., Flory J.D., Abu-Amara D., Hood L., Doyle F.J.. A DNA methylation clock associated with age-related illnesses and mortality is accelerated in men with combat PTSD. Mol. Psychiatry (2020) 26 4999-5009. DOI: 10.1038/s41380-020-0755-z 147. Horvath S., Levine A.J.. HIV-1 Infection Accelerates Age According to the Epigenetic Clock. J. Infect. Dis. (2015) 212 1563-1573. DOI: 10.1093/infdis/jiv277 148. Pellegrini C., Pirazzini C., Sala C., Sambati L., Yusipov I., Kalyakulina A., Ravaioli F., Kwiatkowska K.M., Durso D.F., Ivanchenko M.. A Meta-Analysis of Brain DNA Methylation Across Sex, Age, and Alzheimer’s Disease Points for Accelerated Epigenetic Aging in Neurodegeneration. Front. Aging Neurosci. (2021) 13 639428. DOI: 10.3389/fnagi.2021.639428 149. Horvath S., Langfelder P., Kwak S., Aaronson J., Rosinski J., Vogt T.F., Eszes M., Faull R.L.M., Curtis M.A., Waldvogel H.J.. Huntington’s disease accelerates epigenetic aging of human brain and disrupts DNA methylation levels. Aging (2016) 8 1485-1512. DOI: 10.18632/aging.101005 150. Fingelkurts A.A., Fingelkurts A.A.. Operational Architectonics of the Human Brain Biopotential Field: Towards Solving the Mind-Brain Problem. Brain Mind. (2001) 2 261-296. DOI: 10.1023/A:1014427822738 151. Fingelkurts A.A., Fingelkurts A.A., Neves C.F.H.. Natural world physical, brain operational, and mind phenomenal space–time. Phys. Life Rev. (2010) 7 195-249. DOI: 10.1016/j.plrev.2010.04.001 152. Fingelkurts A.A., Fingelkurts A.A., Neves C.F.H.. Consciousness as a phenomenon in the operational architectonics of brain organization: Criticality and self-organization considerations. Chaos Solitons Fractals (2013) 55 13-31. DOI: 10.1016/j.chaos.2013.02.007 153. Mitina M., Young S., Zhavoronkov A.. Psychological aging, depression, and well-being. Aging (2020) 12 18765-18777. DOI: 10.18632/aging.103880 154. Zhavoronkov A., Kochetov K., Diamandis P., Mitina M.. PsychoAge and SubjAge: Development of deep markers of psychological and subjective age using artificial intelligence. Aging (2020) 12 23548-23577. DOI: 10.18632/aging.202344 155. Westerhof G.J., Wurm S., Diehl M., Wahl H.-W.. Longitudinal Research on Subjective Aging, Health, and Longevity: Current Evidence and New Directions for Research. Annual Review of Gerontology and Geriatrics (2015) Volume 35 145-165 156. Stephan Y., Sutin A.R., Luchetti M., Terracciano A.. An older subjective age is related to accelerated epigenetic aging. Psychol. Aging (2021) 36 767-772. DOI: 10.1037/pag0000607 157. Stephan Y., Sutin A.R., Terracciano A.. Subjective Age and Mortality in Three Longitudinal Samples. Psychosom. Med. (2018) 80 659-664. DOI: 10.1097/PSY.0000000000000613 158. Lahav Y., Avidor S., Stein J.Y., Zhou X., Solomon Z.. Telomere Length and Depression Among Ex-Prisoners of War: The Role of Subjective Age. J. Gerontol. Ser. B (2020) 75 21-29. DOI: 10.1093/geronb/gby006 159. Preston R.J.. Telomeres, Telomerase and Chromosome Stability. Radiat. Res. (1997) 147 529. DOI: 10.2307/3579618 160. Hayflick L.. The limited in vitro lifetime of human diploid cell strains. Exp. Cell Res. (1965) 37 614-636. DOI: 10.1016/0014-4827(65)90211-9 161. Herbig U., Ferreira M., Condel L., Carey D., Sedivy J.M.. Cellular Senescence in Aging Primates. Science (2006) 311 1257. DOI: 10.1126/science.1122446 162. Campisi J., d’Adda di Fagagna F.. Cellular senescence: When bad things happen to good cells. Nat. Rev. Mol. Cell Biol. (2007) 8 729-740. DOI: 10.1038/nrm2233 163. Keyes C.L., Westerhof G.J.. Chronological and subjective age differences in flourishing mental health and major depressive episode. Aging Ment. Health (2012) 16 67-74. DOI: 10.1080/13607863.2011.596811 164. Choi N.G., DiNitto D.M.. Felt age and cognitive-affective depressive symptoms in late life. Aging Ment. Health (2014) 18 833-837. DOI: 10.1080/13607863.2014.886669 165. Segel-Karpas D., Palgi Y., Shrira A.. The reciprocal relationship between depression and physical morbidity: The role of subjective age. Health Psychol. (2017) 36 848-851. DOI: 10.1037/hea0000542 166. Stephan Y., Caudroit J., Jaconelli A., Terracciano A.. Subjective Age and Cognitive Functioning: A 10-Year Prospective Study. Am. J. Geriatr. Psychiatry (2014) 22 1180-1187. DOI: 10.1016/j.jagp.2013.03.007 167. Stephan Y., Sutin A.R., Terracciano A.. Subjective Age and Personality Development: A 10-Year Study. J. Personal. (2015) 83 142-154. DOI: 10.1111/jopy.12090 168. Kwak S., Kim H., Chey J., Youm Y.. Feeling How Old I Am: Subjective Age Is Associated With Estimated Brain Age. Front. Aging Neurosci. (2018) 10 168. DOI: 10.3389/fnagi.2018.00168 169. Elliott M.L., Caspi A., Houts R.M., Ambler A., Broadbent J.M., Hancox R.J., Harrington H.L., Hogan S., Keenan R., Knodt A.. Disparities in the pace of biological aging among midlife adults of the same chronological age have implications for future frailty risk and policy. Nat. Aging (2021) 1 295-308. DOI: 10.1038/s43587-021-00044-4 170. Horowitz A.M., Villeda S.A.. Therapeutic potential of systemic brain rejuvenation strategies for neurodegenerative disease. F1000Research (2017) 6 1291. DOI: 10.12688/f1000research.11437.1 171. Ewing G., Ewing E., Hankey A.. Virtual scanning: A new method of health assessment and treatment. Part I: Assessment. J. Altern. Complement. Med. (2007) 13 271-285. DOI: 10.1089/acm.2006.5207 172. Besedovsky H.O., del Rey A.. Immune-neuro-endocrine interactions: Facts and hypotheses. Endocr. Rev. (1996) 17 64-102. DOI: 10.1210/edrv-17-1-64 173. Harbuz M.. Neuroendocrine-immune interactions. Trends Endocrinol. Metab. (2003) 14 51-52. DOI: 10.1016/S1043-2760(03)00007-9 174. Fingelkurts A.A., Fingelkurts A.A.. Brain space and time in mental disorders: Paradigm shift in biological psychiatry. Int. J. Psychiatry Med. (2019) 54 53-63. DOI: 10.1177/0091217418791438 175. Jellinger K., Kurt A.. Vienna; Austria Cerebral Multimorbidity in Aging. J. Neurol. Neuromed. (2016) 1 1-5. DOI: 10.29245/2572.942X/2017/1.1001 176. Dow W.H., Philipson T.J., Sala-I-Martin X.. Longevity Complementarities Under Competing Risks. Am. Econ. Rev. (1999) 89 1358-1371. DOI: 10.1257/aer.89.5.1358 177. Bonkowski M.S., Sinclair D.A.. Slowing ageing by design: The rise of NAD+ and sirtuin-activating compounds. Nat. Rev. Mol. Cell Biol. (2016) 17 679-690. DOI: 10.1038/nrm.2016.93 178. Guarente L.. Calorie restriction and sirtuins revisited. Genes Dev. (2013) 27 2072-2085. DOI: 10.1101/gad.227439.113 179. Zhu X.-H., Lu M., Lee B.-Y., Ugurbil K., Chen W.. In vivo NAD assay reveals the intracellular NAD contents and redox state in healthy human brain and their age dependences. Proc. Natl. Acad. Sci. USA (2015) 112 2876-2881. DOI: 10.1073/pnas.1417921112 180. Lautrup S., Sinclair D.A., Mattson M.P., Fang E.F.. NAD+ in Brain Aging and Neurodegenerative Disorders. Cell Metab. (2019) 30 630-655. DOI: 10.1016/j.cmet.2019.09.001 181. Satoh A., Brace C.S., Rensing N., Cliften P., Wozniak D.F., Herzog E.D., Yamada K.A., Imai S.-I.. Sirt1 Extends Life Span and Delays Aging in Mice through the Regulation of Nk2 Homeobox 1 in the DMH and LH. Cell Metab. (2013) 18 416-430. DOI: 10.1016/j.cmet.2013.07.013 182. Mamoshina P., Kochetov K., Putin E., Cortese F., Aliper A., Lee W.-S., Ahn S.-M., Uhn L., Skjodt N., Kovalchuk O.. Population Specific Biomarkers of Human Aging: A Big Data Study Using South Korean, Canadian, and Eastern European Patient Populations. J. Gerontol. Ser. A (2018) 73 1482-1490. DOI: 10.1093/gerona/gly005 183. Mamoshina P., Volosnikova M., Ozerov I.V., Putin E., Skibina E., Cortese F., Zhavoronkov A.. Machine Learning on Human Muscle Transcriptomic Data for Biomarker Discovery and Tissue-Specific Drug Target Identification. Front. Genet. (2018) 9 242. DOI: 10.3389/fgene.2018.00242 184. Galkin F., Mamoshina P., Aliper A., Putin E., Moskalev V., Gladyshev V.N., Zhavoronkov A.. Human Gut Microbiome Aging Clock Based on Taxonomic Profiling and Deep Learning. iScience (2020) 23 101199. DOI: 10.1016/j.isci.2020.101199 185. Horvath S.. DNA methylation age of human tissues and cell types. Genome Biol. (2013) 14 R115. DOI: 10.1186/gb-2013-14-10-r115 186. Hannum G., Guinney J., Zhao L., Zhang L., Hughes G., Sadda S., Klotzle B., Bibikova M., Fan J.-B., Gao Y.. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Mol. Cell (2013) 49 359-367. DOI: 10.1016/j.molcel.2012.10.016 187. Lu A.T., Quach A., Wilson J.G., Reiner A.P., Aviv A., Raj K., Hou L., Baccarelli A.A., Li Y., Stewart J.D.. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (2019) 11 303-327. DOI: 10.18632/aging.101684 188. Carmona J.J., Michan S.. Biology of Healthy Aging and Longevity. Rev. Investig. Clin. (2016) 68 7-16. PMID: 27028172 189. Gomez-Verjan J.C., Vazquez-Martinez E.R., Rivero-Segura N.A., Medina-Campos R.H.. The RNA world of human ageing. Hum. Genet. (2018) 137 865-879. DOI: 10.1007/s00439-018-1955-3 190. Shireby G.L., Davies J.P., Francis P.T., Burrage J., Walker E.M., Neilson G.W.A., Dahir A., Thomas A.J., Love S., Smith R.G.. Recalibrating the epigenetic clock: Implications for assessing biological age in the human cortex. Brain (2020) 143 3763-3775. DOI: 10.1093/brain/awaa334 191. Lunnon K., Smith R., Hannon E., De Jager P.L., Srivastava G., Volta M., Troakes C., Al-Sarraj S., Burrage J., Macdonald R.. Methylomic profiling implicates cortical deregulation of ANK1 in Alzheimer’s disease. Nat. Neurosci. (2014) 17 1164-1170. DOI: 10.1038/nn.3782 192. Yu L., Chibnik L.B., Srivastava G.P., Pochet N., Yang J., Xu J., Kozubek J., Obholzer N., Leurgans S.E., Schneider J.A.. Association of brain DNA methylation in SORL1, ABCA7, HLA-DRB5, SLC24A4, and BIN1 with pathological diagnosis of Alzheimer disease. JAMA Neurol. (2015) 72 15-24. DOI: 10.1001/jamaneurol.2014.3049 193. Levine M.E., Lu A.T., Bennett D.A., Horvath S.. Epigenetic age of the prefrontal cortex is associated with neuritic plaques, amyloid load, and Alzheimer’s disease related cognitive functioning. Aging (2015) 7 1198-1211. DOI: 10.18632/aging.100864 194. Klein H.-U., Bennett D.A., De Jager P.L.. The epigenome in Alzheimer’s disease: Current state and approaches for a new path to gene discovery and understanding disease mechanism. Acta Neuropathol. (2016) 132 503-514. DOI: 10.1007/s00401-016-1612-7 195. El Khoury L.Y., Gorrie-Stone T., Smart M., Hughes A., Bao Y., Andrayas A., Burrage J., Hannon E., Kumari M., Mill J.. Systematic underestimation of the epigenetic clock and age acceleration in older subjects. Genome Biol. (2019) 20 283. DOI: 10.1186/s13059-019-1810-4 196. Teeuw J., Ori A.P.S., Brouwer R.M., de Zwarte S.M.C., Schnack H.G., Hulshoff Pol H.E., Ophoff R.A.. Accelerated aging in the brain, epigenetic aging in blood, and polygenic risk for schizophrenia. Schizophr. Res. (2021) 231 189-197. DOI: 10.1016/j.schres.2021.04.005 197. Purves D., Augustine G.J., Fitzpatrick D., Hall W.C., LaMantia A.S., McNamara J.O., White L.E.. Neuroscience (2008) 198. Fjell A.M., Walhovd K.B.. Structural Brain Changes in Aging: Courses, Causes and Cognitive Consequences. Rev. Neurosci. (2010) 21 187-222. DOI: 10.1515/REVNEURO.2010.21.3.187 199. Global, regional, and national burden of neurological disorders, 1990–2016: A systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. (2019) 18 459-480. DOI: 10.1016/S1474-4422(18)30499-X 200. Svennerholm L., Boström K., Jungbjer B.. Changes in weight and compositions of major membrane components of human brain during the span of adult human life of Swedes. Acta Neuropathol. (1997) 94 345-352. DOI: 10.1007/s004010050717 201. Andrews-Hanna J.R., Snyder A.Z., Vincent J.L., Lustig C., Head D., Raichle M.E., Buckner R.L.. Disruption of Large-Scale Brain Systems in Advanced Aging. Neuron (2007) 56 924-935. DOI: 10.1016/j.neuron.2007.10.038 202. Schaie K.W.. The course of adult intellectual development. Am. Psychol. (1994) 49 304-313. DOI: 10.1037/0003-066X.49.4.304 203. Park D.C., Lautenschlager G., Hedden T., Davidson N.S., Smith A.D., Smith P.K.. Models of visuospatial and verbal memory across the adult life span. Psychol. Aging (2002) 17 299-320. DOI: 10.1037/0882-7974.17.2.299 204. Dickie D.A., Job D.E., Gonzalez D.R., Shenkin S.D., Ahearn T.S., Murray A.D., Wardlaw J.M.. Variance in brain volume with advancing age: Implications for defining the limits of normality. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0084093 205. Bittner N., Jockwitz C., Franke K., Gaser C., Moebus S., Bayen U.J., Amunts K., Caspers S.. When your brain looks older than expected: Combined lifestyle risk and BrainAGE. Anat. Embryol. (2021) 226 621-645. DOI: 10.1007/s00429-020-02184-6 206. Jagust W.. Vulnerable Neural Systems and the Borderland of Brain Aging and Neurodegeneration. Neuron (2013) 77 219-234. DOI: 10.1016/j.neuron.2013.01.002 207. Efranke K., Eristow M., Egaser C.. Gender-specific impact of personal health parameters on individual brain aging in cognitively unimpaired elderly subjects. Front. Aging Neurosci. (2014) 6 94. DOI: 10.3389/fnagi.2014.00094 208. Eavani H., Habes M., Satterthwaite T.D., An Y., Hsieh M.-K., Honnorat N., Erus G., Doshi J., Ferrucci L., Beason-Held L.L.. Heterogeneity of structural and functional imaging patterns of advanced brain aging revealed via machine learning methods. Neurobiol. Aging (2018) 71 41-50. DOI: 10.1016/j.neurobiolaging.2018.06.013 209. Bartrés-Faz D., Arenaza-Urquijo E.M.. Structural and Functional Imaging Correlates of Cognitive and Brain Reserve Hypotheses in Healthy and Pathological Aging. Brain Topogr. (2011) 24 340-357. DOI: 10.1007/s10548-011-0195-9 210. Ziegler G., Dahnke R., Gaser C.. Models of the aging brain structure and individual decline. Front. Neuroinform. (2012) 6 3. DOI: 10.3389/fninf.2012.00003 211. Sperling R., Mormino E., Johnson K.. The Evolution of Preclinical Alzheimer’s Disease: Implications for Prevention Trials. Neuron (2014) 84 608-622. DOI: 10.1016/j.neuron.2014.10.038 212. Elliott M.L.. MRI-Based biomarkers of accelerated aging and dementia risk in midlife: How close are we?. Ageing Res. Rev. (2020) 61 101075. DOI: 10.1016/j.arr.2020.101075 213. Franke K., Ziegler G., Klöppel S., Gaser C.. Estimating the age of healthy subjects from T1-weighted MRI scans using kernel methods: Exploring the influence of various parameters. NeuroImage (2010) 50 883-892. DOI: 10.1016/j.neuroimage.2010.01.005 214. Cole J.H., Franke K.. Predicting Age Using Neuroimaging: Innovative Brain Ageing Biomarkers. Trends Neurosci. (2017) 40 681-690. DOI: 10.1016/j.tins.2017.10.001 215. Cole J.H., Poudel R.P., Tsagkrasoulis D., Caan M.W., Steves C., Spector T.D., Montana G.. Predicting brain age with deep learning from raw imaging data results in a reliable and heritable biomarker. NeuroImage (2017) 163 115-124. DOI: 10.1016/j.neuroimage.2017.07.059 216. Liem F., Varoquaux G., Kynast J., Beyer F., Masouleh S.K., Huntenburg J.M., Lampe L., Rahim M., Abraham A., Craddock R.C.. Predicting brain-age from multimodal imaging data captures cognitive impairment. Neuroimage (2017) 148 179-188. DOI: 10.1016/j.neuroimage.2016.11.005 217. Enzinger C., Fazekas F., Matthews P.M., Ropele S., Schmidt H., Smith S., Schmidt R.. Risk factors for progression of brain atrophy in aging: Six-year follow-up of normal subjects. Neurology (2005) 64 1704-1711. DOI: 10.1212/01.WNL.0000161871.83614.BB 218. Franke K., Gaser C., Manor B., Novak V.. Advanced BrainAGE in older adults with type 2 diabetes mellitus. Front. Ageing Neurosci. (2013) 5 90 219. Fingelkurts A.A., Fingelkurts A.A., Bagnato S., Boccagni C., Galardi G.. Dissociation of vegetative and minimally conscious patients based on brain Operational Architectonics: Factor of etiology. Clin. EEG Neurosci. (2013) 44 209-220. DOI: 10.1177/1550059412474929 220. Rudrauf D.. Structure-Function Relationships behind the Phenomenon of Cognitive Resilience in Neurology: Insights for Neuroscience and Medicine. Adv. Neurosci. (2014) 2014 462765. DOI: 10.1155/2014/462765 221. Luria A.. The Man with a Shattered World: The History of a Brain Wound (1987) 222. Lee R.G., van Donkelaar P.. Mechanisms underlying functional recovery following stroke. Can. J. Neurol. Sci. (1995) 22 257-263. DOI: 10.1017/S0317167100039445 223. Noppeney U., Friston K.J., Price C.J.. Degenerate neuronal systems sustaining cognitive functions. J. Anat. (2004) 205 433-442. DOI: 10.1111/j.0021-8782.2004.00343.x 224. Philippi C.L., Feinstein J.S., Khalsa S.S., Damasio A., Tranel D., Landini G., Williford K., Rudrauf D.. Preserved Self-Awareness following Extensive Bilateral Brain Damage to the Insula, Anterior Cingulate, and Medial Prefrontal Cortices. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0038413 225. Bagnato S., Boccagni C., Sant’Angelo A., Fingelkurts A.A., Fingelkurts A.A., Galardi G.. Emerging from an unresponsive wakefulness syndrome: Brain plasticity has to cross a threshold level. Neurosci. Biobehav. Rev. (2013) 37 2721-2736. DOI: 10.1016/j.neubiorev.2013.09.007 226. Fingelkurts A.A., Fingelkurts A.A.. Longitudinal dynamics of 3-dimensional components of selfhood after severe traumatic brain injury: A qEEG case study. Clin. EEG Neurosci. (2017) 48 327-337. DOI: 10.1177/1550059417696180 227. Price J.L., Morris J.C.. Tangles and plaques in nondemented ageing and “preclinical” Alzheimer’s disease. Ann. Neurol. (1999) 45 358-368. DOI: 10.1002/1531-8249(199903)45:3<358::AID-ANA12>3.0.CO;2-X 228. Dubois B., Feldman H.H., Jacova C., Cummings J.L., DeKosky S.T., Barberger-Gateau P., Delacourte A., Frisoni G., Fox N.C., Galasko D.. Revising the definition of Alzheimer’s disease: A new lexicon. Lancet Neurol. (2010) 9 1118-1127. DOI: 10.1016/S1474-4422(10)70223-4 229. Penfield W., Field J.. Neurophysiological Basis of the Higher Functions of the Nervous System—Introduction. Handbook of Physiology. A Critical, Comprehensive Presentation of Physiological Knowledge and Concepts (1960) 1441-1445 230. Lopes da Silva F.H.. Neural mechanisms underlying brain waves: From neural membranes to networks. Electroencephalogr. Clin. Neurophysiol (1991) 79 81-93. DOI: 10.1016/0013-4694(91)90044-5 231. Nunez P.L., Williamson S.J.. Neocortical Dynamics and Human EEG Rhythms (1995) 232. Nuwer M.. Assessment of digital EEG, quantitative EEG, and EEG brain mapping: Report of the American Academy of Neurology and the American Clinical Neurophysiology Society. Neurology (1997) 49 277-292. DOI: 10.1212/WNL.49.1.277 233. Gasser T., Bächer P., Steinberg H.. Test-retest reliability of spectral parameters of the EEG. Electroencephalogr. Clin. Neurophysiol. (1985) 60 312-319. DOI: 10.1016/0013-4694(85)90005-7 234. Stassen H.H., Bomben G., Propping P.. Genetic aspects of the EEG: An investigation into the within-pair similarity of monozigotic and dyzigotic twins with a new method of analysis. Electroencephalogr. Clin. Neurophysiol. (1987) 66 489-501. DOI: 10.1016/0013-4694(87)90095-2 235. Stassen H.H., Bomben G., Hell D.. Familial brain wave patterns: Study of a 12-sib family. Psychiatr. Genet. (1998) 8 141-153. DOI: 10.1097/00041444-199800830-00003 236. Pollock V., Schneider L., Lyness S.. Reliability of topographic quantitative EEG amplitude in healthy late-middle-aged and elderly subjects. Electroencephalogr. Clin. Neurophysiol. (1991) 79 20-26. DOI: 10.1016/0013-4694(91)90152-T 237. Fingelkurts A.A., Fingelkurts A.A., Ermolaev V.A., Kaplan A.Y.. Stability, reliability and consistency of the compositions of brain oscillations. Int. J. Psychophysiol. (2006) 59 116-126. DOI: 10.1016/j.ijpsycho.2005.03.014 238. Posthuma D., Neale M.C., Boomsma D.I., de Geus E.J.C.. Are Smarter Brains Running Faster? Heritability of Alpha Peak Frequency, IQ, and Their Interrelation. Behav. Genet. (2001) 31 567-579. DOI: 10.1023/A:1013345411774 239. van Beijsterveldt C., van Baal G.. Twin and family studies of the human electroencephalogram: A review and a meta-analysis. Biol. Psychol. (2002) 61 111-138. DOI: 10.1016/S0301-0511(02)00055-8 240. Anokhin A.P., Müller V., Lindenberger U., Heath A.C., Myers E.. Genetic influences on dynamic complexity of brain oscillations. Neurosci. Lett. (2006) 397 93-98. DOI: 10.1016/j.neulet.2005.12.025 241. Nyberg L., Pudas S.. Successful Memory Aging. Annu. Rev. Psychol. (2019) 70 219-243. DOI: 10.1146/annurev-psych-010418-103052 242. da Silva F.L., Vos J., Mooibroek J., van Rotterdam A.. Relative contributions of intracortical and thalamo-cortical processes in the generation of alpha rhythms, revealed by partial coherence analysis. Electroencephalogr. Clin. Neurophysiol. (1980) 50 449-456. DOI: 10.1016/0013-4694(80)90011-5 243. Bhattacharya B.S., Coyle D., Maguire L.P.. A thalamo-cortico-thalamic neural mass model to study alpha rhythms in Alzheimer’s disease. Neural Netw. (2011) 24 631-645. DOI: 10.1016/j.neunet.2011.02.009 244. Garcés P., Vicente R., Wibral M., Pineda-Pardo J., López M.E., Aurtenetxe S., Marcos A., de Andrés M.E., Yus M., Sancho M.. Brain-wide slowing of spontaneous alpha rhythms in mild cognitive impairment. Front. Aging Neurosci. (2013) 5 100. DOI: 10.3389/fnagi.2013.00100 245. Klimesch W.. EEG alpha and theta oscillations reflect cognitive and memory performance: A review and analysis. Brain Res. Rev. (1999) 29 169-195. DOI: 10.1016/S0165-0173(98)00056-3 246. Bressler S.L., Kelso J.. Cortical coordination dynamics and cognition. Trends Cogn. Sci. (2001) 5 26-36. DOI: 10.1016/S1364-6613(00)01564-3 247. Fingelkurts A.A., Fingelkurts A.A.. Short-term EEG spectral pattern as a single event in EEG phenomenology. Open Neuroimaging J. (2010) 4 130-156. DOI: 10.2174/1874440001004010130 248. Başar E.. A review of alpha activity in integrative brain function: Fundamental physiology, sensory coding, cognition and pathology. Int. J. Psychophysiol. (2012) 86 1-24. DOI: 10.1016/j.ijpsycho.2012.07.002 249. Matejcek M.. Some relationships between occipital E.E.G. activity and age. A spectral analytic study. Rev. Electroencephalogr. Neurophysiol. Clin. (1980) 10 122-130. DOI: 10.1016/S0370-4475(80)80044-X 250. Köpruner V., Pfurtscheller G., Auer L.. Quantitative EEG in Normals and in Patients with Cerebral Ischemia. Prog. Brain Res. (1984) 62 29-50. DOI: 10.1016/s0079-6123(08)62168-8 251. Babiloni C., Binetti G., Cassarino A., Forno G.D., Del Percio C., Ferreri F., Ferri R., Frisoni G.B., Galderisi S., Hirata K.. Sources of cortical rhythms in adults during physiological aging: A multicentric EEG study. Hum. Brain Mapp. (2006) 27 162-172. DOI: 10.1002/hbm.20175 252. Knyazeva M.G., Barzegaran E., Vildavski V.Y., Demonet J.-F.. Aging of human alpha rhythm. Neurobiol. Aging (2018) 69 261-273. DOI: 10.1016/j.neurobiolaging.2018.05.018 253. Purdon P.L., Pavone K.J., Akeju O., Smith A.C., Sampson A.L., Lee J., Zhou D.W., Solt K., Brown E.N.. The Ageing Brain: Age-dependent changes in the electroencephalogram during propofol and sevoflurane general anaesthesia. Br. J. Anaesth. (2015) 115 i46-i57. DOI: 10.1093/bja/aev213 254. Dukart J., Schroeter M.L., Mueller K.. Age Correction in Dementia—Matching to a Healthy Brain. PLoS ONE (2011) 6. DOI: 10.1371/journal.pone.0022193 255. Szelies B., Mielke R., Kessler J., Heiss W.D.. EEG power changes are related with regional cerebral glucose metbolism in vascular dementia. Clin. Neurophysiol. (1999) 110 615-620. DOI: 10.1016/S1388-2457(98)00052-2 256. Leocani L., Locatelli T., Martinelli V., Rovaris M., Falautano M., Filippi M., Magnani G., Comi G.. Electroencephalographic coherence analysis in multiple sclerosis: Correlation with clinical, neuropsychological, and MRI findings. J. Neurol. Neurosurg. Psychiatry (2000) 69 192-198. DOI: 10.1136/jnnp.69.2.192 257. Fernández A., Hornero R., Mayo A., Poza J., Gil-Gregorio P., Ortiz T.. EEG spectral profile in Alzheimer’s disease and mild cognitive impairment. Clin. Neurophysiol. (2006) 117 306-314. DOI: 10.1016/j.clinph.2005.10.017 258. Medaglia J.D., Pasqualetti F., Hamilton R.H., Thompson-Schill S.L., Bassett D.S.. Brain and cognitive reserve: Translation via network control theory. Neurosci. Biobehav. Rev. (2017) 75 53-64. DOI: 10.1016/j.neubiorev.2017.01.016 259. Stern Y.. Cognitive Reserve and Alzheimer Disease. Alzheimer Dis. Assoc. Disord. (2006) 20 112-117. DOI: 10.1097/01.wad.0000213815.20177.19 260. Cespón J., Miniussi C., Pellicciari M.C.. Interventional programmes to improve cognition during healthy and pathological ageing: Cortical modulations and evidence for brain plasticity. Ageing Res. Rev. (2018) 43 81-98. DOI: 10.1016/j.arr.2018.03.001 261. Lã3Pez M.E., Aurtenetxe S., Pereda E., Cuesta P., Castellanos N., Bruã±A R., Niso G., Maestu F., Bajo R.. Cognitive reserve is associated with the functional organization of the brain in healthy aging: A MEG study. Front. Aging Neurosci. (2014) 6 125. DOI: 10.3389/fnagi.2014.00125 262. Zarahn E., Rakitin B., Abela D., Flynn J., Stern Y.. Age-related changes in brain activation during a delayed item recognition task. Neurobiol. Aging (2007) 28 784-798. DOI: 10.1016/j.neurobiolaging.2006.03.002 263. Steffener J., Stern Y.. Exploring the neural basis of cognitive reserve in aging. Biochim. Biophys. Acta Mol. Basis Dis. (2012) 1822 467-473. DOI: 10.1016/j.bbadis.2011.09.012 264. Stern Y., Habeck C., Moeller J., Scarmeas N., Anderson K.E., Hilton H.J., Flynn J., Sackeim H., van Heertum R.. Brain Networks Associated with Cognitive Reserve in Healthy Young and Old Adults. Cereb. Cortex (2005) 15 394-402. DOI: 10.1093/cercor/bhh142 265. Al Zoubi O., Wong C.K., Kuplicki R.T., Yeh H.-W., Mayeli A., Refai H., Paulus M., Bodurka J.. Predicting Age From Brain EEG Signals—A Machine Learning Approach. Front. Aging Neurosci. (2018) 10 184. DOI: 10.3389/fnagi.2018.00184 266. Sun H., Paixao L., Oliva J.T., Goparaju B., Carvalho D., van Leeuwen K.G., Akeju O., Thomas R.J., Cash S.S., Bianchi M.T.. Brain age from the electroencephalogram of sleep. Neurobiol. Aging (2019) 74 112-120. DOI: 10.1016/j.neurobiolaging.2018.10.016 267. Gontier G., Iyer M., Shea J.M., Bieri G., Wheatley E.G., Ramalho-Santos M., Villeda S.A.. Tet2 Rescues Age-Related Regenerative Decline and Enhances Cognitive Function in the Adult Mouse Brain. Cell Rep. (2018) 22 1974-1981. DOI: 10.1016/j.celrep.2018.02.001 268. Fahy G.M., Brooke R.T., Watson J.P., Good Z., Vasanawala S.S., Maecker H., Leipold M.D., Lin D.T.S., Kobor M.S., Horvath S.. Reversal of epigenetic aging and immunosenescent trends in humans. Aging Cell (2019) 18 e13028. DOI: 10.1111/acel.13028 269. Katsimpardi L., Litterman N.K., Schein P.A., Miller C.M., Loffredo F.S., Wojtkiewicz G.R., Chen J.W., Lee R.T., Wagers A.J., Rubin L.L.. Vascular and Neurogenic Rejuvenation of the Aging Mouse Brain by Young Systemic Factors. Science (2014) 344 630-634. DOI: 10.1126/science.1251141 270. Lindahl M., Saarma M., Lindholm P.. Unconventional neurotrophic factors CDNF and MANF: Structure, physiological functions and therapeutic potential. Neurobiol. Dis. (2017) 97 90-102. DOI: 10.1016/j.nbd.2016.07.009 271. White C.W., Pratt K., Villeda S.A.. OPCs on a Diet: A Youthful Serving of Remyelination. Cell Metab. (2019) 30 1004-1006. DOI: 10.1016/j.cmet.2019.11.009 272. Horowitz A.M., Fan X., Bieri G., Smith L.K., Sanchez-Diaz C.I., Schroer A.B., Gontier G., Casaletto K.B., Kramer J.H., Williams K.E.. Blood factors transfer beneficial effects of exercise on neurogenesis and cognition to the aged brain. Science (2020) 369 167-173. DOI: 10.1126/science.aaw2622 273. Navarro A., Gómez C.G., López-Cepero J.M., Boveris A.. Beneficial effects of moderate exercise on mice aging: Survival, behavior, oxidative stress, and mitochondrial electron transfer. Am. J. Physiol. Integr. Comp. Physiol. (2004) 286 R505-R511. DOI: 10.1152/ajpregu.00208.2003 274. Melov S., Tarnopolsky M.A., Beckman K., Felkey K., Hubbard A.. Resistance Exercise Reverses Aging in Human Skeletal Muscle. PLoS ONE (2007) 2. DOI: 10.1371/journal.pone.0000465 275. Poljsak B., Milisav I.. NAD+ as the Link Between Oxidative Stress, Inflammation, Caloric Restriction, Exercise, DNA Repair, Longevity, and Health Span. Rejuvenation Res. (2016) 19 406-413. DOI: 10.1089/rej.2015.1767 276. Quach A., Levine M.E., Tanaka T., Lu A.T., Chen B.H., Ferrucci L., Ritz B., Bandinelli S., Neuhouser M.L., Beasley J.M.. Epigenetic clock analysis of diet, exercise, education, and lifestyle factors. Aging (2017) 9 419-446. DOI: 10.18632/aging.101168 277. Masoro E.J.. Overview of caloric restriction and ageing. Mech. Ageing Dev. (2005) 126 913-922. DOI: 10.1016/j.mad.2005.03.012 278. Vera E., de Jesus B.B., Foronda M., Flores J.M., Blasco M.A.. Telomerase Reverse Transcriptase Synergizes with Calorie Restriction to Increase Health Span and Extend Mouse Longevity. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0053760 279. Waterland R.A., Jirtle R.L.. Transposable Elements: Targets for Early Nutritional Effects on Epigenetic Gene Regulation. Mol. Cell. Biol. (2003) 23 5293-5300. DOI: 10.1128/MCB.23.15.5293-5300.2003 280. Hore T.A.. Modulating epigenetic memory through vitamins and TET: Implications for regenerative medicine and cancer treatment. Epigenomics (2017) 9 863-871. DOI: 10.2217/epi-2017-0021 281. Alehagen U., Aaseth J., Alexander J., Johansson P.. Still reduced cardiovascular mortality 12 years after supplementation with selenium and coenzyme Q10 for four years: A validation of previous 10-year follow-up results of a prospective randomized double-blind placebo-controlled trial in elderly. PLoS ONE (2018) 13. DOI: 10.1371/journal.pone.0193120 282. Comhaire F., Decleer W.. Can the biological mechanisms of ageing be corrected by food supplementation. The concept of health care over sick care. Aging Male (2020) 23 1146-1157. DOI: 10.1080/13685538.2020.1713080 283. Proshkina E., Shaposhnikov M., Moskalev A.. Genome-Protecting Compounds as Potential Geroprotectors. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21124484 284. Nur S.M., Rath S., Ahmad V., Ahmad A., Ateeq B., Khan M.I.. Nutritive vitamins as epidrugs. Crit. Rev. Food Sci. Nutr. (2021) 61 1712674. DOI: 10.1080/10408398.2020.1712674 285. Stanfel M.N., Shamieh L.S., Kaeberlein M., Kennedy B.K.. The TOR pathway comes of age. Biochim. Biophys. Acta (2009) 1790 1067-1074. DOI: 10.1016/j.bbagen.2009.06.007 286. Johnson S.C., Rabinovitch P.S., Kaeberlein M.. mTOR is a key modulator of ageing and age-related disease. Nature (2013) 493 338-345. DOI: 10.1038/nature11861 287. Kennedy B., Lamming D.W.. The Mechanistic Target of Rapamycin: The Grand ConducTOR of Metabolism and Aging. Cell Metab. (2016) 23 990-1003. DOI: 10.1016/j.cmet.2016.05.009 288. Saxton R.A., Sabatini D.M.. mTOR Signaling in Growth, Metabolism, and Disease. Cell (2017) 168 960-976. DOI: 10.1016/j.cell.2017.02.004 289. Shea T.B., Remington R.. Nutritional supplementation for Alzheimer’s disease?. Curr. Opin. Psychiatry (2015) 28 141-147. DOI: 10.1097/YCO.0000000000000138 290. Dominguez L.J., Barbagallo M.. Nutritional prevention of cognitive decline and dementia. Acta Biomed. (2018) 89 276-290. PMID: 29957766 291. Corpas R., Griñán-Ferré C., Rodriguez-Farre E., Pallàs M., Sanfeliu C.. Resveratrol Induces Brain Resilience Against Alzheimer Neurodegeneration Through Proteostasis Enhancement. Mol. Neurobiol. (2019) 56 1502-1516. DOI: 10.1007/s12035-018-1157-y 292. Orozco-Solis R., Ramadori G., Coppari R., Sassone-Corsi P.. SIRT1 Relays Nutritional Inputs to the Circadian Clock Through the Sf1 Neurons of the Ventromedial Hypothalamus. Endocrinology (2015) 156 2174-2184. DOI: 10.1210/en.2014-1805 293. Imai S.-I., Guarente L.. NAD+ and sirtuins in aging and disease. Trends Cell Biol. (2014) 24 464-471. DOI: 10.1016/j.tcb.2014.04.002 294. Verdin E.. The Many Faces of Sirtuins: Coupling of NAD metabolism, sirtuins and lifespan. Nat. Med. (2014) 20 25-27. DOI: 10.1038/nm.3447 295. Chastin S.F.M., Dall P., Tigbe W., Grant P.M., Ryan C., Rafferty D., Granat M.. Compliance with physical activity guidelines in a group of UK-based postal workers using an objective monitoring technique. Eur. J. Appl. Physiol. (2009) 106 893-899. DOI: 10.1007/s00421-009-1090-x 296. Theofilou P., Saborit A.R.. Adherence and physical activity. Health Psychol. Res. (2013) 1 e6. DOI: 10.4081/hpr.2013.701 297. Gomes T.N., Katzmarzyk P.T., Hedeker D., Fogelholm M., Standage M., Onywera V., Lambert E.V., Tremblay M.S., Chaput J.-P., Tudor-Locke C.. Correlates of compliance with recommended levels of physical activity in children. Sci. Rep. (2017) 7 16507. DOI: 10.1038/s41598-017-16525-9 298. Heilbronn L.K., de Jonge L., Frisard M.I., DeLany J.P., Larson-Meyer D.E., Rood J., Nguyen T., Martin C.K., Volaufova J., Most M.M.. Effect of 6-month calorie restriction on biomarkers of longevity, metabolic adaptation, and oxidative stress in overweight individuals: A randomized controlled trial. JAMA (2006) 295 1539-1548. DOI: 10.1001/jama.295.13.1539 299. Ravussin E., Redman L.M., Rochon J., Das S.K., Fontana L., Kraus W.E., Romashkan S., Williamson D.A., Meydani S.N., Villareal D.T.. A 2-Year Randomized Controlled Trial of Human Caloric Restriction: Feasibility and Effects on Predictors of Health Span and Longevity. J. Gerontol. Ser. A (2015) 70 1097-1104. DOI: 10.1093/gerona/glv057 300. Romashkan S.V., Das S.K., Villareal D.T., Ravussin E., Redman L.M., Rochon J., Bhapkar M., Kraus W.E.. Safety of two-year caloric restriction in non-obese healthy individuals. Oncotarget (2016) 7 19124-19133. DOI: 10.18632/oncotarget.8093 301. O’Connor S.G., Boyd P., Bailey C.P., Shams-White M.M., Agurs-Collins T., Hall K., Reedy J., Sauter E.R., Czajkowski S.M.. Perspective: Time-Restricted Eating Compared with Caloric Restriction: Potential Facilitators and Barriers of Long-Term Weight Loss Maintenance. Adv. Nutr. Int. Rev. J. (2021) 12 325-333. DOI: 10.1093/advances/nmaa168 302. Madeo F., Carmona-Gutierrez D., Hofer S.J., Kroemer G.. Caloric Restriction Mimetics against Age-Associated Disease: Targets, Mechanisms, and Therapeutic Potential. Cell Metab. (2019) 29 592-610. DOI: 10.1016/j.cmet.2019.01.018 303. Guerrieri D., Moon H.Y., van Praag H.. Exercise in a Pill: The Latest on Exercise-Mimetics. Brain Plast. (2017) 2 153-169. DOI: 10.3233/BPL-160043 304. Fingelkurts A.A., Fingelkurts A.A., Neves C.F.H.. Neuro-assessment of leadership training. Coaching Int. J. Theory Res. Pract. (2020) 13 107-145. DOI: 10.1080/17521882.2019.1619796 305. Wilson K.A., Chamoli M., Hilsabeck T.A., Pandey M., Bansal S., Chawla G., Kapahi P.. Evaluating the beneficial effects of dietary restrictions: A framework for precision nutrigeroscience. Cell Metab. (2021) 33 2142-2173. DOI: 10.1016/j.cmet.2021.08.018 306. Kirkland J.L., Stout M.B., Sierra F.. Resilience in Aging Mice. J. Gerontol. Ser. A (2016) 71 1407-1414. DOI: 10.1093/gerona/glw086 307. Miller B.F., Seals D.R., Hamilton K.L.. A viewpoint on considering physiological principles to study stress resistance and resilience with aging. Ageing Res. Rev. (2017) 38 1-5. DOI: 10.1016/j.arr.2017.06.004 308. DE Grey A.D.N.J., Ames B.N., Andersen J.K., Bartke A., Campisi J., Heward C.B., McCARTER R.J.M., Stock G.. Time to Talk SENS: Critiquing the Immutability of Human Aging. Ann. N. Y. Acad. Sci. (2002) 959 452-462. DOI: 10.1111/j.1749-6632.2002.tb02115.x 309. Fingelkurts A.A., Fingelkurts A.A.. Operational Architectonics methodology for EEG analysis: Theory and results. Neuromethods (2015) 91 1-59 310. Kurgansky A.V.. Functional Organization of the Human Brain in the Resting State. Neurosci. Behav. Physiol. (2019) 49 1135-1144. DOI: 10.1007/s11055-019-00850-9 311. Buzsáki G.. Rhythms of the Brain (2006) 312. Lazarev V.V.. The relationship of theory and methodology in EEG studies of mental activity. Int. J. Psychophysiol. (2006) 62 384-393. DOI: 10.1016/j.ijpsycho.2006.01.006 313. Wacker J., Chavanon M.-L., Stemmler G.. Resting EEG signatures of agentic extraversion: New results and meta-analytic integration. J. Res. Pers. (2010) 44 167-179. DOI: 10.1016/j.jrp.2009.12.004 314. Rogala J., Kublik E., Krauz R., Wróbel A.. Resting-state EEG activity predicts frontoparietal network reconfiguration and improved attentional performance. Sci. Rep. (2020) 10 5064. DOI: 10.1038/s41598-020-61866-7 315. Fingelkurts A.A., Fingelkurts A.A., Kallio-Tamminen T.. Selfhood triumvirate: From phenomenology to brain activity and back again. Conscious. Cogn. (2020) 86 103031. DOI: 10.1016/j.concog.2020.103031 316. Ge Y., Grossman R.I., Babb J.S., Rabin M.L., Mannon L.J., Kolson D.L.. Age-Related Total Gray Matter and White Matter Changes in Normal Adult Brain. Part I: Volumetric MR Imaging Analysis. Am. J. Neuroradiol. (2002) 23 1327-1333. PMID: 12223373 317. Raz N., Lindenberger U., Rodrigue K.M., Kennedy K.M., Head D., Williamson A., Dahle C., Gerstorf D., Acker J.D.. Regional Brain Changes in Aging Healthy Adults: General Trends, Individual Differences and Modifiers. Cereb. Cortex (2005) 15 1676-1689. DOI: 10.1093/cercor/bhi044 318. Salthouse T.A.. When does age-related cognitive decline begin?. Neurobiol. Aging (2009) 30 507-514. DOI: 10.1016/j.neurobiolaging.2008.09.023 319. Singh-Manoux A., Kivimaki M., Glymour M.M., Elbaz A., Berr C., Ebmeier K., Ferrie J.E., Dugravot A.. Timing of onset of cognitive decline: Results from Whitehall II prospective cohort study. BMJ (2012) 344 d7622. DOI: 10.1136/bmj.d7622 320. Scraggs T.L.. EEG maturation: Viability through adolescence. Neurodiagnostic J. (2012) 52 176-203 321. Kaminska A., Eisermann M., Plouin P., Levin K.H., Chauvel P.. Child EEG (and Maturation). Handbook of Clinical Neurology (2019) Volume 160 125-142 322. Giedd J.N., Blumenthal J., Jeffries N.O., Castellanos F., Liu H., Zijdenbos A., Paus T., Evans A.C., Rapoport J.L.. Brain development during childhood and adolescence: A longitudinal MRI study. Nat. Neurosci. (1999) 2 861-863. DOI: 10.1038/13158 323. Sobel H.. When Does Human Aging Start?. Gerontologist (1966) 6 17-22. DOI: 10.1093/geront/6.1.17 324. Gordon E., Cooper N., Rennie C., Hermens D., Williams L.. Integrative Neuroscience: The Role of a Standardized Database. Clin. EEG Neurosci. (2005) 36 64-75. DOI: 10.1177/155005940503600205 325. Prichep L.S.. Use of Normative Databases and Statistical Methods in Demonstrating Clinical Utility of QEEG: Importance and Cautions. Clin. EEG Neurosci. (2005) 36 82-87. DOI: 10.1177/155005940503600207 326. Thatcher R.W., Lubar J.F., Budzinsky T., Budzinski H., Evans J., Abarbanel A.. History of the Scientific Standards of QEEG Normative Databases. Introduction to QEEG and Neurofeedback: Advanced Theory and Applications (2008) 29-62 327. Beck A.T., Ward C.H., Mendelson M., Mock J., Erbaugh J.. An inventory for measuring depression. Arch. Gen. Psychiatry (1961) 4 561-571. DOI: 10.1001/archpsyc.1961.01710120031004 328. Beck A.T., Epstein N., Brown G., Steer R.A.. An inventory for measuring clinical anxiety: Psychometric properties. J. Consult. Clin. Psychol. (1988) 56 893-897. DOI: 10.1037/0022-006X.56.6.893 329. Hamilton M.. The Assessment of Anxiety States by Rating. Psychol. Psychother. Theory Res. Pract. (1959) 32 50-55. DOI: 10.1111/j.2044-8341.1959.tb00467.x 330. John O.P., Donahue E.M., Kentle R.L.. The Big Five Inventory-Versions 4a and 54 (1991) 331. Potter G.D.M., Cade J., Grant P.J., Hardie L.J.. Nutrition and the circadian system. Br. J. Nutr. (2016) 116 434-442. DOI: 10.1017/S0007114516002117 332. Suga H., Asakura K., Sasaki S., Nojima M., Okubo H., Hirota N., Notsu A., Fukui M., Date C.. Effect of seasonality on the estimated mean value of nutrients and ranking ability of a self-administered diet history questionnaire. Nutr. J. (2014) 13 51. DOI: 10.1186/1475-2891-13-51 333. Semeraro M.D., Smith C., Kaiser M., Levinger I., Duque G., Gruber H.-J., Herrmann M.. Physical activity, a modulator of aging through effects on telomere biology. Aging (2020) 12 13803-13823. DOI: 10.18632/aging.103504 334. Perneger T.V.. What’s wrong with Bonferroni adjustments. BMJ (1998) 316 1236-1238. DOI: 10.1136/bmj.316.7139.1236 335. Bland M.. An Introduction to Medical Statistics (2000) 336. Pavanello S., Campisi M., Fabozzo A., Cibin G., Tarzia V., Toscano G., Gerosa G.. The biological age of the heart is consistently younger than chronological age. Sci. Rep. (2020) 10 10752. DOI: 10.1038/s41598-020-67622-1 337. Kioussis B., Tuttle C.S., Heard D.S., Kennedy B.K., Lautenschlager N.T., Maier A.B.. Targeting impaired nutrient sensing with repurposed therapeutics to prevent or treat age-related cognitive decline and dementia: A systematic review. Ageing Res. Rev. (2021) 67 101302. DOI: 10.1016/j.arr.2021.101302 338. Amen D.G., Taylor D.V., Ojala K., Kaur J., Willeumier K.. Effects of brain-directed nutrients on cerebral blood flow and neuropsychological testing: A randomized, double-blind, placebo-controlled, crossover trial. Adv. Mind. Body Med. (2013) 27 24-33 339. Morelli M.B., Gambardella J., Castellanos V., Trimarco V., Santulli G.. Vitamin C and Cardiovascular Disease: An Update. Antioxidants (2020) 9. DOI: 10.3390/antiox9121227 340. Young J.I., Züchner S., Wang G.. Regulation of the Epigenome by Vitamin C. Annu. Rev. Nutr. (2015) 35 545-564. DOI: 10.1146/annurev-nutr-071714-034228 341. Chong T.L., Ahearn E.L., Cimmino L.. Reprogramming the Epigenome With Vitamin C. Front. Cell Dev. Biol. (2019) 7 128. DOI: 10.3389/fcell.2019.00128 342. Kriaucionis S., Heintz N.. The Nuclear DNA Base 5-Hydroxymethylcytosine Is Present in Purkinje Neurons and the Brain. Science (2009) 324 929-930. DOI: 10.1126/science.1169786 343. Szulwach K.E., Li X., Li Y., Song C.X., Wu H., Dai Q., Irier H., Upadhyay A.K., Gearing M., Levey A.I.. 5-hmC-mediated epigenetic dynamics during postnatal neurodevelopment and aging. Nat. Neurosci. (2011) 14 1607-1616. DOI: 10.1038/nn.2959 344. Kaas G.A., Zhong C., Eason D.E., Ross D.L., Vachhani R.V., Ming G.-L., King J.R., Song H., Sweatt J.D.. TET1 Controls CNS 5-Methylcytosine Hydroxylation, Active DNA Demethylation, Gene Transcription, and Memory Formation. Neuron (2013) 79 1086-1093. DOI: 10.1016/j.neuron.2013.08.032 345. Rudenko A., Dawlaty M.M., Seo J., Cheng A.W., Meng J., Le T., Faull K.F., Jaenisch R., Tsai L.-H.. Tet1 Is Critical for Neuronal Activity-Regulated Gene Expression and Memory Extinction. Neuron (2013) 79 1109-1122. DOI: 10.1016/j.neuron.2013.08.003 346. Ming G.-L., Song H.. Adult Neurogenesis in the Mammalian Brain: Significant Answers and Significant Questions. Neuron (2011) 70 687-702. DOI: 10.1016/j.neuron.2011.05.001 347. Erickson K.I., Voss M.W., Prakash R.S., Basak C., Szabo A., Chaddock L., Kim J.S., Heo S., Alves H., White S.M.. Exercise training increases size of hippocampus and improves memory. Proc. Natl. Acad. Sci. USA (2011) 108 3017-3022. DOI: 10.1073/pnas.1015950108 348. Minhas P.S., Latif-Hernandez A., McReynolds M.R., Durairaj A.S., Wang Q., Rubin A., Joshi A.U., He J.Q., Gauba E., Liu L.. Restoring metabolism of myeloid cells reverses cognitive decline in ageing. Nature (2021) 590 122-128. DOI: 10.1038/s41586-020-03160-0 349. Etgen T., Sander D., Bickel H., Sander K., Förstl H.. Vitamin D Deficiency, Cognitive Impairment and Dementia: A Systematic Review and Meta-Analysis. Dement. Geriatr. Cogn. Disord. (2012) 33 297-305. DOI: 10.1159/000339702 350. Haussler M.R., Whitfield G.K., Haussler C.A., Sabir M.S., Khan Z., Sandoval R., Jurutka P.W.. 1,25-dihydroxyvitamin d and klotho: A tale of two renal hormones coming of age. Vitam. Horm. (2016) 100 165-230. PMID: 26827953 351. Kuro-O M., Matsumura Y., Aizawa H., Kawaguchi H., Suga T., Utsugi T., Ohyama Y., Kurabayashi M., Kaname T., Kume E.. Mutation of the mouse klotho gene leads to a syndrome resembling ageing. Nature (1997) 390 45-51. DOI: 10.1038/36285 352. Hsu S.C., Huang S.M., Chen A., Sun C.Y., Lin S.H., Chen J.S., Liu S.T., Hsu Y.J.. Resveratrol increases anti-aging Klotho gene expression via the activating transcription factor 3/c-Jun complex-mediated signaling pathway. Int. J. Biochem. Cell Biol. (2014) 53 361-371. DOI: 10.1016/j.biocel.2014.06.002 353. Papadopoli D., Boulay K., Kazak L., Pollak M., Mallette F.A., Topisirovic I., Hulea L.. mTOR as a central regulator of lifespan and aging. F1000Research (2019) 8 998. DOI: 10.12688/f1000research.17196.1 354. Kaeberlein M.. Rapamycin and Ageing: When, for How Long, and How Much?. J. Genet. Genom. (2014) 41 459-463. DOI: 10.1016/j.jgg.2014.06.009 355. Weichhart T.. mTOR as Regulator of Lifespan, Aging, and Cellular Senescence: A Mini-Review. Gerontology (2018) 64 127-134. DOI: 10.1159/000484629 356. Caccamo A., Majumder S., Richardson A., Strong R., Oddo S.. Molecular Interplay between Mammalian Target of Rapamycin (mTOR), Amyloid-β, and Tau: Effects on Cognitive Impairments. J. Biol. Chem. (2010) 285 13107-13120. DOI: 10.1074/jbc.M110.100420 357. Li X., Alafuzoff I., Soininen H., Winblad B., Pei J.J.. Levels of mTOR and its downstream targets 4E-BP1, eEF2, and eEF2 kinase in relationships with tau in Alzheimer’s disease brain. FEBS J. (2005) 272 4211-4220. DOI: 10.1111/j.1742-4658.2005.04833.x 358. Tramutola A., Triplett J.C., Di Domenico F., Niedowicz D.M., Murphy M.P., Coccia R., Perluigi M., Butterfield D.A.. Alteration of mTOR signaling occurs early in the progression of Alzheimer disease (AD): Analysis of brain from subjects with pre-clinical AD, amnestic mild cognitive impairment and late-stage AD. J. Neurochem. (2015) 133 739-749. DOI: 10.1111/jnc.13037 359. Lisse T.S., Hewison M.. Vitamin D: A new player in the world of mTOR signalling. Cell Cycle (2011) 10 1888-1889. DOI: 10.4161/cc.10.12.15620 360. Gupta S.C., Patchva S., Aggarwal B.B.. Therapeutic Roles of Curcumin: Lessons Learned from Clinical Trials. AAPS J. (2013) 15 195-218. DOI: 10.1208/s12248-012-9432-8 361. Van Aller G.S., Carson J.D., Tang W., Peng H., Zhao L., Copeland R.A., Tummino P.J., Luo L.. Epigallocatechin gallate (EGCG), a major component of green tea, is a dual phosphoinositide-3-kinase/mTOR inhibitor. Biochem. Biophys. Res. Commun. (2011) 406 194-199. DOI: 10.1016/j.bbrc.2011.02.010 362. Chen Z., Zhang Y., Jia C., Wang Y., Lai P., Zhou X., Wang Y., Song Q., Lin J., Ren Z.. mTORC1/2 targeted by n-3 polyunsaturated fatty acids in the prevention of mammary tumorigenesis and tumor progression. Oncogene (2014) 33 4548-4557. DOI: 10.1038/onc.2013.402 363. Wang Y., Li X., Guo Y., Chan L., Guan X.. α-Lipoic acid increases energy expenditure by enhancing adenosine monophosphate–activated protein kinase–peroxisome proliferator-activated receptor-γ coactivator-1α signaling in the skeletal muscle of aged mice. Metabolism (2010) 59 967-976. DOI: 10.1016/j.metabol.2009.10.018 364. Salminen A., Kaarniranta K.. AMP-activated protein kinase (AMPK) controls the aging process via an integrated signaling network. Ageing Res. Rev. (2012) 11 230-241. DOI: 10.1016/j.arr.2011.12.005 365. Reznick R.M., Zong H., Li J., Morino K., Moore I.K., Yu H.J., Liu Z.-X., Dong J., Mustard K.J., Hawley S.A.. Aging-Associated Reductions in AMP-Activated Protein Kinase Activity and Mitochondrial Biogenesis. Cell Metab. (2007) 5 151-156. DOI: 10.1016/j.cmet.2007.01.008 366. Ljubicic V., Hood D.A.. Diminished contraction-induced intracellular signaling towards mitochondrial biogenesis in aged skeletal muscle. Aging Cell (2009) 8 394-404. DOI: 10.1111/j.1474-9726.2009.00483.x 367. Poels J., Spasić M.R., Callaerts P., Norga K.K.. Expanding roles for AMP-activated protein kinase in neuronal survival and autophagy. Bioessays (2009) 31 944-952. DOI: 10.1002/bies.200900003 368. Yang T.Y.T., Wang L.W.L., Zhu M.Z.M., Yan L.Y.L.. Properties and molecular mechanisms of resveratrol: A review. Die Pharm. -Int. J. Pharm. Sci. (2015) 70 501-506. DOI: 10.1691/PH.2015.5571 369. Haigis M.C., Sinclair D.A.. Mammalian Sirtuins: Biological Insights and Disease Relevance. Annu. Rev. Pathol. Mech. Dis. (2010) 5 253-295. DOI: 10.1146/annurev.pathol.4.110807.092250 370. Cantó C., Gerhart-Hines Z., Feige J.N., Lagouge M., Noriega L., Milne J.C., Elliott P.J., Puigserver P., Auwerx J.. AMPK regulates energy expenditure by modulating NAD+ metabolism and SIRT1 activity. Nature (2009) 458 1056-1060. DOI: 10.1038/nature07813 371. Lan F., Cacicedo J.M., Ruderman N., Ido Y.. SIRT1 modulation of the acetylation status, cytosolic localization, and activity of LKB1. Possible role in AMP-activated protein kinase activation. J. Biol. Chem. (2008) 283 27628-27635. DOI: 10.1074/jbc.M805711200 372. Sinha K., Chaudhary G., Gupta Y.K.. Protective effect of resveratrol against oxidative stress in middle cerebral artery occlusion model of stroke in rats. Life Sci. (2002) 71 655-665. DOI: 10.1016/S0024-3205(02)01691-0 373. Wang Q., Xu J., Rottinghaus G.E., Simonyi A., Lubahn D., Sun G.Y., Sun A.Y.. Resveratrol protects against global cerebral ischemic injury in gerbils. Brain Res. (2002) 958 439-447. DOI: 10.1016/S0006-8993(02)03543-6 374. Amri A., Chaumeil J.C., Sfar S., Charrueau C.. Administration of resveratrol: What formulation solutions to bioavailability limitations?. J. Control. Release (2012) 158 182-193. DOI: 10.1016/j.jconrel.2011.09.083 375. Sarubbo F., Moranta D., Asensio V.J., Miralles A., Esteban S.. Effects of Resveratrol and Other Polyphenols on the Most Common Brain Age-Related Diseases. Curr. Med. Chem. (2017) 24 4245-4266. DOI: 10.2174/0929867324666170724102743 376. Cosín-Tomàs M., Senserrich J., Arumí-Planas M., Alquézar C., Pallàs M., Martín-Requero Á., Suñol C., Kaliman P., Sanfeliu C.. Role of Resveratrol and Selenium on Oxidative Stress and Expression of Antioxidant and Anti-Aging Genes in Immortalized Lymphocytes from Alzheimer’s Disease Patients. Nutrients (2019) 11. DOI: 10.3390/nu11081764 377. Chang H.-C., Guarente L.. SIRT1 Mediates Central Circadian Control in the SCN by a Mechanism that Decays with Aging. Cell (2013) 153 1448-1460. DOI: 10.1016/j.cell.2013.05.027 378. Nakahata Y., Sahar S., Astarita G., Kaluzova M., Sassone-Corsi P.. Circadian Control of the NAD. Science (2009) 324 654-657. DOI: 10.1126/science.1170803 379. Braidy N., Berg J., Clement J., Khorshidi F., Poljak A., Jayasena T., Grant R., Sachdev P.. Role of nicotinamide sdenine dinucleotide and related precursors as therapeutic targets for age-related degenerative diseases: Rationale, biochemistry, pharmacokinetics, and outcomes. Antioxid. Redox Signal. (2019) 30 251-294. DOI: 10.1089/ars.2017.7269 380. Siegmund K.D., Connor C.M., Campan M., Long T.I., Weisenberger D.J., Biniszkiewicz D., Jaenisch R., Laird P.W., Akbarian S.. DNA Methylation in the Human Cerebral Cortex Is Dynamically Regulated throughout the Life Span and Involves Differentiated Neurons. PLoS ONE (2007) 2. DOI: 10.1371/journal.pone.0000895 381. Sharif R., Thomas P., Zalewski P., Fenech M.. Zinc supplementation influences genomic stability biomarkers, antioxidant activity, and zinc transporter genes in an elderly Australian population with low zinc status. Mol. Nutr. Food Res. (2015) 59 1200-1212. DOI: 10.1002/mnfr.201400784 382. Aaseth J., Alexander J., Bjørklund G., Hestad K., Dusek P., Roos P.M., Alehagen U.. Treatment strategies in Alzheimer’s disease: A review with focus on selenium supplementation. Biometals (2016) 29 827-839. DOI: 10.1007/s10534-016-9959-8 383. Macaron T., Giudici K.V., Bowman G.L., Sinclair A., Stephan E., Vellas B., Barreto P.D.S.. Associations of Omega-3 fatty acids with brain morphology and volume in cognitively healthy older adults: A narrative review. Ageing Res. Rev. (2021) 67 101300. DOI: 10.1016/j.arr.2021.101300 384. Lin P.-Y., Chiu C.-C., Huang S.-Y., Su K.-P.. A Meta-Analytic Review of Polyunsaturated Fatty Acid Compositions in Dementia. J. Clin. Psychiatry (2012) 73 1245-1254. DOI: 10.4088/JCP.11r07546 385. Bo Y., Zhang X., Wang Y., You J., Cui H., Zhu Y., Pang W., Liu W., Jiang Y., Lu Q.. The n-3 Polyunsaturated Fatty Acids Supplementation Improved the Cognitive Function in the Chinese Elderly with Mild Cognitive Impairment: A Double-Blind Randomized Controlled Trial. Nutrients (2017) 9. DOI: 10.3390/nu9010054 386. Avallone R., Vitale G., Bertolotti M.. Omega-3 Fatty Acids and Neurodegenerative Diseases: New Evidence in Clinical Trials. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20174256 387. White III C.W., Fan X., Maynard J.C., Wheatley E.G., Bieri G., Couthouis J., Burlingame A.L., Villeda S.A.. Age-related loss of neural stem cell O-GlcNAc promotes a glial fate switch through STAT3 activation. Proc. Natl. Acad. Sci. USA (2020) 117 22214-22224. DOI: 10.1073/pnas.2007439117 388. Moreno-Jiménez E.P., Flor-García M., Terreros-Roncal J., Rábano A., Cafini F., Pallas-Bazarra N., Ávila J., Llorens-Martín M.. Adult hippocampal neurogenesis is abundant in neurologically healthy subjects and drops sharply in patients with Alzheimer’s disease. Nat. Med. (2019) 25 554-560. DOI: 10.1038/s41591-019-0375-9 389. Moradi H.R., Hajali V., Khaksar Z., Vafaee F., Forouzanfar F., Negah S.S.. The next step of neurogenesis in the context of Alzheimer’s disease. Mol. Biol. Rep. (2021) 48 5647-5660. DOI: 10.1007/s11033-021-06520-9 390. Akan I., Halim A., Vakhrushev S., Clausen H., Hanover J.. Drosophila. Cells (2021) 10. DOI: 10.3390/cells10051026 391. Colcombe S.J., Kramer A.F.. Fitness effects on the cognitive function of older adults: A meta-analytic study. Psychol. Sci. (2003) 14 125-130. DOI: 10.1111/1467-9280.t01-1-01430 392. Colcombe S.J., Erickson K.I., Raz N., Webb A.G., Cohen N.J., McAuley E., Kramer A.F.. Aerobic Fitness Reduces Brain Tissue Loss in Aging Humans. J. Gerontol. Ser. A (2003) 58 M176-M180. DOI: 10.1093/gerona/58.2.M176 393. Piepmeier A.T., Etnier J.L.. Brain-derived neurotrophic factor (BDNF) as a potential mechanism of the effects of acute exercise on cognitive performance. J. Sport Health Sci. (2015) 4 14-23. DOI: 10.1016/j.jshs.2014.11.001 394. Gupta R., Khan R., Cortes C.J.. Forgot to Exercise? Exercise Derived Circulating Myokines in Alzheimer’s Disease: A Perspective. Front. Neurol. (2021) 12 649452. DOI: 10.3389/fneur.2021.649452 395. Willcox B.J., Willcox D.C.. Caloric restriction, caloric restriction mimetics, and healthy aging in Okinawa: Controversies and clinical implications. Curr. Opin. Clin. Nutr. Metab. Care. (2014) 17 51-58. DOI: 10.1097/MCO.0000000000000019 396. Weindruch R., Walford R.L., Charles C.. The Retardation of Aging and Disease by Dietary Restriction (1988) 397. Anderson R.M., Le Couteur D.G., De Cabo R.. Caloric Restriction Research: New Perspectives on the Biology of Aging. J. Gerontol. Ser. A (2017) 73 1-3. DOI: 10.1093/gerona/glx212 398. Hwangbo D.-S., Lee H.-Y., Abozaid L.S., Min K.-J.. Mechanisms of Lifespan Regulation by Calorie Restriction and Intermittent Fasting in Model Organisms. Nutrients (2020) 12. DOI: 10.3390/nu12041194 399. Stadelmann C., Timmler S., Barrantes-Freer A., Simons M.. Myelin in the Central Nervous System: Structure, Function, and Pathology. Physiol. Rev. (2019) 99 1381-1431. DOI: 10.1152/physrev.00031.2018 400. Neumann B., Baror R., Zhao C., Segel M., Dietmann S., Rawji K.S., Foerster S., McClain C.R., Chalut K., van Wijngaarden P.. Metformin restores CNS remyelination capacity by rejuvenating aged stem cells. Cell Stem. Cell (2019) 25 473-485.e8. DOI: 10.1016/j.stem.2019.08.015 401. Lindholm P., Saarma M.. Novel CDNF/MANF family of neurotrophic factors. Dev. Neurobiol. (2010) 70 360-371. DOI: 10.1002/dneu.20760 402. Lindholm P., Peränen J., Andressoo J.-O., Kalkkinen N., Kokaia Z., Lindvall O., Timmusk T., Saarma M.. MANF is widely expressed in mammalian tissues and differently regulated after ischemic and epileptic insults in rodent brain. Mol. Cell. Neurosci. (2008) 39 356-371. DOI: 10.1016/j.mcn.2008.07.016 403. Yang S., Yang H., Chang R., Yin P., Yang Y., Yang W., Huang S., Gaertig M.A., Li S., Li X.-J.. MANF regulates hypothalamic control of food intake and body weight. Nat. Commun. (2017) 8 579. DOI: 10.1038/s41467-017-00750-x 404. Sousa-Victor P., Neves J., Cedron-Craft W., Ventura P.B., Liao C.-Y., Riley R.R., Soifer I., van Bruggen N., Kolumam G.A., Villeda S.A.. MANF regulates metabolic and immune homeostasis in ageing and protects against liver damage. Nat. Metab. (2019) 1 276-290. DOI: 10.1038/s42255-018-0023-6 405. Poehlman E.T., Turturro A., Bodkin N., Cefalu W., Heymsfield S., Holloszy J., Kemnitz J.. Caloric Restriction Mimetics: Physical Activity and Body Composition Changes. J. Gerontol. Ser. A (2001) 56 45-54. DOI: 10.1093/gerona/56.suppl_1.45 406. Carey A.L., Kingwell B.A.. Novel pharmacological approaches to combat obesity and insulin resistance: Targeting skeletal muscle with ‘exercise mimetics’. Diabetologia (2009) 52 2015-2026. DOI: 10.1007/s00125-009-1420-x 407. Goldman D.P., Cutler D., Rowe J.W., Michaud P.-C., Sullivan J., Peneva D., Olshansky S.J.. Substantial Health And Economic Returns From Delayed Aging May Warrant A New Focus For Medical Research. Health Aff. (2013) 32 1698-1705. DOI: 10.1377/hlthaff.2013.0052 408. Goldman D.. The economic returns to delayed aging: Promise and pitfalls. Innov. Aging (2017) 1 1082. DOI: 10.1093/geroni/igx004.3969 409. Li X., Xu S., Yu M., Wang K., Tao Y., Zhou Y., Shi J., Zhou M., Wu B., Yang Z.. Risk factors for severity and mortality in adult COVID-19 inpatients in Wuhan. J. Allergy Clin. Immunol. (2020) 146 110-118. DOI: 10.1016/j.jaci.2020.04.006 410. Bischof E., Siow R.C., Zhavoronkov A., Kaeberlein M.. The potential of rapalogs to enhance resilience against SARS-CoV-2 infection and reduce the severity of COVID-19. Lancet (2021) 2 E105-E111. DOI: 10.1016/S2666-7568(20)30068-4 411. Galkin F., Parish A., Bischof E., Zhang J., Mamoshina P., Zhavoronkov A.. Increased Pace of Aging in COVID-Related Mortality. Life (2021) 11. DOI: 10.3390/life11080730 412. Olshansky S.J., Goldman D.P., Zheng Y., Rowe J.W.. Aging in America in the Twenty-first Century: Demographic Forecasts from the MacArthur Foundation Research Network on an Aging Society. Milbank Q. (2009) 87 842-862. DOI: 10.1111/j.1468-0009.2009.00581.x 413. Davis J.K.. New Methuselahs: The Ethics of Life Extension (2018) 414. Pyrkov T.V., Avchaciov K., Tarkhov A.E., Menshikov L.I., Gudkov A.V., Fedichev P.O.. Longitudinal analysis of blood markers reveals progressive loss of resilience and predicts human lifespan limit. Nat. Commun. (2021) 12 2765. DOI: 10.1038/s41467-021-23014-1 415. Armstrong N.J., Mather K.A., Thalamuthu A., Wright M.J., Trollor J.N., Ames D., Brodaty H., Schofield P.R., Sachdev P.S., Kwok J.B.. Aging, exceptional longevity and comparisons of the Hannum and Horvath epigenetic clocks. Epigenomics (2017) 9 689-700. DOI: 10.2217/epi-2016-0179 416. Demidenko O., Barardo D., Budovskii V., Finnemore R., Palmer F.R., Kennedy B.K., Budovskaya Y.V.. Rejuvant. Aging (2021) 13 24485-24499. DOI: 10.18632/aging.203736 417. Goldsmith R.E., Heiens R.A.. Subjective age: A test of five hypotheses. Gerontologist (1992) 32 312-317. DOI: 10.1093/geront/32.3.312 418. Westerhof G.. Forever young? A comparison of age identities in the United States and Germany. Res. Aging (2003) 25 366-383. DOI: 10.1177/0164027503025004002 419. Montepare J.M., Lachman M.E.. “You’re only as old as you feel”: Self-perceptions of age, fears of aging, and life satisfaction from adolescence to old age. Psychol. Aging (1989) 4 73-78. DOI: 10.1037/0882-7974.4.1.73 420. Weiss D., Lang F.R.. “They” are old but “I” feel younger: Age-group dissociation as a self-protective strategy in old age. Psychol. Aging (2012) 27 153-163. DOI: 10.1037/a0024887 421. Valenzuela M.J., Sachdev P.. Brain reserve and dementia: A systematic review. Psychol. Med. (2006) 36 441-454. DOI: 10.1017/S0033291705006264 422. Christensen H., Anstey K.J., Leach L.S., Mackinnon A.J., Craik F.I.M., Salthouse T.A.. Intelligence, Education, and the Brain Reserve Hypothesis. The Handbook of Aging and Cognition (2008) 133-188 423. Garibotto V., Borroni B., Sorbi S., Cappa S.F., Padovani A., Perani D.. Education and occupation provide reserve in both ApoE ε4 carrier and noncarrier patients with probable Alzheimer’s disease. Neurol. Sci. (2012) 33 1037-1042. DOI: 10.1007/s10072-011-0889-5 424. Sun F.W., Stepanovic M.R., Andreano J., Barrett L.F., Touroutoglou A., Dickerson B.C.. Youthful Brains in Older Adults: Preserved Neuroanatomy in the Default Mode and Salience Networks Contributes to Youthful Memory in Superaging. J. Neurosci. (2016) 36 9659-9668. DOI: 10.1523/JNEUROSCI.1492-16.2016 425. Balduino E., de Melo B.A.R., Silva L.D.S.M.D., Martinelli J.E., Cecato J.F.. The “SuperAgers” construct in clinical practice: Neuropsychological assessment of illiterate and educated elderly. Int. Psychogeriatr. (2020) 32 191-198. DOI: 10.1017/S1041610219001364 426. Cook A.H., Sridhar J., Ohm D., Rademaker A., Mesulam M.-M., Weintraub S., Rogalski E.. Rates of Cortical Atrophy in Adults 80 Years and Older With Superior vs Average Episodic Memory. JAMA (2017) 317 1373-1375. DOI: 10.1001/jama.2017.0627 427. Rogalski E.J., Gefen T., Shi J., Samimi M., Bigio E., Weintraub S., Geula C., Mesulam M.-M.. Youthful Memory Capacity in Old Brains: Anatomic and Genetic Clues from the Northwestern SuperAging Project. J. Cogn. Neurosci. (2013) 25 29-36. DOI: 10.1162/jocn_a_00300 428. Brody A.L., Mandelkern M.A., Jarvik M.E., Lee G.S., Smith E.C., Huang J.C., London E.D.. Differences between smokers and nonsmokers in regional gray matter volumes and densities. Biol. Psychiat. (2004) 55 77-84. DOI: 10.1016/S0006-3223(03)00610-3 429. Almeida O.P., Garrido G.J., Lautenschlager N.T., Hulse G.K., Jamrozik K., Flicker L.. Smoking Is Associated With Reduced Cortical Regional Gray Matter Density in Brain Regions Associated With Incipient Alzheimer Disease. Am. J. Geriatr. Psychiatry (2008) 16 92-98. DOI: 10.1097/JGP.0b013e318157cad2 430. Pfefferbaum A., Lim K.O., Zipursky R.B., Mathalon D.H., Rosenbloom M.J., Lane B., Ha C.N., Sullivan E.V.. Brain Gray and White Matter Volume Loss Accelerates with Aging in Chronic Alcoholics: A Quantitative MRI Study. Alcohol. Clin. Exp. Res. (1992) 16 1078-1089. DOI: 10.1111/j.1530-0277.1992.tb00702.x 431. Ning K., Zhao L., Matloff W., Sun F., Toga A.W.. Association of relative brain age with tobacco smoking, alcohol consumption, and genetic variants. Sci. Rep. (2020) 10 10. DOI: 10.1038/s41598-019-56089-4 432. Gu Y., Scarmeas N., Short E.E., Luchsinger J.A., DeCarli C., Stern Y., Manly J.J., Schupf N., Mayeux R., Brickman A.M.. Alcohol intake and brain structure in a multiethnic elderly cohort. Clin. Nutr. (2014) 33 662-667. DOI: 10.1016/j.clnu.2013.08.004 433. Reale M., Costantini E., Jagarlapoodi S., Khan H., Belwal T., Cichelli A.. Relationship of Wine Consumption with Alzheimer’s Disease. Nutrients (2020) 12. DOI: 10.3390/nu12010206 434. Dobrohotova T.A., Bragina N.N.. The Left-Handers (1994) 435. Marks J.S., Williamson D.F.. Left-handedness and life expectancy. N. Engl. J. Med. (1991) 325 1041-1043. PMID: 1886627 436. Harris L.J.. Do left-handers die sooner than right-handers? Commentary on Coren and Halpern’s (1991) “Left-handedness: A marker for decreased survival fitness”. Psychol. Bull. (1993) 114 203-234. DOI: 10.1037/0033-2909.114.2.203 437. de Leon M.J., la Regina M.E., Ferris S.H., Gentes C.I., Miller J.D.. Reduced incidence of left-handedness in clinically diagnosed dementia of the Alzheimer type. Neurobiol. Aging (1986) 7 161-164. DOI: 10.1016/0197-4580(86)90037-0 438. Doody R.S., Vacca J.L., Massman P.J., Liao T.-Y.. The Influence of Handedness on the Clinical Presentation and Neuropsychology of Alzheimer Disease. Arch. Neurol. (1999) 56 1133-1137. DOI: 10.1001/archneur.56.9.1133 439. Ruigrok A.N., Salimi-Khorshidi G., Lai M.-C., Baron-Cohen S., Lombardo M.V., Tait R.J., Suckling J.. A meta-analysis of sex differences in human brain structure. Neurosci. Biobehav. Rev. (2014) 39 34-50. DOI: 10.1016/j.neubiorev.2013.12.004 440. Kiesow H., Dunbar R.I.M., Kable J.W., Kalenscher T., Vogeley K., Schilbach L., Marquand A.F., Wiecki T.V., Bzdok D.. 10,000 social brains: Sex differentiation in human brain anatomy. Sci. Adv. (2020) 6 eaaz1170. DOI: 10.1126/sciadv.aaz1170 441. Jiang R., Calhoun V.D., Fan L., Zuo N., Jung R., Qi S., Lin D., Li J., Zhuo C., Song M.. Gender Differences in Connectome-based Predictions of Individualized Intelligence Quotient and Sub-domain Scores. Cereb. Cortex (2020) 30 888-900. DOI: 10.1093/cercor/bhz134 442. Coffey C., Lucke J., Saxton J., Ratcliff G., Unitas L., Billig B., Bryan R.N.. Sex differences in brain aging: A quantitative magnetic resonance imaging study. Arch Neurol. (1998) 55 169-179. DOI: 10.1001/archneur.55.2.169 443. Xu J., Kobayashi S., Yamaguchi S., Iijima K.-I., Okada K., Yamashita K.. Gender Effects on Age-Related Changes in Brain Structure. Am. J. Neuroradiol. (2000) 21 112-118. PMID: 10669234 444. Ngun T.C., Ghahramani N., Sánchez F.J., Bocklandt S., Vilain E.. The genetics of sex differences in brain and behavior. Front. Neuroendocr. (2011) 32 227-246. DOI: 10.1016/j.yfrne.2010.10.001 445. Trabzuni D., Ramasamy A., Imran S., Walker R., Smith C., Weale M.E., Hardy J., Ryten M.. North American Brain Expression Consortium Widespread sex differences in gene expression and splicing in the adult human brain. Nat. Commun. (2013) 4 2771. DOI: 10.1038/ncomms3771 446. Flanagan E., Lamport D., Brennan L., Burnet P., Calabrese V., Cunnane S.C., de Wilde M.C., Dye L., Farrimond J.A., Lombardo N.E.. Nutrition and the ageing brain: Moving towards clinical applications. Ageing Res. Rev. (2020) 62 101079. DOI: 10.1016/j.arr.2020.101079
--- title: Lyoprotectant Formulation and Optimization of the J-Aggregates Astaxanthin/BSA/Chitosan Nanosuspension authors: - Yingyuan Zhao - Zhaoxuan Wang - Shuxian Liu - Shiying Xie - Yinchun Xie - Ruifang Li - Hiroaki Oda journal: Biomolecules year: 2023 pmcid: PMC10046545 doi: 10.3390/biom13030496 license: CC BY 4.0 --- # Lyoprotectant Formulation and Optimization of the J-Aggregates Astaxanthin/BSA/Chitosan Nanosuspension ## Abstract Astaxanthin is a carotenoid with excellent antioxidant activity. However, this small lipid-soluble molecule is insoluble in water and has low stability. Although this situation can be improved when astaxanthin is prepared as a nanosuspension, the aqueous form is still not as convenient and safe as the dry powder form for storage, transport, and use. The lyophilization process provides better protection for thermosensitive materials, but this leads to collapse and agglomeration between nanoparticles. To improve this situation, appropriate lyophilization protectants are needed to offer support between the nanoparticles, such as sugars, amino acids, and hydroxy alcohols. The purpose of this work is to screen lyophilization protectants by single-factor experiments and response surface optimization experiments and then explore the optimal ratio of compound lyophilization protectants, and finally, make excellent astaxanthin/BSA/chitosan nanosuspension (ABC-NPs) lyophilized powder. The work shows that the optimal ratio of the compounding lyophilization protectant is $0.46\%$ oligomeric mannose, $0.44\%$ maltose, and $0.05\%$ sorbitol (w/v). The ABC-NPs lyophilized powder prepared under the above conditions had a re-soluble particle size of 472 nm, with a ratio of 1.32 to the particle size of the sample before lyophilization. The lyophilized powder was all in the form of a pink layer. The sample was fluffy and dissolved entirely within 10 s by shaking with water. Consequently, it is expected to solve the problem of inconvenient storage and transportation of aqueous drugs and to expand the application of nanomedicine powders and tablets. ## 1. Introduction Astaxanthin is a carotenoid with the chemical formula C40H52O4, which has a β-viologen ring at each end and can scavenge free radicals to act as an antioxidant [1,2]. Astaxanthin is mainly derived from Haematococcus pluvialis [3,4]. It has many functions, such as anti-inflammatory, vision protection, anti-aging, immunity enhancement, anti-tumor, and maintenance of the central nervous system [5,6,7]. Lu et al. [ 8] reported that hydrophobic astaxanthin monomer molecules can aggregate in hydrated solvents, resulting in two significantly different aggregates. The color, structure, optical properties, and physiological activity of the various aggregates can be dramatically different [9]. There are two forms of aggregation, the card-pack structure for H-type aggregates, and the head–tail structure for J-type aggregates. In this case, the card-pack system is composed of astaxanthin monomers arranged face-to-face in parallel. The head-to-tail system comprises of astaxanthin monomers arranged in parallel and staggered manner [9,10]. However, astaxanthin is a lipid-soluble active small molecule. Its water solubility and stability are poor [11]. In this study, we selected whey protein and chitosan as nanocarrier materials. We then used molecular self-assembly and spin evaporation techniques to encapsulate astaxanthin in a core-shell structure formed by whey protein and chitosan, resulting in the preparation of astaxanthin/bovine serum albumin/chitosan nanoparticles (ABC-NPs). This method increased the stability of astaxanthin and improved its bioavailability of astaxanthin. Chitosan [12] is a product of natural chitin deacetylation linked by β-(1→4) glycosidic bonds. It is the only polysaccharide in nature that contains a free amino group. In addition, chitosan is safe, non-toxic, degradable, and biocompatible. Whey protein [13] is a valuable protein extracted from milk. It contains α-lactalbumin (α-La), β-lactoglobulin (β-Lg), and bovine serum albumin (BSA), which can bind to a variety of cations, anions, and other small molecules, and thus has a variety of activities. In addition, it has high nutritional value and is easily digested and absorbed, so it is a suitable carrier for drugs. In a specific ratio of water and organic solution, astaxanthin small molecules spontaneously self-aggregate to form aggregates. They are present in the hydrophobic micro-regions of nanocarriers constructed from chitosan and whey protein. As aqueous solutions, nanosuspensions are not as convenient and safe to store, transport, and use as dry powders. Freeze drying allows better preservation of heat-sensitive materials, such as the material’s color, structure, and chemical properties. However, freeze-drying is the process of removing inter-particle moisture, which leads to the collapse and aggregation of the nanocomposites. In addition, the proteins on the surface of the nanoparticles may also undergo structural changes due to mechanical forces [14]. Appropriate lyophilization protectants can provide support in ABC-NPs, such as sugars, hydroxy alcohols, amino acids, and monosaccharide mixtures [15,16]. Among them, sugars are classified as monosaccharides, disaccharides, and glycans. Monosaccharides include glucose; disaccharides include maltose and lactose; oligosaccharides include oligomannose and oligomaltose; alcohols include mannitol, oligomannitol, and sorbitol; amino acids include glycine, lysine, and glutamic acid [17]; mixed lyophilization protectants are obtained from the compounding of sugars and alcohols or sugars and amino acids [18]. In general, there are two theories to explain the mechanism of action of lyophilization protectants, namely the water substitution hypothesis and the glassy state hypothesis. Andreani et al. [ 19] explained the water substitution hypothesis as the interaction between sugar molecules and nanoparticles. They suggested that the abundant hydroxyl groups in sugar molecules can form hydrogen bonds with sugars and proteins on the nanoparticle surface. In the case of dehydration, sugar can replace water molecules to interact with the nanoparticle surface and inhibit the dehydration-induced shortening of the distance between adjacent nanoparticles, thus stabilizing the nanosuspension system. The glassy state is the state of existence of a substance under certain conditions. In the glassy state, the substance exists in a non-crystalline form, the viscosity of the solution increases, the diffusion coefficient of the solute decreases, and crystallization is not easily formed. In the drying process, with the reduction of water, the concentration of the solution will increase significantly [20]. If the highly concentrated protectant solution does not crystallize, the mixture of protectant and water will be glassy and form a glassy state. This highly viscous glassy state surrounds the protein, creating a glassy carbohydrate. The movement of non-covalently bonded chains of proteins and biomolecules in glassy solutions is prevented, thus preserving the original conformation and physicochemical properties of the proteins and protecting the architecture of nanoparticles [21]. In this study, we selected six effective lyophilization protectants: oligosaccharides, chitosan, maltose, mannitol, sorbitol, and L-lysine. Sucrose has been used as a lyophilized protective agent for many years, but in recent years it has been found that sucrose metabolizes to fructose and produces potential cardiovascular toxicity, so this option is ruled out in this study [22]. The ABC-NPs complex was used as a model for the nanocarriers, and the water molecules in the complex were purged using a freeze-drying technique. Various characterizations of the lyophilized powders were used as evaluation criteria, such as particle size, polydispersity index (PDI), and re-solubilization time [23]. Then, three of the six lyophilization protectants were selected and compounded. Response surface analysis method was used to explore the optimal ratio of lyophilized protective agents. The above study is expected to solve the problem of inconvenient storage and transportation of ABC-NPs nanosuspension and expand the application of nanomedicine powder and tablet. ## 2.1. Materials Chitosan (degree of deacetylation $90.25\%$) and chitooligosaccharides (degree of deacetylation $91.10\%$) were purchased from Zhejiang Aoxing Biotechnology Co., Ltd. (Zhejiang, China). Bovine serum protein (purity $97\%$) was obtained from Beijing Solarbio Technology Co. Astaxanthin (purity $99\%$) was purchased from Aladdin Biochemical Technology (Shanghai, China). Sorbitol, mannose, L/D lysine, and lactose were purchased from Solarbio, Beijing, China (Beijing, China). Oligomannose and oligomaltose were purchased from Jiangsu Duoyang Bioengineering Technology Co. Ltd. (Jiangsu, China). Sodium hydroxide (purity $99.7\%$), hydrochloric acid (purity $99.7\%$), and anhydrous ethanol (purity $99.7\%$) were purchased from Sinopharm Chemical Reagents Co. Ltd. (Beijing, China). ## 2.2. Fabrication of J Aggregates—Astaxanthin/Bovine Serum Albumin/Chitosan Nanoparticles Dissolve 1000 mg of chitosan in 50 mL of sterilized ultrapure water, then add 3 mL of hydrochloric acid solution with pH 1 (concentration 0.1 mol/mL) and stir magnetically for one hour at room temperature. After the chitosan powder was fully hydrated, the hydrochloric acid solution with a concentration of 0.05 mol/mL was added dropwise and stirred until the chitosan solution was clear and transparent. Subsequently, sodium hydroxide solution with pH 13 was added to the chitosan solution to adjust its pH to 5–6. Finally, sterilized ultrapure water was added to the chitosan solution and fixed the volume at 200 mL. The chitosan stock solution with a concentration of 5 mg/mL was finally obtained. Dissolve 3 mg of astaxanthin in 100 mL of anhydrous ethanol and magnetic stirring for 25 min at 25 °C and protect it from light to obtain 0.03 mg/mL of astaxanthin ethanol solution. Accurately weigh 10 mg of bovine serum albumin and dissolve it in 100 mL of sterile ultrapure water. After stirring magnetically for 20–30 min at 25 °C, a solution of bovine serum albumin at a concentration of 0.1 mg/mL was obtained. After preparing the above three solutions, store them in a refrigerator at 4 °C. The ABC-NPs were prepared following the procedure described by our research team with some optimization [24]. Under light-proof conditions, we slowly poured 50 mL of astaxanthin solution with a concentration of 0.03 mg/mL into 50 mL of chitosan solution with a concentration of 0.1 mg/mL at an addition rate of 43 mL/min for about 70 s, followed by magnetic stirring for 20 min. Then 50 mL of bovine serum albumin solution with a concentration of 0.1 mg/mL was slowly added to the above-mixed solution with a spiking flow rate of 30 mL/min for 100 s, followed by magnetic stirring for 20 min. The rotary evaporator was turned on, the water bath was set at 35 °C, and the solution was rotated at 15 rpm for 70 min. When the mixture was evaporated to approximately 60 mL, we stopped the rotary evaporator, from which J Aggregates—astaxanthin/bovine serum albumin/chitosan nanoparticles, referred to as ABC-NPs, were prepared and stored in a refrigerator at 4 °C. ## 2.3. Determination of Particle Size and Zeta—Potential of ABC-NPs The Malvern Nano ZS90 dynamic light scattering (DLS) instrument was chosen to measure the mean particle size, PDI, and potential of the ABC-NPs. Freshly prepared 1 mL ABC-NPs was added to the cuvette and placed in the sample chamber with an equilibration time of 120 s. The dispersant medium was water with a dispersant RI of 1.330 and a material RI of 1.45. Each sample was measured for three times and analyzed the average value [24]. ## 2.4. Freeze-Drying Parameters of ABC-NPs The bulk-prepared ABC-NPs were divided into seven vials of 2 cm diameter, of which all seven vials were mixed with 2.5 mL of ABC-NPs and 0.5 mL of different concentrations of lyophilized protectant so that the final concentrations of the oligomaltose solution, chitosan solution, mannitol solution, and sorbitol solution were $0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, $0.25\%$, and $0.5\%$ (w/v) so that the final concentrations of maltose solution were $0.05\%$, $0.25\%$, $0.5\%$, $1\%$, and $2\%$, and so that the final concentrations of L-lysine solution were $0.0001\%$, $0.0002\%$, $0.001\%$, $0.005\%$, and $0.025\%$. The other two vials were 2.5 mL of ABC-NPs and 0.5 mL of sterile water mixture, both as blank control samples. The height of the sample in the vial is about 1 cm. After dispensing, they were sealed with a sealing film and rubber band tied with air permeability holes and put into a −80 °C refrigerator for 12 h of pre-freeze drying and then put into a freeze dryer (World Bank Technology Co., Ltd.) for 24 h of main freeze drying with a pressure of 0.03 MPa. ## 2.5. Observation of the Appearance of ABC-NPs Lyophilized Powder Freshly prepared lyophilized powder of ABC-NPs was removed from the freeze dryer. Then, the samples were immediately photographed within 20 min. Subsequently, the situation of lyophilized powder and lyophilized powder wers observed after re-dissolution and the Tyndall effect. ## 2.6. Resolubilisation of ABC-NPs Lyophilized Powder The prepared ABC-NPs lyophilized powder was re-dissolved with sterilized ultrapure water within 30 min. Then, researcher shook the lyophilized powder by hand and recorded the time to completely dissolve in the solid state. The particle size and the polydispersity index PDI (see Section 2.3 the lyophilized ABC-NPs lyophilized powder were measured by Malvern Nano ZS90 within one hour after the re-dissolution and recorded as re-dissolution particle size Sa. Finally, we evaluated the amount of protective agent added and the corresponding effect. ## 2.7. Single-Factor Experimental Screening According to the references [25] and the results of the preliminary experiments, the following six lyophilization protectants, which are L-lysine [26], maltose [27], oligomannose [28], chitosan [29], mannitol [30], and sorbitol [15] were selected. The final concentrations of oligomannose, chitosan, and mannitol lyophilized protectants were $0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, and $0.5\%$ (w/v). The final concentrations of L-lysine were $0.0001\%$, $0.0002\%$, $0.0010\%$, $0.0050\%$, and $0.0250\%$ (w/v). The concentration of lyophilization protectant can be defined as $0.5\%$ (w/v) at 0.5 g/100 mL. Then the corresponding concentration of lyophilization protectant solution was prepared and added to ABC-NPs by the addition method. The ABC-NPs were prepared into lyophilized powder by freeze-drying procedure, and then the status and re-solubilization time of the lyophilized powder was observed and recorded. Finally, the particle size, potential, PDI, and particle size change ratio Sa/Sb (short for Size after/Size before freeze drying) of the lyophilized powder after re-solubilization were measured. ## 2.8. Single Lyophilisation Protectant Optimisation In the above steps, six types of lyophilized protectants were screened firstly. Next, different lyophilized protectants were optimized one by one through controlling the material concentration and preparation conditions. Then the most suitable three lyophilized protectants were selected by using the state of the lyophilized powder and the particle size of the ABC-NPs complex after re-solubilization as evaluation criteria. Finally, the concentration continued to be refined. ## 2.9. Response Surface Optimization Experiments Based on the single-factor experiments, an excellent compounded lyophilized protectant was prepaerd. The Box–Behnken experimental design with Design-Expert V8.0.6 software was used to establish a mathematical regression model to analyze the data using the particle size of ABC-NPs lyophilized powder as the evaluation index. A Box–Behnken design with three independent variables is shown in Table 1. Three independent variables were coded at three levels (−1, 0, 1) on the concentration of lyoprotectant, which produced an experimental design with 17-run experiments using a Design-Expert 8.0.6 [31]. All the experiments were conducted at random with the intent of minimizing the unexplained variability caused by systematic errors. A second-order polynomial equation was formed to study the effects of variables on the concentration of lyoprotectant. The equation indicates the effect of variables inline accordance with linear, quadratic, and cross-product terms: Y=β0+∑$i = 13$β0Xi+∑$i = 13$βiiXi2+∑$i = 13$∑j=i+12βijXiXj where Y is the concentration of lyoprotectant (%), Xi and Xj are the levels of variables, β0 is the constant term, βi is the coefficient of the linear terms, βii is the coefficient of the quadratic terms, and βij is the coefficient of the cross-product terms. Three-dimensional surface plots and contour plots were analyzed by the fitted polynomial equation. F-value and p-value were used to check the significance of the regression coefficient, while R2 and adjusted R2 were used to assess the model adequacy [31]. ## 2.10. Analysis of Data Experimental data were plotted for statistical analysis using Origin 9.0 and Excel, with different letters indicating significant differences ($p \leq 0.05$). Linear regression and ANOVA were performed on the data obtained from response surface tests using Design-Expert 8.0.6 software (the software is from Stat-Ease, Minneapolis, MN, USA). ## 2.11. Freeze-Drying Flow The entire experimental process begins with the preparation of ABC-NPs in large quantities and then uses it as a backup storage solution (as shown in Figure 1). Follow-up experiments were conducted by taking some samples from a large batch of ABC-NPs. After ABC-NPs were removed, lyophilized protectants were added and then freeze-dried at −80 °C for 2 h and −47 °C for 12 h. Finally, the freeze-dried powder was re-dissolved, and the particle size, PDI, potential, and other data were measured. ## 3.1.1. Preparation of ABC-NPs The bulk astaxanthin/bovine serum albumin/chitosan nanoparticles (ABC-NPs) prepared according to the method in 2.4 are shown in Figure 2. The freshly prepared ABC-NPs with J-type astaxanthin aggregates nanosuspensionwas pink-purple and a clear Tyndal pathway could be observed by illumination with a red laser beam. The above phenomenon indicated that the ABC-NPs was well dispersed and homogeneous [24]. ## 3.1.2. Protection of Nanoparticles via Glycoconjugate Lyophilization Protectors Following the method in 2.5, oligomeric mannose ($0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, and $0.5\%$ (w/v)), chitosan ($0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, and $0.5\%$ (w/v)), and maltose ($0.05\%$, $0.25\%$, $0.5\%$, $1\%$, and $2\%$ (w/v)) were added to the ABC-NPs nanosuspension at different concentrations. After freeze-drying, the prepared ABC-NPs lyophilized powder is shown as O1–E1 in Figure 3, and the O1, A1, B1, C1, D1, and E1 lyophilized powder samples in Figure 3A–C are all in skeletonized blocks, and sample E1 was in the state of adhesion. According to the method in 2.6, the samples were re-dissolved, and it was found that the re-dissolution time was gradually shortened as the concentration of sugar lyophilized protectant increased. All of them could be completely dissolved, and no pink granular precipitation was seen. The ABC-NPs after re-solubilization were irradiated with a laser pointer, as shown in Figure 3(O3–E3), and a clear red pathway appeared in almost all samples, except for the lyophilized powder at 1–$2\%$ chitosan concentration, which could not be completely re-solubilized, indicating good dispersion and homogeneity of the samples. The particle size of the samples showed a gradual decrease with increasing oligomeric mannose concentration, especially at a concentration of 0.5 with the particle size reached 371 nm (Figure 3A,D), and Sa/Sb (short for Size after/Size before freeze drying) reached 1.02. However, all samples showed no clear trend in PDI. The values of PDI were below 0.4 and the potentials were almost always above +40 mV. The apparent situation indicated that the nanosuspension system was relatively stable. From Figure 3B, we observed that with the increase of chitosan concentration, the particle size and PDI both experienced a trend of decreasing and then increasing; the particle size and PDI became smaller and smaller when the chitosan concentration was $0.005\%$, $0.025\%$, and $0.05\%$. The smallest particle size reached 480 nm, the smallest PDI reached 0.27, the smallest Sa/Sb was 1.32, and the potentials are all above +40 mV. The above situation indicated that the nanosuspension system was relatively stable in the range of $0.005\%$ to $0.05\%$. When the chitosan concentration was higher than $0.05\%$, both the particle size and PDI increased substantially and the potential decreased substantially, indicating that the system was unstable in this range. In addition, the particle size became smaller and then larger with increasing maltose concentration, and the PDI becomes smaller and smaller (Figure 3C,F). The smallest particle size was 454 nm at $0.5\%$ maltose concentration, and the smallest Sa/Sb was 1.25. The smallest PDI was 0.30 at $0.005\%$ maltose concentration, and the average potential was around +53 mV. Sugars are the most commonly used freeze-drying protectants. Carbohydrate not only do not crystallize under normal conditions but also produce low eutectic mixtures or glassy substances with water molecules during the freezing process. Such substances prevent the growth of ice crystals and allow them to exist in amorphous or microcrystalline form, thus reducing the agglomeration and destruction of nanoparticles during lyophilization [32]. In addition, sugars contain a large number of hydroxyl groups, so when the protein-bound water is removed from the nanoparticle surface, sugars can replace the bound water and, thus, inhibit the conformational change of the protein [33,34]. ## 3.1.3. Protective Effect of Alcohols on Nanoparticles Following the method in 2.5, different concentrations of mannitol ($0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, and $0.5\%$ (w/v)) and sorbitol ($0.005\%$, $0.025\%$, $0.05\%$, $0.25\%$, and $0.5\%$ (w/v)) were added to the ABC-NPs nanosuspension. The above mixture was freeze-dried, and the prepared ABC-NPs lyophilized powder was shown as O1–E1 in Figure 4A,B below. As the concentration of mannitol increased, the state of the lyophilized powder in the blank control samples showed irregular dispersion. The four lyophilized powder samples marked as A1, B1, C1, and D1 in Figure 4A were in the form of loose powder lumps, while sample E was in the state of adhesion. When the samples were re-dissolved according to the method in 2.6, the dissolution rate of the samples gradually increased with the increase of mannitol concentration. However, as shown in O2–E2 in Figure 4A, the flocculent pink complex was still present in the sample with a mannitol concentration of $0.005\%$, even after shaking for more than 5 min. When the samples were irradiated with a laser beam, the status of the samples was shown as O3–E3 in Figure 4A, and a clear red pathway was visible in each sample, indicating good dispersion of all samples. Figure 4B indicated that the particle size gradually decreased with the increase of mannitol concentration, the PDI gradually decreased, and the minimum Sa/Sb reached 1.32. The ABC-NPs lyophilized powder prepared according to the method in 2.5 was shown as O1–E1 in Figure 4B, but the four samples named O1, A1, B1, and C1 in Figure 4B were all irregularly shaped, while the two samples named D1 and E1 were in the form of adhesion. The samples were re-dissolved according to the method in 2.6, and the redissolution time was gradually reduced with the increase of sorbitol concentration. Finally, the samples were all completely dissolved, and no pink granular precipitation was seen. The samples were irradiated with the laser pointer, and the status of the samples is shown in Figure 4B(O3–E3). It can be seen that there is a clear red pathway in all samples, which indicates that the dispersion and homogeneity of the samples are good. Figure 4B exhibited that the particle size first gradually decreased and then gradually increased with the increase of sorbitol concentration. When the sorbitol concentration was $0.05\%$, the particle size was the smallest, reaching 454 nm, and the smallest Sa/Sb was 1.25. In addition, the PDI had a tendency to decrease gradually, and all of them are stable below 0.35. The above situation indicates that the samples have good homogeneity. The potentials are all above +40 mV, which indicates that the nanosuspensions are stable. There are several reasons for the above situation. Polyhydric alcohols have a large number of hydroxyl groups in their structure. As the concentration of the protectant increases, the water molecules around the nanoparticles in suspension are replaced by the hydroxyl groups of mannitol, and the hydroxyl groups also interact with the hydrogen bonds in the protein molecules. The above situation keeps the protein structure on the surface of nanoparticles unchanged, avoiding agglomeration of nanoparticles and maintaining nanoscale stability [35]. Sorbitol is a low molecular weight, so it may adsorb back to the nanoparticle surface and protect them [15]. In addition, sorbitol has a shallow redox potential and can replace proteins that are oxidized first, and is, therefore, considered an antioxidant [34]. ## 3.1.4. Protective Effect of Amino Acids on Nanoparticles The ABC-NPs lyophilized powder prepared according to the method in 2.5 is shown as O1–E1 in Figure 5A below. With the increase of the added L-lysine concentration ($0.0001\%$, $0.0002\%$, $0.001\%$, $0.005\%$, and $0.025\%$ (w/v)), these three lyophilized powder samples named O1, A1, and B1 in Figure 5A are powder blocks with a uniform skeleton. In contrast, the state of the lyophilized powder of the four samples called B1, C1, and D1 is irregularly scattered loose small particles, and sample E is noticeable loose gel-like particles. When the samples were re-dissolved according to the method in 2.6, the re-dissolution time was gradually extended as the concentration of L-lysine added increased, and even pink granular precipitates appeared at the bottom of the bottle for some samples. As shown in Figure 5A, the pink, coarse sediments in these four samples, named B2, C2, D2, and E2, increased with the increase of L-lysine concentration. The samples were irradiated with a laser pointer, and the status of the samples is shown in Figure 5A for O3–E3. There is a clear red pathway in only two samples, the blank sample, and sample A3, which indicates that the dispersion and homogeneity of the samples are good. However, the red pathways in the remaining four samples were unclear, indicating that the distribution and uniformity of the samples were relatively poor. It can be seen from Figure 5B that the particle size decreases and the PDI increase with increasing concentration for the samples with L-lysine concentrations of $0.0001\%$, $0.0002\%$, and $0.001\%$. However, the particle size and PDI increased substantially after the L-lysine concentration reached $0.005\%$. The minimum Sa/Sb was only 2.24, and the maximum was 11.6, with a negative potential. The studies of Kundu et al. [ 36,37] indicated that the positively charged side chain groups of lysine could interact with the groups of proteins to form spacers between nanoparticles, thus preventing nanoparticle agglomeration effectively. However, the study by Mohammed et al. [ 38] reported that the cryoprotection mechanism of lysine is biphasic (the cyroprotection profiles of the three amino acids tested were biphasic). As the concentration of amino acids increased, the lyophilization protection of amino acids showed first a positive and then a negative effect.. When the lysine:liposome concentration was 4:1, the vesicle size after lyophilization was comparable to the original size; when the lysine: liposome concentration was 10:1, it led to the aggregation of the vesicles, whose size increased significantly to 10 times the original size. Similar biphasic properties are also seen in the lyophilized cryoprotectants [38]. ## 3.2. Box–Behnken Design Analysis The particle size ratio Sa/Sb, the amount of lyophilization protectant, and the sample re-dissolution time can be used to screen the lyophilization protectant. Finally, three lyophilization protectants were selected from six types of lyophilization protectants, namely oligomeric mannose, maltose, and sorbitol. After re-dissolution, the minimum particle size ratios of the lyophilized powders prepared with the above-lyophilized protectants were 1.02, 1.25, and 1.25, respectively. According to the principle of lowest concentration to achieve the highest efficiency, after refinement of concentration, the optimal concentration range was selected as follows: the final concentration of oligomeric mannose was $0.3\%$, $0.4\%$, and $0.5\%$ (w/v), and the final concentration of oligomeric mannose $0.4\%$ (w/v) was selected as the center point of optimization; the final concentration of maltose was $0.4\%$, $0.5\%$, and $0.6\%$ (w/v), and the final concentration of maltose $0.4\%$ (w/v) as the center point of optimization; and final sorbitol concentrations of $0.15\%$, $0.25\%$, and $0.35\%$ (w/v), and final sorbitol concentration of $0.25\%$ (w/v) was selected as the center point of optimization. ## 3.2.1. Statistical Analysis and the Model Fitting Response surface is an effective optimization method when multiple factors affect the effect of lyophilization protectants on nanoparticle re-solubilization [39]. Based on single-factor experiments with the final concentrations of oligomeric mannose, maltose, and sorbitol as variables, a three-factor with three-level optimization experimental design was conducted. There were 17 runs for optimizing the three individual parameters, and the results are shown in Table 1. By applying multiple regression analysis, the relationship between response variables and the test variables was obtained as follows: $Y = 445.80$ − 8.63 × A − 21.5 × B + 0.13 × C + 5.00 × AB + 7.25 × AC − 2.00 × BC + 27.48 × A2 + 20.73 × B2 − 0.025 × C2. In this case, Y is the particle size of ABC-NPs after re-solubilization, and A, B, and C are the coding variables for oligomeric mannose, maltose, and sorbitol, respectively. We analyzed the regression equation, and the obtained ANOVA results were shown in Table 2. The p-value of the quadratic regression model was 0.0059, which was less than 0.01 (significant), while the lack-of-fit value was 0.7498, which was greater than 0.05 (not significant), indicating that the selection of the data model was reasonable; R2 (determination coefficient) and R2Adj (adjusted determination coefficient) were 0.7990 and 0.9120, indicating a good fit of the quadratic regression equation to the test results. In addition, the lower C.V$.\%$ (coefficient of variation) is 2.49, which shows high reliability and accuracy. Based on the magnitude of the F-value, the strength of the effect of three key factors of lyophilized protectants on the particle size (variation) is maltose > oligomeric mannose > sorbitol. In summary, the model can analyze and predict the particle size change after nanoparticle lyophilization protectant addition and determine the optimal protectant compounding conditions. ## 3.2.2. Analysis of Response Surface Plot The interactions between the optimal parameters and factors can be visualized from response surface analysis plots [40], while the shape of the contours can reflect the trend in the strength of the interactions between the factors [41]. Within the response surface methodology design, the interaction factors were AB, AC, and BC, and their response surfaces and contours are shown in Figure 6. The three-dimensional contour plots of oligomeric mannose concentration and maltose concentration are shown in Figure 6A. When the concentration of sorbitol is certain, the particle size of ABC-NPs shows a trend of increasing and then decreasing with the increase of oligomeric mannose and maltose concentrations. However, the interactions between oligomeric mannose and sorbitol and maltose and sorbitol were weak, as shown in Figure 6B,C. ## 3.3. Optimization of Response Surface Experimental Results The quadratic polynomial regression model was optimally solved using Design Expert 8.0.6 software, and nine scenarios were obtained. The optimal values of the final concentrations of oligomeric mannose, maltose, and sorbitol in the scenarios were $0.46\%$, $0.45\%$, and $0.05\%$ (w/v), respectively. The theoretical minimum particle size value of ABC-NPs lyophilized powder prepared with the above concentrations of lyophilized protectant was 357 nm after re-dissolution. To check the reliability of the results, three validation experiments were conducted using the optimal conditions optimized by the response surface. The experimental results were 472 nm, and the ratio of the re-solubilized particle size to the initial particle size Sa/Sb of ABC-NPs suspension was 1.32. The values of the experimental results were close to the theoretical values, indicating that the model can better predict the protective effect of lyophilized protectant concentration on ABC-NPs. In our research, the natural polymers such as polysaccharides have the advantages of biodegradability, safety, low toxicity, and good stability, and they are in ionic form in water and can spontaneously bind to proteins. In addition, the protein–polysaccharide system can prevent the structure and properties of biochemical drugs from being destroyed, and nanoparticles for nanocarriers can achieve slow and controlled release of drugs, so it has become a central tool for nanodrug delivery research [42]. Maltose and sorbitol are commonly used as lyophilization protectants [43,44,45,46]. Our experiments demonstrated that maltose and sorbitol can effectively protect ABC-NPs with J-type astaxanthin aggregates which have various bioactive function [47,48], presumably because their surfaces contain a large number of hydrogen bonds that can provide support for nanoparticles and have a high glass transition temperature [15,16,20]. Oligomannose is not commonly seen in previous studies. This experiment verified that oligomannose has a practical protective effect on nanoparticles in the protein–polysaccharide system. The samples after adding the compound lyophilized protectant were pink in color; the samples were fluffy and completely dissolved within 10 s after shaking with water, indicating that the effect of the compound protectant was significantly better than that of the single protectant. In other international studies, the particle size compounding ratio is generally between 1.02 and 1.52 [49,50,51,52,53], while our particle size compounding ratio is 1.32. This indicates that our results are at a stable level internationally and can provide theoretical support for future studies on protein–polysaccharide systems and lyophilized protectants. ## 4. Conclusions Freeze drying is an effective way to improve the long-term physical stability of nanosuspensions during preservation. Freeze-drying process agglomerates the particles, whereas the addition of cryoprotectants and ultrasonication procedure can reduce the particle size. In this study, the mass-produced homogeneous ABC-NPs was successfully carried out, and the particle sizes were about 320–360 nm, the PDIs were all below 0.3, and the Zeta potentials were all around +15 mV. The ABC-NPs nano-suspension was freeze-dried to obtain a pink fluffy lyophilized powder with a redissolving particle size about 550–680 nm. The optimal ratio of lyophilized protectants was obtained through response surface methods as $0.46\%$ oligomannose, $0.44\%$ maltose, and $0.05\%$ sorbitol (w/v). Under these conditions, the re-dissolved particle size of ABC-NPs lyophilized powder was 472 nm and the Sa/Sb value was 1.32. The color of lyophilized powder remained unchanged and the structure was looser, indicating that the structure and activity of astaxanthin nanoparticles were well preserved. The Sa/Sb value is an important factor in evaluating the effectiveness of lyophilized protectants.1) These results of this study provide theoretical foundation and academic references for the long-term preservation, deep processing, and resource utilization of ABC-NPs, and also expand the applications of nanocarriers in food, pharmaceutical, and biological fields. ## References 1. Liu Y.J., Yang L., Guo Y.L., Zhang T., Qiao X., Wang J.F., Xu J., Xue C.H.. Hydrophilic astaxanthin: PEGylated astaxanthin fights diabetes by enhancing the solubility and oral absorbability. J. Agric. Food Chem. (2020) 68 3649-3655. DOI: 10.1021/acs.jafc.0c00784 2. Mularczyk M., Bourebaba N., Marycz K., Bourebaba L.. Astaxanthin Carotenoid Modulates Oxidative Stress in Adipose-Derived Stromal Cells Isolated from Equine Metabolic Syndrome Affected Horses by Targeting Mitochondrial Biogenesis. Biomolecules (2022) 12. DOI: 10.3390/biom12081039 3. Li F., Cai M., Lin M., Huang X., Wang J., Ke H., Wang C., Zheng X., Chen D., Yang S.. Enhanced Biomass and Astaxanthin Production of Haematococcus pluvialis by a Cell Transformation Strategy with Optimized Initial Biomass Density. Mar. Drugs (2020) 18. DOI: 10.3390/md18070341 4. Li F., Cai M., Lin M., Huang X., Wang J., Ke H., Zheng X., Chen D., Wang C., Wu S.. Differences between Motile and Nonmotile Cells of Haematococcus pluvialis in the Production of Astaxanthin at Different Light Intensities. Mar. Drugs (2019) 17. DOI: 10.3390/md17010039 5. Liu G., Hu M., Zhao Z., Lin Q., Wei D., Jiang Y.. Enhancing the stability of astaxanthin by encapsulation in poly (l-lactic acid) microspheres using a supercritical anti-solvent process. Particuology (2018) 44 54-62. DOI: 10.1016/j.partic.2018.04.006 6. Oslan S.N.H., Shoparwe N.F., Yusoff A.H., Rahim A.A., Chang C.S., Tan J.S., Oslan S.N., Arumugam K.A., Ariff A.B., Sulaiman A.Z.. A Review on Haematococcus pluvialis Bioprocess Optimization of Green and Red Stage Culture Conditions for the Production of Natural Astaxanthin. Biomolecules (2021) 11. DOI: 10.3390/biom11020256 7. Fania L., Sampogna F., Ricci F., Hyeraci M., Abeni D.. Systemic Photoprotection in Skin Cancer Prevention: Knowledge among Dermatologists. Biomolecules (2021) 11. DOI: 10.3390/biom11020332 8. Lu L.P., Hu T.P., Xu Z.G.. Structural characterization of astaxanthin aggregates as revealed by analysis and simulation of optical spectra. Spectrochim. Part A Mol. Biomol. Spectrosc. (2017) 185 85-92. DOI: 10.1016/j.saa.2017.05.031 9. Dai M., Li C., Yang Z., Sui Z., Li J., Dong P., Liang X.. The Astaxanthin Aggregation Pattern Greatly Influences Its Antioxidant Activity: A Comparative Study in Caco-2 Cells. Antioxidants (2020) 9. DOI: 10.3390/antiox9020126 10. Zajac G., Machalska E., Kaczor A., Kessler J., Bour P., Baranska M.. Structure of supramolecular astaxanthin aggregates revealed by molecular dynamics and electronic circular dichroism spectroscopy. Phys. Chem. Chem. Phys. (2018) 20 18038-18046. DOI: 10.1039/C8CP01742E 11. Mao X., Tian Y., Sun R., Wang Q., Huang J., Xia Q.. Stability study and in vitro evaluation of astaxanthin nanostructured lipid carriers in food industry. Integr. Ferroelectr. (2019) 200 208-216. DOI: 10.1080/10584587.2019.1592626 12. Wang Q., Zhao Y.Y., Guan L., Zhang Y.P., Dang Q.F., Dong P., Li J., Liang X.G.. Preparation of astaxanthin-loaded DNA/chitosan nanoparticles for improved cellular uptake and antioxidation capability. Food Chem. (2017) 227 9-15. DOI: 10.1016/j.foodchem.2017.01.081 13. Shen X., Zhao C., Lu J., Guo M.. Physicochemical Properties of Whey-Protein-Stabilized Astaxanthin Nanodispersion and Its Transport via a Caco-2 Monolayer. J. Agric. Food Chem. (2018) 66 1472-1478. DOI: 10.1021/acs.jafc.7b05284 14. Salar-Behzadi S., Wu S., Toegel S., Hofrichter M., Altenburger I., Unger F.M., Viernstein H.. Impact of heat treatment and spray drying on cellular properties and culturability of Bifidobacterium bifidum BB-12. Food Res. Int. (2013) 54 93-101. DOI: 10.1016/j.foodres.2013.05.024 15. Ambros S., Hofer F., Kulozik U.. Impact of protectants on drying kinetics and viability of microwave freeze-dried Lactobacillus paracasei ssp. paracasei F19. J. Food Process. Preserv. (2019) 43 e13859. DOI: 10.1111/jfpp.13859 16. Luo W.C., Beringhs A.O., Kim R., Zhang W., Patel S.M., Boghner R.H., Robin H., Lu X.L.. Impact of formulation on the quality and stability of freeze-dried nanoparticles. J. Eur. J. Pharm. Biopharm. (2021) 169 256-267. DOI: 10.1016/j.ejpb.2021.10.014 17. Mattern M., Winter G., Kohnert U., Lee G.. Formulation of Pro-teins in Vacuum-Dried Glasses, T. Process and Storage Stability in Sugar-free Amino Acid Systems. Pharm. Dev. Technol. (1999) 4 199-208. DOI: 10.1081/PDT-100101354 18. Cheng Z.Y., Yan X., Wu J.Y., Weng P.F., Wu Z.F.. Effects of freeze drying in complex lyoprotectants on the survival, and membrane fatty acid composition of Lactobacillus plantarum L1 and Lactobacillus fermentum L2. Cryobiology (2022) 105 1-9. DOI: 10.1016/j.cryobiol.2022.01.003 19. Andreani T., Kiill C.P., de Souza A.L.R., Fangueiro J.F., Doktorovová S., Garcia M.L., Silva A.M., Souto E.B.. Effect of cryoprotectants on the reconstitution of silica nanoparticles produced by sol–gel technology. J. Therm. Anal. Calorim. (2014) 120 1001-1007. DOI: 10.1007/s10973-014-4275-4 20. Santivarangkna C., Aschenbrenner M., Kulozik U., Forest P.. Role of glassystate on stabilities of freeze-dried probiotics. J. Food Sci. (2011) 76 R152-R156. DOI: 10.1111/j.1750-3841.2011.02347.x 21. Levine H., Slade L.. Another view of trehalose for drying and stabilizing biological materials. BioPharm (1992) 5 36-40 22. Mattar P., Toledo-Valenzuela L., Hernández-Cáceres M.P., Peña-Oyarzún D., Morselli E., Perez-Leighton C.. Integrating the effects of sucrose intake on the brain and white adipose tissue: Could autophagy be a possible link?. Obesity (2022) 30 1143-1155. DOI: 10.1002/oby.23411 23. Weiwei L., Peng W., Zhaokai W., Xiangzhi L.. Study on the vacuum freeze-drying process on microalgae oil nanoemulsion powder. Mar. Sci. (2017) 41 51-56 24. Zhao Y., Liu J., Zhang S., Wang Z., Jia H., Oda H., Li R.. Fabrication and characterization of the H/J-type aggregates astaxanthin/bovine serum albumin/chitosan nanoparticles. Int. J. Biol. Macromol. (2022) 223 (Pt A) 1186-1195. DOI: 10.1016/j.ijbiomac.2022.11.006 25. Sameti M., Bohr G., Kumar M.N.V.R., Kneuer C., Bakowsky U., Nacken M., Schimdt H., Lehr C.M.. Stabilisation by freeze-drying of cationically modified silica nanoparticles for gene delivery. Int. J. Pharm. (2003) 266 51-60. DOI: 10.1016/S0378-5173(03)00380-6 26. Cacela C., Hincha D.K.. Monosaccharide composition, chain length and linkage type influence the inter-actions of oligosaccharides with dry phosphatidylcholine membranes. Biochim. Et Biophys. Acta (2006) 1758 680-691. DOI: 10.1016/j.bbamem.2006.04.005 27. Man B.G., Sue H.C., Sung W.K.. Some observations in freeze-drying ofrecombinant bioluminescent Escherichia coli for toxicity monitoring. J. Biotechnol. (2001) 88 95-105. DOI: 10.1081/PDT-100101354 28. Hincha D.K., Livingston D.P., Premakumar R., Zuther E., Obel N., Cacela C., Heyer A.G.. Fructans from oat and rye: Composition and effects on membrane stability during drying. Biochim. Et Biophys. Acta (2007) 1768 1611-1619. DOI: 10.1016/j.bbamem.2007.03.011 29. Chen W., Chen H., Xia Y., Zhao J., Tian F., Zhang H.. Production, purification, and characterization of a potentialthermostable galactosidase for milk lactose hydrolysis from Bacillus stearothermophilus. J. Dairy Sci. (2008) 91 1751-1758. DOI: 10.3168/jds.2007-617 30. Holm T.P., Meng-Lund H., Rantanen J., Jorgensen L., Grohganz H.. Screening of novel excipients for freeze-dried protein formulations. Eur. J. Pharm. Biopharm. (2021) 160 55-64. DOI: 10.1016/j.ejpb.2021.01.008 31. Yu M., Xiao B., Hao X., Tan J., Gu J., Wang G., Wang W., Zhang Y.. Pumpkin polysaccharide preparation, simulated gastrointestinal digestion, and in vivo biodistribution. Int. J. Biol. Macromol. (2019) 1 1293-1303. DOI: 10.1016/j.ijbiomac.2019.09.037 32. Kadoya S., Fujii K., Izutsu K., Yonemochi E., Terada K., Yomota C., Kawanishi T.. Freeze-drying of proteins with glass-forming oligosaccharide-derived sugar alcohols. Int. J. Pharm. (2010) 389 107-113. DOI: 10.1016/j.ijpharm.2010.01.027 33. Kurtmann L., Carlsen C.U., Skibsted L.H., Risbo J.. Water activity-temperature state diagrams of freeze-driedLactobacillus acidophilus (La-5): Influence of physical state on bacterial survival during storage. Biotechnol. Prog. (2009) 25 265-270. DOI: 10.1002/btpr.96 34. Santivarangkna C., Higl B., Foerst P.. Protection mechanisms of sugars during different stages of preparation process of dried lactic acid starter cultures. Food Microbiol. (2008) 25 429-441. DOI: 10.1016/j.fm.2007.12.004 35. Li W., Zhang X., Jing S., Xin X., Chen K., Chen D., Hu H.. Preparation of Calcium Phosphate/pDNA Nanoparticles for Exogenous Gene Delivery by Co-Precipitation Method: Optimization of Formulation Variables Using Box-Behnken Design. J. Pharm. Sci. (2017) 106 2053-2059. DOI: 10.1016/j.xphs.2017.04.049 36. Kundu C.N., Das K., Majumder G.C.. Effect of Amino Acids on Goat Cauda Epididymal Sperm Cryopreservation Using a Chemically Defined Model System. Cryobiology (2001) 42 21-27. DOI: 10.1006/cryo.2001.2296 37. Lalonde R.J., Lepock J.R., Kruuv J.. Site of freeze-thaw damage and cryoprotection by amino acids of the calcium ATPase of sarcoplasmic reticulum. Biochim. Et Biophys. Acta (BBA)-Protein Struct. Mol. Enzymol. (1991) 1079 128-138. DOI: 10.1016/0167-4838(91)90117-I 38. Mohammed A.R., Coombes A.G.A., Perrie Y.. Amino acids as cryoprotectants for liposomal delivery systems. Eur. J. Pharm. Sci. (2007) 30 406-413. DOI: 10.1016/j.ejps.2007.01.001 39. Xie Y., Cao W.J., Krishnan S., Lin H., Cauchon N.. Characterization of mannitol polymorphic forms in lyophilized protein formulations using a multivariate curve resolution (MCR)-based Raman spectroscopic method. Pharm. Res. (2008) 25 2292-2301. DOI: 10.1007/s11095-008-9624-1 40. Patil P., Khairnar G., Naik J.. Preparation and statistical optimization of Losartan Potassium loaded nanoparticles using Box Behnken factorial design: Microreactor precipitation. Chem. Eng. Res. Des. (2015) 104 98-109. DOI: 10.1016/j.cherd.2015.07.021 41. Muralidhar R.V., Chirumamila R.R.. A response surface approach for the comparison of lipase production by canida cylindracea using two different carbon sources. Biochem. Eng. J. (2001) 9 17-23. DOI: 10.1016/S1369-703X(01)00117-6 42. Umerska A., Paluch K.J., Martinez M.-J.S., Corrigan O.I., Medina C., Tajber L.. Freeze drying of polyelectrolyte complex nanoparticles: Effect of nanoparticle composition and cryoprotectant selection. Int. J. Pharm. (2018) 552 27-38. DOI: 10.1016/j.ijpharm.2018.09.035 43. Wong C.Y., Al-Salami H., Dass C.R.. Lyophilisation Improves Bioactivity and Stability of Insulin-Loaded Polymeric-Oligonucleotide Nanoparticles for Diabetes Treatment. AAPS PharmSciTech (2020) 21 108. DOI: 10.1208/s12249-020-01648-6 44. Fonte P., Andrade F., Azevedo C., Pinto J., Seabra V., van de Weert M., Reis S., Sarmento B.. Effect of the Freezing Step in the Stability and Bioactivity of Protein-Loaded PLGA Nanoparticles Upon Lyophilization. Pharm. Res. (2016) 33 2777-2793. DOI: 10.1007/s11095-016-2004-3 45. Walayat N., Xiong H., Xiong Z., Moreno H.M., Nawaz A., Niaz N., Randhawa M.A.. Role of Cryoprotectants in Surimi and Factors Affecting Surimi Gel Properties: A Review. Food Rev. Int. (2020) 38 1103-1122. DOI: 10.1080/87559129.2020.1768403 46. Garay-Jimenez J.C., Turos E.. A convenient method to prepare emulsified polyacrylate nanoparticles from for drug delivery applications. Bioorg. Med. Chem. Lett. (2011) 21 4589-4591. DOI: 10.1016/j.bmcl.2011.05.104 47. Ding L., Yang J., Yin K., Cheng H., Li J., Xue C.. The spatial arrangement of astaxanthin in bilayers greatly influenced the structural stability of DPPC liposomes. Colloids Surf. B Biointerfaces (2022) 212 112383. DOI: 10.1016/j.colsurfb.2022.112383 48. Zhao Y., Liu J., Guan L., Zhang Y., Dong P., Li J., Liang X., Komiyama M.. Fabrication of aqueous nanodispersion from natural DNA and chitosan as eminent carriers for water-insoluble bioactives. Int. J. Biol. Macromol. (2018) 118 (Pt A) 263-270. DOI: 10.1016/j.ijbiomac.2018.05.054 49. Bozdag S., Dillen K., Vandervoort J., Ludwig A.. The effect of freeze-drying with different cryoprotectants and gamma-irradiation sterilization on the characteristics of ciprofloxacin HCl-loaded poly(D,L-lactide-glycolide) nanoparticles. J. Pharm. Pharmacol. (2005) 57 699-707. DOI: 10.1211/0022357056145 50. Amis T.M., Renukuntla J., Bolla P.K., Clark B.A.. Selection of Cryoprotectant in Lyophilization of Progesterone-Loaded Stearic Acid Solid Lipid Nanoparticles. Pharmaceutics (2020) 12. DOI: 10.3390/pharmaceutics12090892 51. Cao J., Xu H., Xu Q., Chen B., Kan H.. Fine Particulate Matter Constituents and Cardiopulmonary Mortality in a Heavily Polluted Chinese City. Environ. Health Perspect. (2012) 120 373-378. DOI: 10.1289/ehp.1103671 52. Alihosseini F., Ghaffari S., Dabirsiaghi A.R., Haghighat S.. Freeze-drying of ampicillin solid lipid nanoparticles using mannitol as cryoprotectant. Braz. J. Pharm. Sci. (2015) 51 797-802. DOI: 10.1590/S1984-82502015000400005 53. Varshosaz J., Ghaffari S., Khoshayand M.R., Atyabi F., Dehkordi A.J., Kobarfard F.. Optimization of freeze-drying condition of amikacin solid lipid nanoparticles using D-optimal experimental design. Pharm. Dev. Technol. (2010) 17 187-194. DOI: 10.3109/10837450.2010.529149
--- title: Roles of miR-196a and miR-196b in Zebrafish Motor Function authors: - Chunyan Yuan - Huaping Xie - Xiangding Chen - Shunling Yuan journal: Biomolecules year: 2023 pmcid: PMC10046552 doi: 10.3390/biom13030554 license: CC BY 4.0 --- # Roles of miR-196a and miR-196b in Zebrafish Motor Function ## Abstract Background: The exertion of motor function depends on various tissues, such as bones and muscles. miR-196 has been widely studied in cancer and other fields, but its effect on bone and skeletal muscle is rarely reported. In order to explore the role of miR-196 family in bone and skeletal muscle, we used the previously successfully constructed miR-196a-1 and miR-196b gene knockout zebrafish animal models for research. Methods: The behavioral trajectories of zebrafish from 4 days post-fertilization (dpf) to 7 dpf were detected to analyze the effect of miR-196a-1 and miR-196b on motor ability. Hematoxylin-eosin (HE) staining and transmission electron microscopy (TEM) were used to detect the dorsal muscle tissue of zebrafish. The bone tissue of zebrafish was detected by microcomputed tomography (micro-CT). Real-time PCR was used to detect the expression levels of related genes, including vcp, dpm1, acta1b, mylpfb, col1a1a, bmp8a, gdf6a, and fgfr3. Results: The behavioral test showed that the total behavioral trajectory, movement time, and movement speed of zebrafish larvae were decreased in the miR-196a-1 and miR-196b gene knockout lines. Muscle tissue analysis showed that the structure of muscle fibers in the zebrafish lacking miR-196a-1 and miR-196b was abnormal and was characterized by vacuolar degeneration of muscle fibers, intranuclear migration, melanin deposition, and inflammatory cell infiltration. Bone CT examination revealed decreased bone mineral density and trabecular bone number. The real-time PCR results showed that the expression levels of vcp, dpm1, gdf6a, fgfr3, and col1a1a were decreased in the miR-196b gene knockout group. The expression levels of dpm1, acta1b, mylpfb, gdf6a, and col1a1a were decreased, and the expression level of fgfr3 was increased in the miR-196b gene knockout group compared with the wild-type group. Conclusions: miR-196a-1 and miR-196b play an important role in muscle fiber structure, bone mineral density, and bone trabecular quantity by affecting the expression of vcp, dpm1, acta1b, mylpfb, gdf6a, fgfr3, and col1a1a and then affect the function of the motor system ## 1. Introduction The motor system of the body is extremely complex, and its function depends on tissues such as bone and muscle [1,2,3]. Bone or skeletal muscle is needed to maintain the whole body, and abnormal function can cause an inability to properly exercise and lead to a variety of chronic diseases, such as osteoporosis, fractures, osseous arthritis, muscle atrophy, and issues with the lumbar disc. Diseases such as arthritis, gout, and weakening or loss of body movement seriously affect the daily life of an individual and their quality of life, resulting in a major economic burden to society and families [4,5,6]. In recent years, the incidence of motor system-related diseases has increased, and there is no effective treatment for many diseases. Therefore, it is important to study and further explore the role and regulatory mechanism of bone and muscle for the defense, treatment, recovery, and improvement of motor function in motor system diseases. MicroRNAs (miRNAs) are a class of highly conserved noncoding small RNAs that are approximately 22 to 25 nucleotides in length, widely exist in a variety of eukaryotic cells and a large number of species, and play an important role in the growth, development, metabolism, regeneration, and physiological functions of bone and skeletal muscle [7,8,9,10,11]. Previous studies [12,13] have reported that miR-196 is related to bone mineral density and muscle. Studies regulating the expression level of miR-196 showed that this molecule plays an important role in preventing the occurrence of bone and skeletal muscle diseases and improving related diseases, suggesting that miR-196 may be involved in bone or skeletal muscle. miR-196 is encoded by the homologous HOX family, is highly conserved among different species, and includes miR-196a (mature sequence: 3′-GGG UUG UUG UAC UUU GAU GGA U-5′) and miR-196b (mature sequence: 3′-GGG UUG UUG UCC UUU GAU GGA U-5′), which are highly homologous with only one nucleotide difference [14,15]. A study on osteosarcoma reported that the expression level of miR-196a in osteosarcoma cells was significantly higher than that in normal tissues, and this molecule ultimately promoted the metastasis of tumor cells and extracellular matrix transformation by targeting the 3′-UTR of HOXA5 mRNA [16]. Another study [17] on osteosarcoma reported that overexpressed miR-196a promotes cell proliferation and inhibits cell apoptosis through the PTEN/Akt/FOXO1 signaling pathway. Zhong et al. [ 13] found, in a mouse model of osteoporosis, that overexpression of miR-196a could inhibit the expression of the GNAS gene through the Hedgehog signaling pathway and promote osteogenic differentiation in mice. In addition, studies [18] have shown that miR-196a-5p inhibits the formation of osteoclast-like cells and mitochondrial energy metabolism in mouse cells. A miRNA microarray detection of osteosarcoma found that the expression of miR-196a and miR-196b was upregulated [19]. Another exosome experiment [20] examining the source of bone marrow stromal cells found that miR-196a played an important role in the differentiation of osteoblasts and the expression of related osteoblast genes. At present, there have been few studies on miR-196a and miR-196b in the bones and muscles of the motor system. In a study on the expression pattern of miR-196 in zebrafish, it was found that pri-miR-196a1 was expressed in the tail and trunk of 24 zebrafish embryos [21]. In summary, we speculate that miR-196a may play a role in bone or skeletal muscle. In zebrafish, skeletal muscle and bone make up a large part of the body trunk and are highly similar to human muscle both molecularly and histologically, making them suitable for the study of bone and muscle diseases [22]. Additionally, the relatively simple genome and simple genetic manipulation of zebrafish are advantageous in myopathy studies [23].To explore the role of miR-196a-1 and miR-196b in the motor system, we constructed miR-196a and miR-196b gene knockout lines. In this study, we tested the behavior and motor function of miR-196a-1 and miR-196b gene knockout embryos, as well as muscle and bone tissue structure, to explore the influence of miR-196a-1 and miR-196b on the muscle of the operating system. The relevant results will further clarify the role of miR-196 in the body and contribute to its clinical application. ## 2.1. Zebrafish Lines The miR-196a-1 and miR-196b gene knockout lines and the miR-196a-1 and miR-196b gene double knockout zebrafish lines were obtained for this experiment (College of Life Sciences, Hunan Normal University, Changsha, China). As we reported earlier, homozygous mutant zebrafish lines were constructed using CRISPR/Cas9 gene editing technology and were screened by gene sequencing [24]. Homozygous mutants of gene knockout zebrafish showed no abnormal embryonic development after breeding. Twelve zebrafish were randomly selected form each group, for a total of 48 zebrafish. ## 2.2. Zebrafish Maintenance Zebrafish strains were raised in a water temperature of 28 °C and pH 6.5~7.4, with alternating cycles of 14 h light and 10 h dark in the Zebrafish Laboratory of the College of Life Sciences, Hunan Normal University. According to previous reports [25], the embryos were incubated at 28.5 °C water temperature, and paramecium (Heading, Tianjin, China) was given 5 days later. The young fish were transferred to a feeding system rack (ESEN, Beijing, China) after the 14th day and fed with heading (Tianjin, China) twice a day at a fixed time in the morning and evening. ## 2.3. Behavioral Test of the Zebrafish Model The behavioral movement of zebrafish embryos at 4 dpf, 5 dpf, 6 dpf, and 7 dpf was detected in 24-well plates using the ViewPoint system (Lyon, France). Zebrafish embryos were cultured in 24-well cell culture plates filled with 60 μg/mL Instant Ocean salt mix with 1 larva per well. The total distance traveled by larvae at different speeds, the total duration of movement at different speeds, and the number of movements were recorded within 10 min. ## 2.4. HE Staining of Muscle Tissue Zebrafish back muscle tissue was collected after feeding ten months, anhydrous ethanol dehydration (Characters, Shanghai, China) was performed, xylene (SINPHARM, Beijing, China) solution was added, and paraffin infiltration and the embedding of slices (Leica, Heidelberg, Germany) were performed. The muscle tissue was sliced after dewaxing, hydration and hematoxylin (St. Louis, MO, USA) and eosin (Solarbio, Beijing, China) dyeing, and finally sliced and photographed. ## 2.5. Transmission Electron Microscopy of Muscle The back muscle tissue of ten-month-old zebrafish (1 mm3) was fixed in 4 °C $2.5\%$ glutaraldehyde solution. The fixative solution was discarded, and cells were transferred to phosphate-buffered saline. Next, cells were fixed in $1\%$ osmic acid for 1–2 h, and dehydration was carried out by incubation in $30\%$ ethanol for 10 min, $50\%$ ethanol for 10 min, $70\%$ uranyl acetate in ethanol (stained before embedding) for 3 h or overnight, $80\%$ ethanol for 10 min, $95\%$ ethanol for 15 min, $100\%$ ethanol twice for 50 min each, and propylene oxide for 30 min. Next, samples were incubated in propylene oxide: epoxy resin (1:1) for 1–2 h and then in pure epoxy resin for 2–3 h. After embedding in pure epoxy resin, samples were baked in an oven at 40 °C for 12 h and then at 60 °C for 48 h. Samples were then cut into ultra-thin sections and placed on copper grids. Staining was then performed with lead and uranium stain, and images were acquired using a Japan Electronics JEM1400 transmission electron microscope and recorded with a Morada G3 digital camera. ## 2.6. Bone Micro-CT Examination Ten-month-old zebrafish were collected, fixed with $4\%$ paraformaldehyde, and then sent to Xidian University (Xi’an, China) for micro-CT (microcomputed tomography) analysis. A hardware system (Xidian University, Xi’an, China) composed of an X-ray tube, rotary table, detector, and acquisition card was used for shooting detection. The analysis was carried out by the Hiscan Analyzer software system (Hiscan, Suzhou, China). According to the section image information obtained from the scan, graying, binarization, head image segmentation of the target area, and threshold segmentation were carried out sequentially; the influence of cortical bone (or cancellous bone), soft tissue, and fluid in the medullary cavity was excluded; and finally, statistical analyses were carried out. ## 2.7. Real-Time PCR Analysis of Skeletal Muscle and Skeletal-Related Gene Expression Levels The dorsal muscle tissue and bone tissue of 10-year-old zebrafish were collected, total RNA was extracted from tissue, and cDNA was synthesized by reverse transcriptase (Thermo, MA, USA). Then, a SYBR Green qRCR kit (TaKaRa, Tokyo, Japan) and specific primers were used for amplification. The tested genes included vcp (R:CAGAGAAGAACGCACCAGCCATC, F:CCCTTTGCTTGAGTCCGTCCATC); dpm1 (R:AGCCGGAGAAGTAATGCGAA, F:CGGGAGGTTTTCTCGCTCAT); acta1b (R:CTGGCACCACACCTTCTACAATGAG, F:GGTCATCTTCTCACGGTTAGCCTTG); mylpfb (R:GATGTGCTGGCAACAATGGG, F:GCGCCCTTTAGCTTTTCACC); col1a1a (R:GCAGCACTTCCAGCACCCTTAC, F:AGGAGCACCAGCAATACCAGGAG); bmp8a (R:ATGGACAGACACGAGGTTGAGATTG, F:TACACACAGAGGGAGGAAGATGGAC); gdf6a (R:ACCGTCTGGACAGGATTCACTAAGG, F:TCAACAGGTGCTCGTCTACACATTC); and fgfr3 (R:AGATGAGGACGAGGCAGGTAATGG, F:CAGCAGGACAGCGGAACTTGAC). The cycle threshold (CT) value was collected, and the formula 2−ΔΔC was used to calculate the relative expression level of target genes in each sample. ## 2.8. Statistical Analysis Data between different groups are expressed as the mean ± variance. The experiment was repeated three times. SPSS 20.0 software (SPSS, Inc., Chicago, IL, USA) was used for statistical analysis. A two-tailed Student’s t test or one-way analysis of variance was used to compare the differences between different groups. $p \leq 0.05$ was considered statistically significant. ## 3.1. Muscle Behavior Analysis of miR-196a-1 and miR-196b Gene Knockout Zebrafish The motor behaviors and behavioral trajectories of zebrafish embryos on the fourth, fifth, sixth, and seventh days were recorded for 10 min every day, and the experiment was repeated three times. We captured and recorded the activity track of each group of zebrafish in a 24-hole culture plate and selected three holes for each group to display, as shown in Figure 1. The behavior track and activity time of zebrafish larvae of the four groups in the culture plate were collected, and the total movement distance, total duration, movement speed, and movement times of the zebrafish in each group were calculated. The behavior trajectories of zebrafish larvae were observed continuously for 4 days. According to the detection data, the total moving distance and moving times of both wild-type zebrafish and miR-196a-1 and miR-196b gene knockout zebrafish increased obviously with the development of larvae. The behavioral data of the fourth, fifth, sixth, and seventh days were calculated, and we found that the total movement distance of the miR-196a-1 or miR-196b−/− group (Figure 2A) was smaller than that of the wild-type zebrafish, and the differences were statistically significant ($$n = 24$$, $p \leq 0.05$). Compared with that of the miR-196a-1 or miR-196b knockout groups, the total travel distance of the miR-196a-1 and miR-196b double knockout groups was further reduced, but the difference between the groups was not statistically significant. The total movement time of zebrafish was statistically analyzed (Figure 2B). The total movement time of the miR-196a-1 gene knockout group, miR-196b gene knockout group, and double knockout group was decreased compared with that of the wild-type control group. In a comparison of the movement speed of each group (Figure 2C), the speed of the miR-196a-1 or miR-196b gene knockdown group decreased, but the difference between groups was not significant. The number of movements of different groups (Figure 2D) was determined, the number of movements of the miR-196a-1 or miR-196b gene knockdown group decreased, and further decreased after the double knockdown of miR-196a-1 and miR-196b; the differences between some groups were statistically significant. In summary, the experimental results suggest that the total behavioral trajectory of zebrafish larvae is reduced, and the movement time and speed are also reduced after miR-196a-1 and miR-196b gene knockout, indicating that the motor ability is decreased. When both the miR-196a-1 and miR-196b genes were knocked out, the motor ability was further decreased compared with that of the miR-196a-1 or miR-196b single knockout groups. However, there was no significant difference between the miR-196a-1 and miR-196b gene knockdown groups. These results indicated that both the miR-196a-1 and miR-196b genes play a role in the motor ability of zebrafish. ## 3.2. HE Staining of Muscle Tissue of the miR-196a-1 and miR-196b Knockout Zebrafish Model To further explore the influence of the miR-196a-1 and miR-196b genes on motor function, we assessed the muscle and bone tissues of zebrafish with gene knockout. The dorsal muscle tissue of zebrafish from the wild-type, miR-196a-1 gene knockout, miR-196b gene knockout, and miR-196a-1 and miR-196b gene double knockout groups were stained with HE, and the results of longitudinal and transverse sections are shown in Figure 3. The transverse section of the dorsal muscle tissue of wild-type zebrafish (Figure 3A1) showed that the muscle fibers were polyangular and uniform in size. The muscle nuclei were distributed around the muscle fibers, and the nuclear membrane was intact. In the longitudinal section of zebrafish in the wild-type group (Figure 3A2), muscle fibers were arranged neatly with uniform thickness, and a small amount of inflammatory cell infiltration was occasionally observed in the interstitium. In the transverse section and longitudinal section of the muscle in the miR-196a-1 knockout group (Figure 3B1), the tissue structure of the muscle fibers was slightly abnormal, some muscle fibers showed vacuolar degeneration, and a small amount of inflammatory cells could be seen in the intercellular substance. In the miR-196b knockout zebrafish group (Figure 3C), muscle fibers were slightly abnormal, vacuolar degeneration of some muscle fibers was present, and a small amount of inflammatory cells were observed in the transverse section of the muscle. A small number of inflammatory cells in the interstitial tissue and individual melanin deposition could be seen in the longitudinal muscle section (Figure 3C2). The tissue structure of muscle fibers was slightly abnormal in the miR-196a-1 and miR-196b double knockout groups (Figure 3D1). The shape of the muscle fibers was irregular, with enlarged and rounded muscle fibers, vacuolar degeneration of some muscle fibers, infiltration of a small amount of inflammatory cells in the stroma, individual melanin deposition in the tissue, and intranuclear migration of individual muscle cells. Longitudinal sections (Figure 3D2) showed that individual inflammatory cells infiltrated the interstitium, and individual muscle cells were hypertrophic. The results suggest that the absence of miR-196a-1 or miR-196b has an effect on the morphological structure of muscle fibers in dorsal muscle tissue. ## 3.3. Electron Microscopy Results of Muscle Tissue of the miR-196a-1 and miR-196b Knockout Zebrafish Transmission electron microscopy was performed on the dorsal muscle tissue of zebrafish, and the experimental results are shown in Figure 4. The longitudinal muscle cells of the dorsal muscle tissue of wild-type zebrafish (Figure 4A) appeared normal in morphology, with a regular arrangement of intracellular myofibrils, normal sarcomere morphology, normal sarcoplasmic reticulum structure, and clear inner ridges of mitochondria. miR-196a-1 gene knockout zebrafish (Figure 4B) showed irregular myocyte morphology, irregular arrangement of intracellular myofibrils, partial myofilament breakage or focal lysis, sarcoplasmic reticulum expansion, mitochondrial swelling, partial inner ridge breakage or outer membrane rupture, and widened myocyte space. The dorsal muscle tissue of miR-196b gene knockout zebrafish (Figure 4C) was longitudinally cut, and it was found that myocytes were irregular in shape, intracellular myofibrils were arranged irregularly, some myofilaments were broken or focally lysed, the sarcoplasmic reticulum was expanded, mitochondria were swollen, some inner ridges were broken or the outer membrane was ruptured, and the myocyte space was widened. In zebrafish with miR-196a-1 and miR-196b gene knockout (Figure 4D), myocytes were irregular in shape, with myofibrils arranged irregularly, Z-lines mostly blurred and disordered, sarcomere swelling, mitochondrial pyknosis, and slightly dilated sarcoplasmic reticulum. The experimental results showed that after miR-196a-1 or miR-196b gene knockout, myofibrils became irregular, and the cell structure was abnormal; that is, muscle tissue was damaged. ## 3.4. Bone CT Detection Results of the miR-196a-1 and miR-196b Knockout Zebrafish Models Zebrafish with miR-196a-1 and miR-196b gene knockout were collected, and three in each group were tested by micro-CT. The results are shown in Figure 5, and there were no obvious defects in skeletal growth or development in the bones of zebrafish in the experimental groups. The bone mineral density (BMD) and trabecular number (TB.N) of zebrafish were further analyzed. The head, middle, and tail spines were measured for each fish, and the data obtained are shown in Figure 6. The BMD of the spine in different parts of the zebrafish was compared (Figure 6A). The BMDs of the head and middle spine of the miR-196a-1 gene knockout group, miR-196b gene knockout group, and miR-196a-1 and miR-196b double knockout group were significantly decreased compared with those of the wild-type zebrafish. The difference between the two groups was statistically significant ($$n = 3$$, * $p \leq 0.05$). In the caudal vertebra test results, although the bone mineral density decreased in the gene knockout group, there was no significant difference between the groups. These results suggest that miR-196a-1 or miR-196B gene knockdown reduces spinal BMD or BMD. In a comparison of the number of trabecular bones in different groups (Figure 6B), it was found that the number of trabecular bones decreased after miR-196a-1 or miR-196b gene knockout. Compared with that of the control group, the number of trabeculae in the middle and tail spine of the miR-196a-1 gene knockout group, miR-196b gene knockout group, and miR-196a-1 and miR-196b double knockout group was significantly decreased, and the difference between the groups was statistically significant ($$n = 3$$, * $p \leq 0.05$). In the postcranial vertebra, the number of trabecular bones decreased in the miR-196a-1 gene knockout group, but there was no significant difference between the groups compared with the control group. Compared with that of the single gene knockout group, the number of trabecular bones in the miR-196a-1 and miR-196b double knockout groups was further decreased, but only the middle and tail spine trabecular bone numbers between the miR-196b knockout group and the double knockout group were found to be statistically significant ($$n = 3$$, * $p \leq 0.05$), and there was no statistically significant difference between the other groups. The results suggested that miR-196a-1 or miR-196b gene knockdown reduced the number of spinal trabeculae. In conclusion, miR-196a-1 or miR-196b gene knockout can reduce spine BMD or bone mineral density and trabecular bone number in zebrafish, suggesting that the effect of miR-196a-1 or miR-196b deletion on the zebrafish motor system may be related to the reduction in bone BMD or trabecular bone number. ## 3.5. Real-Time PCR Analysis of Skeletal Muscle and Skeletal-Related Gene Expression Levels In this study, we screened vcp, dpm1, acta1b, and mylpfb genes, which are closely related to the function of skeletal muscle, by using biological information analysis and carried out real-time PCR analysis of zebrafish dorsal muscle tissue. The experimental results are shown in Figure 7. In addition, real-time PCR was used to detect related genes in the bone group, including col1a1a, bmp8a, gdf6a, and fgfr3. The experimental results are shown in Figure 8. Real-time PCR showed that the expression level of the vcp gene in the muscle tissue of miR-196a-1 gene knockout zebrafish was significantly lower than that in the control group ($$n = 3$$, * $p \leq 0.05$). The expression level of the vcp gene in the miR-196b knockout group did not change significantly. In the double-knockout group, the expression level of vcp was significantly decreased. Compared with the miR-196a-1 knockout group, there was no significant difference between the groups. Compared with the miR-196b knockout group, the difference between the groups was statistically significant ($$n = 3$$, & $p \leq 0.05$). The expression level of the dpm1 gene in the miR-196a-1 gene knockout group, miR-196b knockout group, and double knockout group was significantly reduced, and the difference between the groups was statistically significant ($$n = 3$$, * $p \leq 0.05$). There was no significant difference in the dpm1 gene between the miR-196a-1 or miR-196b knockout groups; that is, there was no significant difference between the double knockout group and the single knockout group. Compared with the control group, the expression levels of the acta1b and mylpfb genes were not significantly different in the miR-196a-1 knockout group. In the miR-196b knockout group, the relative expression levels of the acta1b and mylpfb genes were significantly reduced, and the difference between the groups was statistically significant ($$n = 3$$, * $p \leq 0.05$). In the double knockout group, the expression levels of these genes were also downregulated, and there was no significant difference between the two groups compared with the miR-196b knockout group. Compared with the miR-196a-1 knockout group, the difference between the two groups was statistically significant ($$n = 3$$, # $p \leq 0.05$). The relative expression level of bone genes was detected. The expression levels of the gdf6a and col1a1a genes in the miR-196a-1 knockout group, miR-196b knockout group, and double knockout group were significantly lower than those in the control group ($$n = 3$$, * $p \leq 0.05$). The relative expression levels of gdf6a and abcc6a in the double knockout group were lower than those in the miR-196b gene knockout group, and the difference between the two groups was statistically significant ($$n = 3$$, & $p \leq 0.05$). Real-time PCR showed that there was no significant difference in the expression of the bmp8a gene in the miR-196a-1 knockout group, the miR-196b knockout group, and the double knockout group; that is, there was no significant difference in the expression of the bmp8a gene compared with the control group. The expression level of the fgfr3 gene in miR-196a-1 knockout zebrafish was significantly decreased, the expression level in the miR-196b knockout group was significantly increased, and the difference between the two groups was statistically significant ($$n = 3$$, * $p \leq 0.05$). The expression level of the fgfr3 gene in the double knockout group was higher than that in the control group, and the difference between the two groups was statistically significant ($$n = 3$$, # $p \leq 0.05$). ## 4. Discussion Muscle is under the command and control of the nervous system; the force produced by skeletal muscle contraction results in its attached bone movement along with lever movement produced by the joint, resulting in various actions. Bones can support body shape, protect internal organs, maintain body posture, and support weight [26,27]. Skeletal muscle is widely distributed in the body. It plays a contractile function through the reflex of the nervous system and participates in the energy metabolism of the body to reserve energy to meet the needs of exercise [28,29]. Bone and skeletal muscle are closely related and influence each other [30,31]. Skeletal muscle is an important link between bone and bone, provides a microenvironment for bone, and is important in regulating bone latent cells, biomechanics, signaling factors, etc. [ 32,33]. The skeletal system also plays a role in regulating muscle; for example, paracrine or endocrine factors of osteocytes and osteoblasts can regulate muscle development and muscle strength [34,35]. These two factors complement each other and cooperate to complete the motor function of the body [2]. Dysfunction of bone and muscle leads to a variety of motor system diseases [36,37]. miR-196a plays an important regulatory role in organisms and has been found to be upregulated in a variety of malignant tumors. miR-196a affects the proliferation, invasion, and apoptosis of tumor cells and is expected to become a new molecular marker or target for tumor diagnosis, treatment, and prognosis, indicating its potential in clinical applications [38,39,40,41]. Several studies have also reported that miR-196b is highly expressed in multiple malignant tumors and has the characteristics of an oncogene to promote tumorigenesis [42,43,44]. At present, the relevant studies mainly focus on cancer or tumors, and the roles of miR-196a-1 and miR-196b in bone and skeletal muscle and even in the motor system are still unclear. Studies have found that miR-196a plays an important regulatory role in osteosarcoma, osteoporosis, bone mineral density, and osteoblast differentiation and proliferation [18,20,45]. The relevant reports rarely involve the study of miR-196a, bone and skeletal muscle, or exercise. It was found that pri-miR-196a1 was expressed in the tail and trunk of 24 h zebrafish embryos [21]. Combined with the existing research reports, the results showed that miR-196a plays an important role in bone and skeletal muscle. miR-196b is rarely reported in bone and muscle. Exploring the role of miR-196a and miR-196b in bone or muscle may help elucidate the value of the miR-196 family in the motor system. Zebrafish are a commonly used animal model for studying bone and muscle. Their bone development has a high degree of similarity to the development process of other vertebrates; their size is small, their feeding cost is low, their reproductive cycle is short, and their reproductive ability is strong [46,47], making them one of the best choices for gene knockout animal models. In this study, miR-196a-1 and miR-196b gene knockout lines were employed. The gene knockout zebrafish lines had no obvious defects in growth and development, indicating that the gene knockout of miR-196a-1 and miR-196b had no significant effect on the overall development of zebrafish and that these embryos could normally develop to maturity. However, whether loss of miR-196a-1 and miR-196b has effects on zebrafish bone and muscle needs to be further studied. Therefore, we detected the effects of miR-196a-1 and miR-196b on zebrafish motor behavior, as well as the effects on muscle and bone microstructure in the motor system. It was found that the total movement distance, movement duration, speed, and movement times of the zebrafish with miR-196a-1 or miR-196b gene knockout were decreased, especially the related indicators in the double knockout zebrafish, which were further decreased, but there was no significant difference between the miR-196a-1 gene knockout and miR-196b gene knockout groups. The experimental results suggested that the deletion of miR-196a-1 and miR-196b had an impact on motor function, so we further tested the effect of miR-196a-1 and miR-196b gene knockout on the motor system of zebrafish, including muscle and bone. HE and transmission electron microscopy (SEM) tests of zebrafish dorsal muscle tissue showed that muscle fiber organization of the miR-196a-1 or miR-196b gene knockout zebrafish is irregular, with sarcoplasmic reticulum expansion, mitochondrial swelling, partial spinal fracture or outer membrane rupture, and broadening muscle cell gap observed; namely, the muscle tissue structure was destroyed, and muscle fibers in the double knockout group structure were irregular. These results suggest that the deletion of miR-196a-1 or miR-196b has a certain effect on the morphology and structure of muscle tissue. The bone density of zebrafish spinal bones with CT detection or bone mineral density and bone trabecular number were assessed, and in the knockout group fish, bone density and bone mineral density were reduced, the trabecular bone volume was reduced, no obvious difference between the different knockout groups was found, and the lack of miR-196a-1 or miR-196b influenced the bone density and bone trabecular number. In conclusion, after miR-196a-1 or miR-196b gene knockout, muscle tissue was damaged, bone mineral density was decreased, and the amount of trabecular bone was decreased, which confirmed the effect of gene deletion on motor behavior. We speculated that the miR-196a-1 and miR-196b genes play a role in the motor system by affecting muscle fiber structure, bone mineral density, and the number of trabecular bone. Valosin-containing protein (vcp) has ATP hydrolysis activity and polyubiquitin modification-dependent protein binding activity and is expressed in various tissues, including muscle and brain [48,49]. Some studies have shown that vcp is the central mediator of lysosomal clearance and biogenesis in skeletal muscle [50]. Philipp [51] and other researchers showed that deletion of vcp in zebrafish damages protein homeostasis and leads to structural and functional defects in striated muscle in vivo. Real-time PCR showed that the expression level of the vcp gene in miR-196a-1 knockout zebrafish was significantly decreased, but there was no significant difference in the miR-196b knockout group. Dolichyl-phosphate mannose-transferase subunit 1 (dpm1) plays an important role in muscle development and is expected to be active in the endoplasmic reticulum. A study in zebrafish showed that dpm1 plays an important role in stabilizing muscle structure and regulating apoptotic cells [52]. The expression of the dpm1 gene was downregulated in the knockout group. It is speculated that the stability of muscle structure will be affected after the downregulation of expression, resulting in uneven thickness of muscle fibers in the knockout group. Actin alpha 1 skeletal muscle b (acta1b) plays a role upstream or downstream of embryonic cardiac tube development and skeletal muscle fiber development and is expressed in the cardiovascular system and myoblasts [53]. Myosin light chain can phosphorylate fast skeletal muscle b (mylpfb), which is expressed in muscle tissue and plays an important role in regulating muscle growth, maintaining muscle structure and function and muscle tissue homeostasis [54]. Real-time PCR showed that the expression levels of the acta1b and mylpfb genes were significantly decreased in the miR-196b knockout group and the double knockout group, but there was no significant difference in the miR-196a-1 knockout group, suggesting that after miR-196b gene knockout, the morphological structure and function of muscle might be affected by the acta1b and mylpfb genes. The differential expression of these genes may be related to the structural and functional changes in muscle. The bone extracellular matrix plays an important role in the dynamic action of osteoblasts and osteoclasts to regulate the process of bone regeneration. Col1a1 is an important part of the bone matrix [55]. Changes in the composition of the bone extracellular matrix can destroy ECM-bone cell signal transduction, resulting in changes in bone mineral density and/or bone microstructure. The expression level of col1a1a was significantly downregulated after miR-196a-1 or miR-196b gene knockout. In combination with previous experimental results, it was found that the bone mineral density and the number of bone trabeculae were decreased in zebrafish after miR-196a-1 and miR-196b gene deletion. It is speculated that the deletion of the miR-196a-1 and miR-196b genes may affect the expression of the collagen gene or affect bone mineral density and the number of bone trabeculae through signal transduction, thus affecting the function of zebrafish bone. Bone morphogenetic protein 8a (bmp8a) is a member of the bone morphogenetic protein family, which is a classical multifunctional growth factor and belongs to the transforming growth factor β superfamily, and bmp8a plays an important regulatory role in bone, muscle, blood vessels, and other tissues [56]. In this study, it was found that there was no significant difference in the expression of the bmp8a gene in the miR-196a-1 or miR-196b gene knockout group. It is speculated that the effect of miR-196a-1/b on bone may be unrelated to bmp8a. Growth differentiation factor 6a (gdf6a) is expressed in a variety of tissues and structures. It is a ligand of the bone morphogenetic protein family that is expressed in the notochord and primitive intestinal endoderm of zebrafish. The human homologous gene of this gene is related to a variety of diseases, including multiple joint fracture syndrome [57]. The expression level of the gdf6a gene was significantly downregulated in the miR-196a-1 or miR-196b gene knockout group and was further downregulated in the double knockout group. Fibroblast growth factor receptor 3 (fgfr3) is expressed in the chondrocytes of zebrafish head cartilage, osteoblasts involved in bone formation, and other cells [58]. In this study, we found that the expression level of fgfr3 was downregulated in the miR-196a-1 gene knockout group and upregulated in the miR-196b gene knockout group. In the double knockout group, the relative expression level of the fgfr3 gene was higher than that in the miR-196a-1 knockout group and lower than that in the miR-196b knockout group. In summary, miR-196a-1 or miR-196b knockout had a certain impact on the bone mineral density and trabecular quantity of zebrafish bone tissue, and there was no significant difference in the changes in bone microstructure between the double knockout group and the single knockout group. It is speculated that in the miR-196a-1 or miR-196b knockout group, these changes in bone microstructure may be related to the changes in the expression levels of gdf6a, fgfr3, and col1a1a. Our results show that the effects of miR-196a-1 and miR-196b on bone and muscle may be related to the expression levels of vcp, dpm1, acta1b, mylpfb, gdf6a, fgfr3, and col1a1a. We expect to study the key factors associated with miR-196a-1 and miR-196b in muscle and bone to further elucidate the mechanism of miR-196a-1 and miR-196b in organisms. ## 5. Conclusions The motor function of zebrafish decreased after miR-196a-1 or miR-196b gene knockout. Microstructure analysis showed that zebrafish with gene knockout had abnormal muscle fiber structure, reduced bone mineral density, and reduced trabecular bone data. We hypothesized that miR-196a-1 or miR-196b causes motor function decline by affecting muscle and bone tissue, and that these effects may be related to the expression of the vcp, dpm1, acta1b, mylpfb, gdf6a, fgfr3, and col1a1a genes. These results suggest that miR-196a-1 and miR-196b play roles in muscle fiber structure, bone mineral density, and bone trabecular quantity by affecting the expression of vcp, dpm1, acta1b, mylpfb, gdf6a, fgfr3, and col1a1a, and then affect the function of the motor system. ## References 1. Ireland A., Ferretti J.L., Rittweger J.. Imaging of the muscle-bone relationship. Curr. Osteoporos. Rep. (2014) 12 486-495. DOI: 10.1007/s11914-014-0216-1 2. Tagliaferri C., Wittrant Y., Davicco M.J., Walrand S., Coxam V.. Muscle and bone, two interconnected tissues. Ageing Res. Rev. (2015) 21 55-70. DOI: 10.1016/j.arr.2015.03.002 3. Colaianni G., Mongelli T., Colucci S., Cinti S., Grano M.. Crosstalk Between Muscle and Bone Via the Muscle-Myokine Irisin. Curr. Osteoporos. Rep. (2016) 14 132-137. DOI: 10.1007/s11914-016-0313-4 4. Wood C.L., Straub V.. Bones and muscular dystrophies: What do we know?. Curr. Opin. Neurol. (2018) 31 583-591. DOI: 10.1097/WCO.0000000000000603 5. Iolascon G., Paoletta M., Liguori S., Curci C., Moretti A.. Neuromuscular Diseases and Bone. Front. Endocrinol. (2019) 10 794. DOI: 10.3389/fendo.2019.00794 6. Torres-Costoso A., Lopez-Munoz P., Martinez-Vizcaino V., Alvarez-Bueno C., Cavero-Redondo I.. Association Between Muscular Strength and Bone Health from Children to Young Adults: A Systematic Review and Meta-analysis. Sport. Med. (2020) 50 1163-1190. DOI: 10.1007/s40279-020-01267-y 7. Lin Y.Y.. Muscle diseases in the zebrafish. Neuromuscul. Disord. (2012) 22 673-684. DOI: 10.1016/j.nmd.2012.04.007 8. Vishnoi A., Rani S.. miRNA Biogenesis and Regulation of Diseases: An Overview. Methods in Molecular Biology (2017) 1509 1-10. PMID: 27826912 9. Wang J., Yang L.Z., Zhang J.S., Gong J.X., Wang Y.H., Zhang C.L., Chen H., Fang X.T.. Effects of microRNAs on skeletal muscle development. Gene (2018) 668 107-113. DOI: 10.1016/j.gene.2018.05.039 10. Hill M., Tran N.. miRNA interplay: Mechanisms and consequences in cancer. Dis. Model. Mech. (2021) 14 dmm047662. DOI: 10.1242/dmm.047662 11. Sun Z., Wang H., Wang Y., Yuan G., Yu X., Jiang H., Wu Q., Yang B., Hu Z., Shi F.. MiR-103-3p targets the m(6) A methyltransferase METTL14 to inhibit osteoblastic bone formation. Aging Cell (2021) 20 e13298. DOI: 10.1111/acel.13298 12. Karabegovic I., Maas S., Medina-Gomez C., Zrimsek M., Reppe S., Gautvik K.M., Uitterlinden A.G., Rivadeneira F., Ghanbari M.. Genetic Polymorphism of miR-196a-2 is Associated with Bone Mineral Density (BMD). Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18122529 13. Zhong L.N., Zhang Y.Z., Li H., Fu H.L., Lv C.X., Jia X.J.. Overexpressed miR-196a accelerates osteogenic differentiation in osteoporotic mice via GNAS-dependent Hedgehog signaling pathway. J. Cell. Biochem. (2019) 120 19422-19431. DOI: 10.1002/jcb.29166 14. Chen C., Zhang Y., Zhang L., Weakley S.M., Yao Q.. MicroRNA-196: Critical roles and clinical applications in development and cancer. J. Cell. Mol. Med. (2011) 15 14-23. DOI: 10.1111/j.1582-4934.2010.01219.x 15. Bastami M., Masotti A., Saadatian Z., Daraei A., Farjam M., Ghanbariasad A., Vahed S.Z., Eyvazi S., Mansoori Y., Nariman-Saleh-Fam Z.. Critical roles of microRNA-196 in normal physiology and non-malignant diseases: Diagnostic and therapeutic implications. Exp. Mol. Pathol. (2021) 122 104664. DOI: 10.1016/j.yexmp.2021.104664 16. Wang X., Zhang L., Zhang X., Xing C., Liu R., Zhang F.. MiR-196a promoted cell migration, invasion and the epithelial-mesenchymal transition by targeting HOXA5 in osteosarcoma. Cancer Biomark. (2020) 29 291-298. DOI: 10.3233/CBM-201674 17. Shang Y., Wang L.Q., Guo Q.Y., Shi T.L.. MicroRNA-196a overexpression promotes cell proliferation and inhibits cell apoptosis through PTEN/Akt/FOXO1 pathway. Int. J. Clin. Exp. Pathol. (2015) 8 2461-2472. PMID: 26045752 18. Takafuji Y., Tatsumi K., Kawao N., Okada K., Muratani M., Kaji H.. MicroRNA-196a-5p in Extracellular Vesicles Secreted from Myoblasts Suppresses Osteoclast-like Cell Formation in Mouse Cells. Calcif. Tissue Int. (2021) 108 364-376. DOI: 10.1007/s00223-020-00772-6 19. Namlos H.M., Meza-Zepeda L.A., Baroy T., Ostensen I.H., Kresse S.H., Kuijjer M.L., Serra M., Burger H., Cleton-Jansen A.M., Myklebost O.. Modulation of the osteosarcoma expression phenotype by microRNAs. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0048086 20. Qin Y., Wang L., Gao Z., Chen G., Zhang C.. Bone marrow stromal/stem cell-derived extracellular vesicles regulate osteoblast activity and differentiation in vitro and promote bone regeneration in vivo. Sci. Rep. (2016) 6 21961. DOI: 10.1038/srep21961 21. He X., Yan Y.L., Eberhart J.K., Herpin A., Wagner T.U., Schartl M., Postlethwait J.H.. miR-196 regulates axial patterning and pectoral appendage initiation. Dev. Biol. (2011) 357 463-477. DOI: 10.1016/j.ydbio.2011.07.014 22. Daya A., Donaka R., Karasik D.. Zebrafish models of sarcopenia. Dis. Model. Mech. (2020) 13 dmm042689. DOI: 10.1242/dmm.042689 23. Gupta V., Kawahara G., Gundry S.R., Chen A.T., Lencer W.I., Zhou Y., Zon L.I., Kunkel L.M., Beggs A.H.. The zebrafish dag1 mutant: A novel genetic model for dystroglycanopathies. Hum. Mol. Genet. (2011) 20 1712-1725. DOI: 10.1093/hmg/ddr047 24. Ting Z., Guifang F., Bingling X., Han D., Huaping X., Yulong Y.. Construction of a gene knockout strain of miR-196a-1 in Zebrafish. Laser Biol. (2020) 29 176-182 25. Ulhaq M., Orn S., Carlsson G., Morrison D.A., Norrgren L.. Locomotor behavior in zebrafish (. Aquat. Toxicol. (2013) 144–145 332-340. DOI: 10.1016/j.aquatox.2013.10.021 26. Ferretti J.L., Cointry G.R., Capozza R.F., Frost H.M.. Bone mass, bone strength, muscle-bone interactions, osteopenias and osteoporoses. Mech. Ageing Dev. (2003) 124 269-279. DOI: 10.1016/S0047-6374(02)00194-X 27. Grabowski P.. Physiology of bone. Endocr. Dev. (2009) 16 32-48. PMID: 19494659 28. Brooks S.V.. Current topics for teaching skeletal muscle physiology. Adv. Physiol. Educ. (2003) 27 171-182. DOI: 10.1152/advan.2003.27.4.171 29. Florencio-Silva R., Sasso G.R., Sasso-Cerri E., Simoes M.J., Cerri P.S.. Biology of Bone Tissue: Structure, Function, and Factors That Influence Bone Cells. BioMed Res. Int. (2015) 2015 421746. DOI: 10.1155/2015/421746 30. Cianferotti L., Brandi M.L.. Muscle-bone interactions: Basic and clinical aspects. Endocrine (2014) 45 165-177. DOI: 10.1007/s12020-013-0026-8 31. Battafarano G., Rossi M., Marampon F., Minisola S., Del Fattore A.. Bone Control of Muscle Function. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21041178 32. Pedersen B.K., Febbraio M.A.. Muscles, exercise and obesity: Skeletal muscle as a secretory organ. Nat. Rev. Endocrinol. (2012) 8 457-465. DOI: 10.1038/nrendo.2012.49 33. Julien A., Kanagalingam A., Martinez-Sarra E., Megret J., Luka M., Menager M., Relaix F., Colnot C.. Direct contribution of skeletal muscle mesenchymal progenitors to bone repair. Nat. Commun. (2021) 12 2860. DOI: 10.1038/s41467-021-22842-5 34. Laurent M.R., Dubois V., Claessens F., Verschueren S.M., Vanderschueren D., Gielen E., Jardi F.. Muscle-bone interactions: From experimental models to the clinic? A critical update. Mol. Cell. Endocrinol. (2016) 432 14-36. DOI: 10.1016/j.mce.2015.10.017 35. Sims N.A.. Cell-specific paracrine actions of IL-6 family cytokines from bone, marrow and muscle that control bone formation and resorption. Int. J. Biochem. Cell Biol. (2016) 79 14-23. DOI: 10.1016/j.biocel.2016.08.003 36. Novotny S.A., Warren G.L., Hamrick M.W.. Aging and the muscle-bone relationship. Physiology (2015) 30 8-16. DOI: 10.1152/physiol.00033.2014 37. Bonewald L.F.. The Role of the Osteocyte in Bone and Nonbone Disease. Endocrinol. Metab. Clin. N. Am. (2017) 46 1-18. DOI: 10.1016/j.ecl.2016.09.003 38. Liu M., Du Y., Gao J., Liu J., Kong X., Gong Y., Li Z., Wu H., Chen H.. Aberrant expression miR-196a is associated with abnormal apoptosis, invasion, and proliferation of pancreatic cancer cells. Pancreas (2013) 42 1169-1181. DOI: 10.1097/MPA.0b013e3182962acb 39. Qiu Y., Han Q., Lu H., Shi C.. miR-196a targeting LRIG3 promotes the proliferation and migration of cervical cancer cells. Cell. Mol. Biol. (2020) 66 180-185. DOI: 10.14715/cmb/2020.66.7.27 40. Zhan B., Huang L., Chen Y., Ye W., Li J., Chen J., Yang S., Jiang W.. miR-196a-mediated downregulation of p27(kip1) protein promotes prostate cancer proliferation and relates to biochemical recurrence after radical prostatectomy. Prostate (2020) 80 1024-1037. DOI: 10.1002/pros.24036 41. Fan Y., Fei M., Li Y., Gao Z., Zhu Y., Dai G., Wu D.. miR-196a-2 Promotes Malignant Progression of Thyroid Carcinoma by Targeting NRXN1. Comput. Math. Methods Med. (2021) 2021 4856820. DOI: 10.1155/2021/4856820 42. Ma R., Zheng G., Shao C., Liu J., Lv H., Zhang G.. Downregulation of miR-196b Promotes Glioma Cell Sensitivity to Temozolomide Chemotherapy and Radiotherapy. Ann. Clin. Lab. Sci. (2018) 48 719-725. PMID: 30610041 43. Han L., Yao Z., Xie L., Li D., Wang C., Yang Y., Yang J., Huang Z., Li K., Zhang Y.. Transcriptome Sequencing reveals the expressed profiles of mRNA and ncRNAs and regulate network via ceRNA mediated molecular mechanism of lung adenocarcinoma bone metastasis in Xuanwei. Transl. Cancer Res. (2021) 10 73-87. DOI: 10.21037/tcr-20-2376 44. Wan T., Sun H., Mao Z., Zhang L., Chen X., Shi Y., Shang Y.. Vitamin D deficiency inhibits microRNA-196b-5p which regulates ovarian granulosa cell hormone synthesis, proliferation, and apoptosis by targeting RDX and LRRC17. Ann. Transl. Med. (2021) 9 1775. DOI: 10.21037/atm-21-6081 45. Kim Y.J., Bae S.W., Yu S.S., Bae Y.C., Jung J.S.. miR-196a regulates proliferation and osteogenic differentiation in mesenchymal stem cells derived from human adipose tissue. J. Bone Miner. Res. (2009) 24 816-825. DOI: 10.1359/jbmr.081230 46. Mushtaq M.Y., Verpoorte R., Kim H.K.. Zebrafish as a model for systems biology. Biotechnol. Genet. Eng. Rev. (2013) 29 187-205. DOI: 10.1080/02648725.2013.801238 47. Busse B., Galloway J.L., Gray R.S., Harris M.P., Kwon R.Y.. Zebrafish: An Emerging Model for Orthopedic Research. J. Orthop. Res. (2020) 38 925-936. DOI: 10.1002/jor.24539 48. Kustermann M., Manta L., Paone C., Kustermann J., Lausser L., Wiesner C., Eichinger L., Clemen C.S., Schroder R., Kestler H.A.. Loss of the novel Vcp (valosin containing protein) interactor Washc4 interferes with autophagy-mediated proteostasis in striated muscle and leads to myopathy in vivo. Autophagy (2018) 14 1911-1927. DOI: 10.1080/15548627.2018.1491491 49. Imamura S., Ojima N., Yamashita M.. Cold-inducible expression of the cell division cycle gene CDC48 and its promotion of cell proliferation during cold acclimation in zebrafish cells. FEBS Lett. (2003) 549 14-20. DOI: 10.1016/S0014-5793(03)00723-3 50. Arhzaouy K., Papadopoulos C., Schulze N., Pittman S.K., Meyer H., Weihl C.C.. VCP maintains lysosomal homeostasis and TFEB activity in differentiated skeletal muscle. Autophagy (2019) 15 1082-1099. DOI: 10.1080/15548627.2019.1569933 51. Voisard P., Diofano F., Glazier A.A., Rottbauer W., Just S.. CRISPR/Cas9-Mediated Constitutive Loss of VCP (Valosin-Containing Protein) Impairs Proteostasis and Leads to Defective Striated Muscle Structure and Function In Vivo. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23126722 52. Marchese M., Pappalardo A., Baldacci J., Verri T., Doccini S., Cassandrini D., Bruno C., Fiorillo C., Garcia-Gil M., Bertini E.. Dolichol-phosphate mannose synthase depletion in zebrafish leads to dystrophic muscle with hypoglycosylated alpha-dystroglycan. Biochem. Biophys. Res. Commun. (2016) 477 137-143. DOI: 10.1016/j.bbrc.2016.06.033 53. Sztal T.E., Zhao M., Williams C., Oorschot V., Parslow A.C., Giousoh A., Yuen M., Hall T.E., Costin A., Ramm G.. Zebrafish models for nemaline myopathy reveal a spectrum of nemaline bodies contributing to reduced muscle function. Acta Neuropathol. (2015) 130 389-406. DOI: 10.1007/s00401-015-1430-3 54. Dhanasiri A.K.S., Johny A., Xue X., Berge G.M., Bogevik A.S., Rise M.L., Faeste C.K., Fernandes J.M.O.. Plant-Based Diets Induce Transcriptomic Changes in Muscle of Zebrafish and Atlantic Salmon. Front. Genet. (2020) 11 575237. DOI: 10.3389/fgene.2020.575237 55. Michou L., Brown J.P.. Genetics of bone diseases: Paget’s disease, fibrous dysplasia, osteopetrosis, and osteogenesis imperfecta. Jt. Bone Spine (2011) 78 252-258. DOI: 10.1016/j.jbspin.2010.07.010 56. Ward-Caviness C.K., Neas L.M., Blach C., Haynes C.S., Larocque-Abramson K., Grass E., Dowdy E., Devlin R.B., Diaz-Sanchez D., Cascio W.E.. Genetic Variants in the Bone Morphogenic Protein Gene Family Modify the Association between Residential Exposure to Traffic and Peripheral Arterial Disease. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0152670 57. Crosier P.S., Kalev-Zylinska M.L., Hall C.J., Flores M.V., Horsfield J.A., Crosier K.E.. Pathways in blood and vessel development revealed through zebrafish genetics. Int. J. Dev. Biol. (2002) 46 493-502. PMID: 12141436 58. Ledwon J.K., Turin S.Y., Gosain A.K., Topczewska J.M.. The expression of fgfr3 in the zebrafish head. Gene Expr. Patterns (2018) 29 32-38. DOI: 10.1016/j.gep.2018.04.002
--- title: Impact of a Training Program on Oncology Nurses’ Confidence in the Provision of Self-Management Support and 5As Behavioral Counseling Skills authors: - Doris Howell - Patrick McGowan - Denise Bryant-Lukosius - Ryan Kirkby - Melanie Powis - Diana Sherifali - Vishal Kukreti - Sara Rask - Monica K. Krzyzanowska journal: Cancers year: 2023 pmcid: PMC10046590 doi: 10.3390/cancers15061811 license: CC BY 4.0 --- # Impact of a Training Program on Oncology Nurses’ Confidence in the Provision of Self-Management Support and 5As Behavioral Counseling Skills ## Abstract ### Simple Summary Cancer patients and their families require support to effectively self-manage the medical, emotional, and lifestyle consequences of cancer. In this paper, we describe a training program that showed improvement in oncology nurses confidence in the microskills required for the provision of self-management support to patients for application in routine care and in roles as cancer coaches before and after training at three cancer centres in Ontario, Canada. Self-management support is lacking in cancer care and little attention has been focused on the required preparation of nurses to provide self-management support and behavior change counseling. Our training program may have potential for improving nurses’ provision of self-management support, however, further testing in a larger population of nurses is required to assess effects on nurses skills and its impact on patient uptake of self-management behaviors and health outcomes. ### Abstract Background: Cancer patients and their families play a central role in the self-management of the medical, emotional, and lifestyle consequences of cancer. Nurses with training in self-management support can enable cancer patients to better manage the effects of cancer and treatment. Methods: As part of a randomized controlled trial, we developed a training program to build nurses’ confidence in the provision of self-management support (SMS). The SMS skills taught were adapted from the Stanford Peer Support training programs and embedded within the 5As (Assess, Advise, Agree, Assist, and Arrange) behavioral counseling process. We evaluated the impact of the training program on oncology nurses’ and coaches’ confidence using a Student’s t-test for paired samples in a nonrandomized, one-group pre/postsurvey. Results: Participants were experienced oncology nurses from three participating cancer centers. A two-tailed Student’s t-test for paired samples showed a significant improvement in nurses’ confidence for the 15 SMS microskills targeted in the training between the pretest and post-test as follows: for Center 1, a mean difference of 0.79 ($t = 7.18$, p ≤ 0.00001); for Center 2, a mean difference of 0.73 ($t = 8.4$, p ≤ 0.00001); for Center 3, a mean difference of 1.57 ($t = 11.45$, p ≤ 0.00001); and for coaches, a mean difference of 0.52 ($t = 7.6$, p ≤ 0.00001). Conclusions: Our training program improved oncology staff nurses’ and cancer coaches’ confidence in 15 SMS microskills and has potential for SMS training of nurses in routine care. ## 1. Background Globally, there is increasing recognition of the central role of cancer patients and caregivers in achieving better health outcomes through effective self-management of the acute and chronic effects of cancer [1,2]. Self-management involves the application of a set of cognitive and behavioral skills to manage the medical aspects of cancer, including physical symptoms and treatment effects, psychosocial consequences, role and lifestyle changes [3,4,5], and other tasks for living with cancer [6]. Individuals with cancer require the knowledge, skills, and confidence to tailor their daily behaviors to manage dynamic disease and symptom fluctuations, reduce complications and late-effect risks, and optimize well-being and survival [7,8]. Cancer self-management can be daunting and may not be in the individuals’ usual repertoire of disease self-management skills or health behaviors [9], and they may require self-management support (SMS) to build their capacity and self-efficacy for effective disease self-management [10]. Healthcare providers who are skilled in the systematic provision of SMS and in coaching patients on the use of health behaviors and core skills such as goal setting, action planning, and problem-solving [11] can empower and enable patients in the self-management of cancer and health [12]. Oncology nurses, as one of the largest service provider groups in cancer care, can play pivotal roles in the provision of SMS in routine care and as cancer coaches [13,14]. Systematic reviews have shown that SMS, delivered by nurses educated and skilled in facilitating patient activation in self-management, can result in positive behavioral change and better clinical outcomes in typical chronic conditions (e.g., reduced blood pressure in hypertension and lower A1c in diabetes) [15,16]. SMS also improves physical symptoms (e.g., fatigue), anxiety, and quality of life in cancer populations [17,18,19]. However, oncology nurses require specific knowledge and skills in the provision of SMS and coaching for behavior change and positive attitudes toward collaborative engagement of patients as partners in care [20]. SMS involves more than just condition-related education: it also involves coaching of patients in the cognitive and behavioral application of self-management behaviors to address specific problems (e.g., cancer fatigue) and the core skills inclusive of goal setting and taking action, system navigation, implementing problem-solving strategies, communicating with healthcare professionals, self-monitoring, decision making, resource utilization and self-tailoring of health behaviors to optimize health during and after cancer treatment [21]. The purpose of this article is to describe the development and impact of a SMS training program on oncology nurses’ confidence in the 15 microskills necessary for delivering effective SMS in routine care and acting as cancer coaches. The training program was developed in the context of a randomized controlled trial entitled the Self-Management and Activation to Reduce Treatment Toxicities (SMARTCare) that targeted patient management of acute treatment toxicities (NCT03849950). Ethics approval for the study was obtained from the Ontario Clinical Oncology Group, Hamilton, ON. Canada. ## 2. Materials and Methods Briefly, the SMARTCare intervention comprised two components: an online self-management education program for patients (I-Can Manage.ca) and telephone-based, nurse-delivered cancer coaching targeting patients diagnosed with lung cancer, colorectal cancer, and lymphoma, who were scheduled to receive first-line or metastatic parenteral or oral treatment. Study participants were randomized to receive the online program and 5 sessions of cancer coaching (approximately 45 min per session) timed for completion across the treatment trajectory or a usual care control condition (Figure 1). Oncology clinic nurses were trained in practical skills of SMS (e.g., teach back) that could be applied in routine practice [22]. Oncology clinic nurses and two nurses from each disease site were purposively selected as cancer coach intervention nurses from ambulatory clinics across 3 regional cancer centers in Ontario, Canada, to participate in the training program. The results of the trial suggest promising results for nurse-delivered SMS coaching for improving patient activation [23]. We adapted an existing telephone-based, peer-led SMS program based on the Stanford approach with demonstrated effectiveness in diabetes [24] to cancer populations. The training was provided by experts in SMS and coaching (PM and DH). All oncology clinic nurses received 4 h of training in the fundamentals of SMS that targeted the application of specific strategies to support patients’ application of core self-management skills for the management of cancer treatment toxicities and health promotion (smoking cessation, physical activity, and healthy eating). SMS techniques and microskills taught in the training program were embedded within the teaching of the 5As (Assess, Advise, Agree, Assist, and Arrange) process for counseling patients in the use of self-management behaviors (Figure 2) [25]. The training emphasized the fundamental microskills of establishing rapport, setting a shared agenda, assessing readiness using rulers (importance and confidence), ask–tell–ask, closing the loop, and teach back for practical application by staff nurses and cancer coaches. This approach is consistent with recommendations of provincial nursing guidelines [26] and based on the Stanford approach for developing peer leaders’ core self-management support skills [24]. The 5As model is recommended for ensuring that healthcare professionals use a sequential set of actions to help individuals to become better self-managers [27]. We also demonstrated the application of the 5As for behavioral counseling of patients in the management of specific treatment toxicities (e.g., fatigue) (Figure 3) and for coaching self-management of treatment toxicities such as immunotherapy (Supplement S1 in Supplementary Materials). The training program was participatory and case-based, and it included skills practiced through modeling and role playing. Nurses seconded as cancer coaches to deliver self-management coaching in the intervention arm of the trial participated in the fundamentals of SMS (4 h), then received an additional 7 h of an enhanced training program that targeted deeper consolidation of skills in applying the 5As and advanced SMS skills, including goal setting and action planning, problem solving, decision making, and motivational interviewing OARS communication skills (open-ended questions, affirmations, reflections, and summarization). Cancer coaching was defined as “a person-centred, collaborative, strength-based care process that educates, engages, and motivates patients within a coaching context to enhance self-management capability and self-efficacy in applying problem-specific self-management strategies, core skills, and health behaviors to reduce the immediate and long-term physical and emotional consequences of cancer and cancer treatment based on health coaching models for other chronic conditions” [28]. The SMARTCare intervention and training program was guided by our cancer coaching model adapted from the diabetes peer support training program that shows the range of skills emphasized in the training [24] (Figure 4). This approach uses the patients’ frame of reference (i.e., their experiences, definitions of health and well-being, values, and preferences) as the starting point to identify health goals, and then provides navigation, education, support, practical guidance, and facilitation of uptake of behaviors and skills to help the client manage symptoms and achieve their health goals. Oncology nurses designated as coaches in the intervention arm were also guided by a manual to standardize the coaching process and support intervention fidelity. The manual consisted of the objectives, content, and processes to be covered in each of the five coaching sessions for the SMARTCare intervention. Recording forms were included for nurses to document each call. Additionally, instruction was provided for making monthly telephone calls and focusing the coaching conversation on four key areas: [1] how patients were managing their cancer and treatment toxicities, medications, home, and daily life; [2] using the problem-solving process; [3] making action plans to support a behavioral goal; and [4] how to access and navigate locate community resources. Cancer coaches also received mentorship through monthly community-of-practice meetings with experienced cancer coaches who were focused on case review and supporting skills consolidation. Cancer coach interactions with a client after the first 2 sessions were observed using a fidelity coaching checklist (Supplement S2 in Supplementary Materials), and coaches were provided with constructive feedback on their application of coaching skills and guidance for further consolidation. We developed a purpose-built, pre/postsurvey to evaluate nurses’ confidence using a Likert scale (from 0 (no confidence) to 5 (very confident)) in the application of the 5As and 15 related micro-skills (Supplement S3 in Supplementary Materials). Descriptive statistics were used to describe sample characteristics and mean scores across survey items and nurses’ satisfaction with the training program. We checked the data for normality using Shapiro–Wilk tests, and the normality assumption was met. Thus, a Student’s t-test for paired samples was used to examine mean differences in oncology nurses’ confidence in using the 15 micro-skills for SMS from the pre- to post-test survey for each of the three centers. A p-value of 0.05 was considered significant. ## 3. Results A total of 40 clinic nurses, who were mostly diploma prepared with >10 years cancer experience and certified in oncology, completed the fundamentals of SMS training and 8 nurses received additional training as cancer coaches (Table 1). Most nurses had limited pretraining awareness of the concepts of self-management, self-management support, and patient activation (Table 2). A statistically significant improvement in the 15 SMS microskills for nurses from the three centers participating in the training program and separately for coaches was observed as follows: for Center 1, an overall mean difference of 0.79 between pre- and postsurvey ($t = 7.18$, p ≤ 0.00001); for Center 2, a mean difference of 0.73 ($t = 8.4$, p ≤ 0.00001); for Center 3, a mean difference of 1.57 ($t = 11.45$, p ≤ 0.00001); and for coaches, a mean difference of 0.52 ($t = 7.6$, p ≤ 0.00001; Table 3). Overall, the training, facilitation, and methods for teaching (i.e., skills practice) were positively viewed, and the time to practice skills using cases and role playing was highly valued (Table 4). ## 4. Discussion We developed a SMS training program to be tested in a feasibility RCT, with preliminary effects shown for improvement in patient activation favoring the online self-management education program plus coaching compared with a usual care control group [23]. In this paper, we described our training program and showed positive benefits for improving nurses’ knowledge and self-confidence in the microskills necessary for the provision of self-management support and coaching within the 5As behavior counseling framework. While there are training programs for health coaches, this is one of the few programs that additionally targets training of oncology nurses in self-management support for application within their practices. SMS is essential to the achievement of a more person-centered, psycho-educational approach that can enable patients to live well with cancer as an acute and chronic disease [29]. Self-management education combined with nurse-led SMS is considered integral for enabling patients to effectively manage chronic conditions, including complex illnesses such as cancer [30,31,32]. However, few studies have addressed SMS training for nurses, and this preparation is lacking in basic and graduate nurse education programs [33]. Recently, a core curriculum was developed through international consensus that can be used to guide future undergraduate and graduate education programs [34]. However, less is known about how to build oncology nurses’ application of SMS in routine care, and a gap was identified in the integration of SMS in routine oncology care [35]. We recognize that skills consolidation requires more than a one-time training program. We did not evaluate the application of SMS in routine care by oncology staff nurses, and this should be examined in future research. We did assess the use of skills by cancer coaches and identified that more time and practice in skill consolidation will be essential in future SMS training and that online learning may facilitate the learning process [36]. The practical issues of training healthcare professionals and their integration of self-management skills within their practice will require consideration. There is an urgent need to understand how a system-wide approach to SMS can become embedded and integrated into routine practice and the essential training components and skill consolidation strategies that are critical for scale and spread. Training in the fundamental skills of SMS could be included as part of new staff orientation and ongoing professional development. However, this will require cancer organizations to set similar expectations for performance for SMS within practice and as a standard of quality care in cancer programs [37]. There is still much to understand regarding the critical elements and how best to build the self-efficacy and capacity of nurses to deliver effective SMS in routine care and as cancer coaches and to ensure their ongoing competence. Further creative multimedia training approaches, such as the use of standardized patients for skills consolidation and rigorous evaluation, are needed to measure the uptake of SMS in the practice repertoire of nurses and other healthcare professionals. Oncology nurses in ambulatory cancer clinics are ideally positioned to deliver SMS that could build cancer patients’ self-efficacy and capacity for self-management of treatment toxicities but will require flexibility in adopting this approach within traditional patient education programs and the time to ensure its delivery in rapid, episodic clinic visits. This study has limitations, including the self-selection of nurses as cancer coaches, the use of a purpose-built survey for program evaluation, and the one-group design without a control comparator. A power calculation was not performed given the feasibility nature of the work; thus, statistical significance should be interpreted in this context. It is also uncertain if SMS skills will be maintained after the trial without ongoing efforts toward embedding and normalizing SMS in routine care. ## 5. Conclusions As noted in the Chronic Disease Self-Management Framework, one of the main objectives of health services is to support self-management, which needs to be embedded within a system and includes knowledgeable and confident individuals, prepared clinicians, and a responsive and flexible service delivery structure [38]. This study demonstrated the feasibility of training oncology nurses to deliver SMS coaching. Further research is needed to evaluate the effectiveness of interventions for integrating and sustaining nurse-led SMS into routine care. ## References 1. Coolbrandt A., Wildiers H., Aertgeerts B., de Casterle B.D., Achterberg T.V., Millisen K.. Systematic development of CHEMO-SUPPORT, a nursing intervention to support adult patients with cancer in dealing with chemotherapy-related symptoms at home. BMC Nurs. (2018) 17. DOI: 10.1186/s12912-018-0297-8 2. Foster C., Wyatt D., Hubert-Williams N.. Survivorship and self-management in cancer care. Cancer and Cancer Care (2015) 3. Barlow J.H., Wright C., Sheasby J., Turner A., Hainsworth J.. Self-management approaches for people with chronic conditions: A review. Patient Educ. Counsel. (2002) 48 177-187. DOI: 10.1016/S0738-3991(02)00032-0 4. Lorig K.R., Holman H.. Self-management education: History, definition, outcomes, and mechanisms. Ann. Behav. Med. (2003) 26 1-7. DOI: 10.1207/S15324796ABM2601_01 5. Hammer M.J., Ercolano E.A., Wright F., Dickson V.V., Chyun D., Melkus G.D.. Self-management for Adult Patients with Cancer: An Integrative Review. Cancer Nurs. (2015) 38 E10-E26. DOI: 10.1097/NCC.0000000000000122 6. Howell D.. Supported self-management for cancer survivors to address long-term biopsychosocial consequences of cancer and treatment to optimize living well. Curr. Opin. Support. Palliat. Care (2018) 12 92-99. DOI: 10.1097/SPC.0000000000000329 7. Stein K.D., Syrjala K.L., Andrykowski M.A.. Physical and psychological long-term and late effects of cancer. Cancer (2008) 112 2577-2592. DOI: 10.1002/cncr.23448 8. Grimmett C., Corbett T., Brunet J., Shepherd J., Pinto B.M., May C.R., Foster C.. Systematic review and meta-analysis of maintenance of physical activity behaviour change in cancer survivors. Int. J. Behav. Nutr. Phys. Act. (2019) 16 37. DOI: 10.1186/s12966-019-0787-4 9. Haase K.R., Avery J., Bryant-Lukosius D., Kryzanowska M., Kukretti V., Liu G., Mayo S.J., Jones J., Howell D.. Patient and clinician perspectives of desired features for a web-based self-management program (icanmanage.ca): Exposing patients “hard work” of managing acute cancer. Support. Care Cancer (2020) 29 1989-1998. DOI: 10.1007/s00520-020-05683-8 10. Ose D., Winkler E.C., Berger S., Baudendistel I., Kamradt M., Eckrich F., Szecsenyi J.. Complexity of care and strategies of self-management in patients with colorectal cancer. Patient Prefer. Adherence (2017) 11 731-742. DOI: 10.2147/PPA.S127612 11. Adams K., Greiner A.C., Corrigan J.M.. Report of a summit. The First Annual Crossing the Quality Chasm Summit: A Focus on Communities (2004) 12. McCorkle R., Ercolano E., Lazenby M., Schulman-Green D., Schilling L.S., Lorig K., Wagner E.H.. Self-management: Enabling and empowering patients living with cancer as a chronic illness. CA Cancer J. Clin. (2011) 61 50-62. DOI: 10.3322/caac.20093 13. Chan R.J., Mayer D., Koczwara B., Loerzel V., Charalambous A., Agbejule O.A., Howell D.. Building capacity in cancer nurses to deliver self-management support: A call for action paper. Cancer Nurs. (2020) 43 341-342. DOI: 10.1097/NCC.0000000000000855 14. Eagen L., Levesque J.. Transforming community cancer care: The Ottawa Regional Cancer Foundation’s cancer coaching practice. Univ. Ott. J. Med. (2017) 7 15. DOI: 10.18192/uojm.v7i1.1944 15. Taylor S.J.C., Pinnock H., Epiphaniou E., Pearce G., Parke H.. A rapid synthesis of the evidence on interventions supporting self-management for people with long- term conditions. (PRISMS Practical Systematic Review of Self-Management Support for long-term conditions). Health Serv. Deliv. Res. (2014) 2 54. DOI: 10.3310/hsdr02530 16. Greene J., Hibbard J.. Why does patient activation matter? An examination of the relationships between patient activation and health-related outcomes. J. Gen. Intern. Med. (2012) 27 520-526. DOI: 10.1007/s11606-011-1931-2 17. Howell D., Harth T., Brown J.A., Bennet C., Boyco S.. Self-management education interventions for patients with cancer: A systematic review. Support Care Cancer (2017) 25 1323-1355. DOI: 10.1007/s00520-016-3500-z 18. Cuthbert C.A., Farragher J.F., Hemmelgarn B.R., Ding Q., McKinnon G.P., Cheung W.Y.. Self-management interventions for cancer survivors: A systematic review and evaluation of intervention content and theories. Psycho-Oncology (2019) 28 2119-2140. DOI: 10.1002/pon.5215 19. Kim S.H., Kim K., Mayer D.K.. Self-management interventions for adult cancer survivors after treatment: A systematic review and meta-analysis. Oncol. Nurs. Forum. (2017) 44 719-728. DOI: 10.1188/17.ONF.719-728 20. Kantilal K., Hardeman W., Whiteside H., Karapanagiotou E., Small M., Bhattacharya D.. Facilitating healthcare practitioners to deliver self-management support in adult cancer survivors: A realist review. Res. Soc. Admin. Pharm. (2022) 18 3870-3883. DOI: 10.1016/j.sapharm.2022.05.011 21. Creer T.L., Holroyd K.A., Baum A., Newman S., Weinman J., West R., McManus C.. Self-management of chronic conditions. Cambridge Handbook of Psychology, Health and Medicine (2006) 22. Dinh H., Thuy T., Bonner A., Clark R., Ramsbotham J., Hines S.. The effectiveness of the teach-back method on adherence and self-management in health education for people with chronic disease: A systematic review. JBI Evid. Synth. (2016) 14 210-247. DOI: 10.11124/jbisrir-2016-2296 23. Howell D., Pond G., Bryant Lukosius D., Powis M., McGowan P., Tutsirai M., Kukretti V., Rask S., Hack S., Krzyzanowska M.K.. Feasibility and effectiveness of self-management online education and telephone coaching on patient activation in managing cancer treatment toxicities. J. Natl. Compr. Cancer Netw. (2023) 21 247-255. DOI: 10.6004/jnccn.2022.7095 24. McGowan P., Lynch S., Hensen F.. The role and effectiveness of telephone peer coaching for adult patients with Type 2 Diabetes. Can. J. Diabetes (2019) 43 399-405. DOI: 10.1016/j.jcjd.2019.03.006 25. Glasgow R.E., Davis C.L., Funnell M.M., Beck A.. Implementing practical interventions to support chronic illness self-management. Jt. Comm. J. Qual. Saf. (2003) 29 563-574. DOI: 10.1016/S1549-3741(03)29067-5 26. 26. Registered Nurses’ Association of Ontario Strategies to Support Self-Management in Chronic Conditions: Collaboration with ClientsRegistered Nurses’ Association of OntarioToronto, ON, Canada2010. Strategies to Support Self-Management in Chronic Conditions: Collaboration with Clients (2010) 27. Hudon C., Chouinard M.C., Diadiou F., Bouliane D., Lambert M., Hudon É.. The Chronic Disease Self-Management Program: The experience of frequent users of health care services and peer leaders. Fam. Pract. (2016) 33 167-171. DOI: 10.1093/fampra/cmw007 28. Wolever R.Q., Simmons L.A., Sforzo G.A., Dill D., Kaye M., Bechard E.M., Southard M.E., Kennedy M., Vosloo J., Yang N.. A systematic review of the literature on health and wellness coaching: Defining a key behavioral intervention in healthcare. Glob. Adv. Health Med. (2013) 2 38-57. DOI: 10.7453/gahmj.2013.042 29. Tharani A., Van Hecke A., Ali T.S., Duprez V.. Factors influencing nurses’ provision of self-management support for patients with chronic illnesses: A systematic mixed studies review. Int. J. Nurs. Stud. (2021) 120 103983. DOI: 10.1016/j.ijnurstu.2021.103983 30. Collins C., Doran G., Patton P., Fitsgerald R., Rochfort A.. Does education of primary care professionals promote self-management and improve outcomes in chronic disease? An updated review. BJGP Open (2021) 5 BJGPO.2020.0186. DOI: 10.3399/BJGPO.2020.0186 31. Franklin M., Lewis S., Willis K., Bourke-Taylor H., Smith L.. Patients’ and healthcare professionals’ perceptions of self-management support interactions: Systematic review and qualitative synthesis. Chronic Illn. (2018) 14 79-103. DOI: 10.1177/1742395317710082 32. Roy D., Mahony F., Horsburgh M., Bycroft J.. Partnering in primary care in New Zealand: Client’ and nurses’ experience of the Flinders Program ™ in the management of long term conditions. J. Nurs. Healthc. Chronic Illn. (2011) 3 140-149. DOI: 10.1111/j.1752-9824.2011.01088.x 33. Duprez V., Vandecasteele T., Verhaeghe S., Beeckman D., Van Hecke A.. The effectiveness of interventions to enhance self-management support competencies in the nursing profession: A systematic review. J. Adv. Nurs. (2016) 73 1807-1824. DOI: 10.1111/jan.13249 34. Chan R.J., Oluvaseyifunmsi A., Crawford-Williams F., Yates P., Koczwara B., Erikson J., So W.K., Howell D.. Development of an international competency framework for nurses in the provision of self-management support to cancer populations. Cancer Nurs. (2022). DOI: 10.1097/NCC.0000000000001143 35. Howell D., Mayer D.K., Fielding R., Eicher M., Verdonck-de Leeuw I.M., Johansen C., Soto-Perez-de-Celis E., Foster C., Chan R., Alfano C.M.. Management of Cancer and Health After the Clinic Visit: A Call to Action for Self-Management in Cancer Care. JNCI (2020) 113 523-531. DOI: 10.1093/jnci/djaa083 36. Lawn S., Zhi X., Morellow A.. An integrative review of e-learning in the delivery of self-management support training for health professionals. BMC Med. Educ. (2017) 17. DOI: 10.1186/s12909-017-1022-0 37. Self-Management in Cancer. Quality Standards (2018) 38. Wagner E.H., Austin B.T., Von Korff M.. Organizing care for patients with chronic illness. Milbank Q (1996) 74 511-544. DOI: 10.2307/3350391 39. Glasgow R.E., Funnell M.M., Bonomi A.E., Davis C., Beckham V., Wagner E.H.. Self-Management aspects of improving chronic illness care breakthrough series: Implementation with diabetes and heart failure teams. Ann. Behav. Med. (2002) 24 80-87. DOI: 10.1207/S15324796ABM2402_04 40. Medland M.. Self-Management Support: A Study and Implementation Guide for Health Care Professionals (2009) 41. Self-Management Toolkit. A Resource for Health Care Providers. (2018) 42. Whitelock E.P., Orleans C.T., Pender N., Allan J.. Evaluating primary care behavioral counselling interventions: An evidence-based approach. Am. J. Prev. Med. (2002) 22 267-284. DOI: 10.1016/S0749-3797(02)00415-4
--- title: Association between Systemic Immune Inflammation Index and Cognitive Impairment after Acute Ischemic Stroke authors: - Yuanfei Bao - Lingling Wang - Chaopin Du - Yan Ji - Yiwei Dai - Wei Jiang journal: Brain Sciences year: 2023 pmcid: PMC10046597 doi: 10.3390/brainsci13030464 license: CC BY 4.0 --- # Association between Systemic Immune Inflammation Index and Cognitive Impairment after Acute Ischemic Stroke ## Abstract Background and Aims: Post-stroke cognitive impairment (PSCI) is one of the major complications after ischemic stroke. PSCI has been shown to be associated with low-grade systemic inflammation. As a novel inflammatory marker, the systemic immune-inflammation (SII) index could reflect clinical outcomes in severe cardiovascular diseases. We therefore performed a prospective study to investigate the correlation between the SII index and the risk of PSCI in patients with ischemic stroke. Methods: We prospectively enrolled 254 patients with ischemic stroke with symptoms onset <72 h. The SII index was detected within 24 h after admission. The Montreal Cognitive Scale (MoCA) was utilized to evaluate cognitive function, and PSCI was defined as a MoCA score of <25 points. Results: During the 3-month follow-up, 70 participants ($27.6\%$) had mild cognitive impairment and 60 ($23.6\%$) had severe cognitive impairment. In binary logistic regression analysis, each one-standard deviation increase in the SII index was significantly associated with the prevalence of PSCI after adjusting for age, sex, and other confounders (odds ratio 2.341; $95\%$ confidence interval, 1.439–3.809, $$p \leq 0.001$$). Similar significant findings were observed when SII was defined as a categorical variable. In addition, the multiple-adjusted spline regression model showed a linear association between the SII index and cognitive impairment ($$p \leq 0.003$$ for linearity). Conclusions: Our study indicated that an increased SII index was closely related to PSCI at 3 months in patients with ischemic stroke. Further research is required to evaluate the efficacy of inflammation management in these patients. ## 1. Introduction Stroke is one of the diseases with the highest disability and morbidity in China and worldwide [1,2,3,4,5]. Post-stroke cognitive impairment (PSCI) is a frequent complication of ischemic stroke, and it affects 20–$80\%$ of patients, depending on diagnostic criteria, countries, and races [6,7,8]. The period between stroke onset and the presence of PSCI can be recognized as a treatment window for early intervention to protect cognitive function [9]. Therefore, determining reliable predictors for PSCI is of vital importance for continuously improving the prognosis of stroke. In recent years, several researchers have progressively recognized the secondary injury of the brain’s inflammatory response after ischemic stroke [10,11]. Preclinical and clinical studies have confirmed a causal link between sterile low-grade inflammation and the pathogenesis of stroke [12,13,14]. Accumulating evidence also reported that several pro-inflammatory and inflammatory molecules were involved in the inflammatory response to cognitive impairment, including C-reactive protein, interleukin-6, and interleukin-10 [15]. The crosstalk between the immune system and the central nervous system during ischemic stroke may be the underlying mechanism of brain tissue injury and complications. The systemic immune-inflammation index (SII) is a new type of comprehensive inflammation index combining peripheral lymphocytes, neutrophils, and platelet counts [16]. The SII was first proposed in a variety of cancer research, aiming to identify patients with a high risk of recurrence or death [16,17,18]. Zhang et al. performed a meta-analysis and reported that increased pretreatment SII significantly correlated with worse overall survival and recurrence-free survival/progression-free survival in biliary tract cancers [18]. SII can be used to comprehensively evaluate the inflammation status better than the neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, platelet-to-lymphocyte ratio, and platelet volume. In the central nervous system, the SII index could differentiate the high- and low-grade gliomas [19]. In recent years, the SII score has been studied as a marker for stroke outcomes [20,21,22,23]. Weng et al. conducted a retrospective study consecutively including 365 patients who were treated with intravenous thrombolysis, which found that SII is correlated with stroke severity and can be a novel prognostic biomarker [22]. Yi et al. retrospectively reviewed prospectively collected data from the stroke database of each institution and found that decreased SII index was associated with favorable clinical outcomes in patients who underwent mechanical thrombectomy for large artery occlusion [23]. However, the results of previous studies on ischemic stroke did not sufficiently demonstrate the effectiveness of the SII score as a potential predictor of PSCI after ischemic stroke. In our prospective cohort study, we aimed to systematically evaluate the relationship between the SII index and the risk of PSCI in patients with acute ischemic stroke. ## 2.1. Study Sample First-ever patients with ischemic stroke who were treated at the Nantong Third People’s Hospital between September 2021 and October 2022 were prospectively selected. Patients were enrolled based on the following criterion: [1] age ≥ 18 years; [2] had time from symptom onset to admission within 72 h; [3] had a baseline National Institutes of Health Stroke Scale (NIHSS) score ≤ 8 points. The exclusion criteria were as follows: [1] patients with obvious disorder of consciousness; [2] patients with obvious apraxia and/or aphasia who were unable to complete the questionnaire; [3] patients complicated with diseases that might affect cognitive function, such as brain tumor, Parkinson’s syndrome, Alzheimer’s disease, frontotemporal dementia, schizophrenia, major depression, and alcohol abuse; [4] patients complicated with diseases which might affect inflammatory conditions, such as malignant tumors, acute infection, and trauma. This study was reviewed and approved by the Medical Research Ethics Committee of the Nantong Third People’s Hospital (MB2020034). The written informed consent for participation in the study was obtained from each participant or their caregivers. ## 2.2. Baseline Data Collection Demographic characteristics, vascular risk factors, clinical data, and laboratory data were all collected after admission. The blood pressure was measured after admission. Hypertension was defined as a history of hypertension and/or the use of antihypertensive medications. Hyperlipidemia was defined as a history of hyperlipidemia and/or having received treatment for dyslipidemia. Diabetes mellitus was defined as a history of diabetes and/or the use of glucose-lowering agents. The baseline stroke severity was assessed by experienced neurologists using the NIHSS [24]. The stroke subtype was classified according to the criteria of Trial of Org 10172 in Acute Stroke Treatment [25]. The laboratory data, including peripheral blood count, total cholesterol, triglyceride, high-density lipoprotein, low-density lipoprotein, fasting blood glucose, high-sensitivity C-reactive protein (Hs-CRP), and SII index, were recorded. Blood samples were routinely obtained within 24 h of hospital admission. The lipid profile and fasting blood glucose were detected by Chemistry Analyzer (AU480, Beckman Coulter, Brea, CA, USA). Then, the cell counts were analyzed by an auto-analyzer (XE-2100, Sysmex, Kobe, Japan) and utilized to calculate the SII index. According to previous studies [21,26,27], the SII index was calculated as: platelet count × neutrophil count/lymphocyte count. ## 2.3. Assessment of Cognitive Function The study outcome was PSCI at 3 months after stroke, assessed by a professionally trained neurologist who was blinded to the clinical and laboratory data, using the Montreal Cognitive Assessment (MoCA) scale [28]. The total MoCA score is 30 points. To correct for errors, 1 point was added to the total MoCA score in participants with <12 years of education. In this analysis, a score of <25 on the total MoCA indicated the presence of PSCI [29,30]. Furthermore, according to the recommended cutoffs, the degree of cognitive impairment was categorized as follows: severe cognitive impairment (0–19), mild cognitive impairment (20–24), and no cognitive impairment (25–30). ## 2.4. Statistical Analysis Statistical analyses were performed using the statistical software SPSS (version 25.0; IBM, New York, NY, USA), and R (version 4.0.1; The R Project for Statistical Computing, Vienna, Austria). Continuous variables were expressed as means with standard deviation (SD) or median with interquartile range (IQR). Categorical variables were described by frequencies with percentages. Differences in baseline characteristics in the patients with and without were compared using an independent sample t-test, an χ2 test, one-way analysis of variance (Bonferroni post hoc test), and a Kruskal–Wallis H-test. Logistic regression analysis was utilized to determine the association of the SII index with the presence and severity of PSCI after adjusting for potential confounders. All multivariate analyses were first adjusted for age and gender (Model 1) and additionally adjusted for all variables (including age, gender, education years, hypertension, diabetes mellitus, stroke subtypes, baseline stroke severity, and Hs-CRP levels; Model 2). Results were shown as adjusted odds ratio (OR) ($95\%$ confidence interval, CI). Furthermore, we performed the restricted cubic splines with 3 knots (at the 5th, 50th, and 95th percentiles) to explore the pattern and magnitude of the relationship between the SII index and clinical outcomes [31]. A p-value < 0.05 was considered statistically significant for all the analyses. ## 3.1. Baseline Characteristics In this study, we identified 254 patients with ischemic stroke who met our inclusion criteria. The eligible participants included 143 men and 111 women with a mean age of 65.5 ± 10.2 years. Among these patients with ischemic stroke, $68.5\%$ had hypertension, $27.6\%$ had diabetes mellitus, and $15.0\%$ had hyperlipidemia. The median (IQR) SII index was 574.2 (337.6–888.9) × 109/L, and the median (IQR) NIHSS score at admission was 3.0 (2.0–5.0) points. The demographic and clinical characteristics of the study population stratified by the SII quartiles are presented in Table 1. Age, diastolic blood pressure, the prevalence of PSCI, and Hs-CRP levels differed significantly with the increasing quartile of the SII index. However, sex and the presence of vascular risk factors did not differ significantly among the categories. ## 3.2. Prevalence and Risk Factors of PSCI During the 3-month follow-up, 130 participants ($51.2\%$) had PSCI. Table 2 demonstrated the data of the comparison between patients with and without PSCI. On univariate analysis, patients with PSCI were older (mean, 67.3 ± 8.9 years versus 63.4 ± 11.1 years; $$p \leq 0.004$$), had a higher prevalence of hypertension ($74.6\%$ versus $62.1\%$; $$p \leq 0.032$$), diabetes mellitus ($33.8\%$ versus $21.0\%$; $$p \leq 0.022$$), and large artery atherosclerosis stroke ($$p \leq 0.035$$), and had a higher baseline NIHSS score (median, 4.0 score versus 3.0 score; $$p \leq 0.001$$), Hs-CRP levels (median, 6.8 mg/L versus 4.5 mg/L; $$p \leq 0.031$$), and SII index (median, 653.9 × 109/L versus 493.1 × 109/L; $$p \leq 0.001$$). Furthermore, educational years were lower in patients with PSCI than in patients without PSCI (median, 9.0 years versus 9.0 years; $$p \leq 0.001$$). ## 3.3. Logistic Regression Analysis for the Relationship between the SII Index and PSCI In order to study whether the SII index could be used as an effective variable for the diagnosis of the presence and severity of PSCI, we utilized a multivariate logistic regression analysis. The results of the logistic analysis were demonstrated in Table 3. After controlling for age and sex, a high SII index is a significant risk factor for PSCI (per 1-SD increase, OR, 2.045; $95\%$ CI, 1.347–3.106; $$p \leq 0.001$$; Model 1). In Model 2, after further controlling for education years, hypertension, diabetes, stroke subtypes, baseline NIHSS score, and Hs-CRP levels, an increased SII index also significantly correlated with 3-month PSCI (per 1-SD increase, OR, 2.341; $95\%$ CI, 1.439–3.809; $$p \leq 0.001$$). Similar significant findings were observed when SII was categorized according to the quartile. In addition, the multiple-adjusted spline regression model showed a linear association between the SII index and cognitive impairment ($$p \leq 0.003$$ for linearity; Figure 1). Among these patients with PSCI, 70 participants ($27.6\%$) had mild PSCI and 60 ($23.6\%$) had severe PSCI. After controlling for the potential confounders, the ordinal regression analysis further confirmed the close association between the SII index and severity of PSCI (per SD increase, OR, 1.879; $95\%$ CI, 1.324–2.537; $$p \leq 0.001$$). ## 4. Discussion This study showed that patients with acute ischemic stroke suffering from cognitive impairment had a significantly higher SII index than those without cognitive impairment. Furthermore, even after adjusting for potential confounders, the positive association between the SII score and PSCI remained significant. SII, as a noninvasive and cost-effective serological inflammatory marker, represents potential prognostic predictors for PSCI screening and management. Previously available studies on PSCI used inconsistent definitions and study populations, leading to discrepancies in incidence rates and associated factors. This study reported that $51.2\%$ of patients who suffered from ischemic stroke present with PSCI at 3 months, which is consistent with earlier literature [6,7,8]. Our data further showed that the PSCI risk was higher in patients with older age and low education levels, which was similar to previous studies [6,32]. Elderly patients often accompany decreased cerebral blood flow and arteriosclerosis, which is more likely to cause cognitive dysfunction. Additionally, several studies have found that patients with a higher education level tend to have a larger cognitive reserve capacity after stroke [33]. We also confirmed that patients with diabetes are at higher risk for cognitive impairment. Diabetes mellitus is often accompanied by a more severe oxidative stress reaction and longer exposure to glycolipid metabolic disorders. These factors significantly mediate the correlation between diabetes and cognitive impairment [34]. The SII index was established by Hu and designed to predict the clinical outcomes in hepatocellular carcinoma patients after operative treatment [16], which showed that the predictive value of the SII index was more accurate than those indexes that use only one or two cell subtypes. In recent years, a higher level of the SII index is a predicting factor for poor outcomes in brain pathologies such as glioma, aneurysmal subarachnoid hemorrhage, ischemic stroke, and post-stroke depression [19,20,21,22,23,24,35]. Weng et al. conducted a retrospective study consecutively including 365 patients who were treated with intravenous thrombolysis, which found that SII is correlated with stroke severity and can be a novel prognostic biomarker [22]. Yi et al. retrospectively reviewed prospectively collected data from the stroke database of each institution and found that a decreased SII index was associated with favorable clinical outcomes in patients who underwent mechanical thrombectomy for large artery occlusion [23]. Additionally, Yun et al. have performed an analysis of inflammatory markers including SII in patients with aneurysmal subarachnoid hemorrhage and reported that an SII index value ≥960 × 109/L was an independent predicting factor for poor prognosis [35]. In addition, data from a 2-year cross-sectional, stratified, multistage probability cluster survey demonstrated that the SII score was significantly higher in subjects with depression after ischemic stroke than those without it [20]. Our study extended the current knowledge about the adverse effect of a higher SII index in ischemic stroke, as it demonstrated a positive association between the SII score and the risk of PSCI at 3 months in patients with ischemic stroke. Although PSCI is a heterogeneous condition after stroke, some researchers supposed that PSCI is a multisystem inflammatory disease [36]. The factors constituting the index might explain the effect of SII in PSCI. Neutrophils infiltrate into the ischemic tissue within several hours after stroke and peak at 48 h. Increased neutrophil levels lead to the increased expression of matrix metalloproteinase-9 and the release of pro-inflammatory mediators, which might damage the blood–brain barrier [37]. Furthermore, acute inflammatory secretions such as reactive oxygen species and proteolytic enzymes produced by neutrophils may further deteriorate these processes [38]. On the contrary, lymphocytes have been found to inhibit endothelial damage by regulating the inflammatory response [39]. The blood–brain barrier dysfunction may induce damage to the white matter and be related to a progression of cognitive impairment [40]. Platelets are an indicator of inflammation after stroke. Platelets interact directly with circulating leukocytes, thereby forming platelet–leukocyte aggregates and activating the innate immune response to ischemia [41]. In addition, the activation of platelets may have a negative effect on the phosphorylation and expression of brain-derived neurotrophic factor, which is a key molecule for memory in the healthy and the pathological brain [42]. Our observations, combined with results obtained in animal models and clinical studies, suggest that immuno-inflammatory responses may mediate cognitive function after ischemic stroke, offering a potential therapeutic target for intervention. The strengths of our study include using a standardized research method, detailed psychological evaluation, and enrolling a homogeneous sample of patients with mild ischemic stroke, all of which make this group appropriate for examining the association between the SII score and the risk of PSCI. However, there are several limitations that should be addressed in this study. Firstly, the SII index used in this study was measured from one blood test only. It is useful to take routine blood tests constantly so as to detect the dynamic changes in SII score after ischemic stroke. Secondly, patients with obvious disorders of consciousness, apraxia and/or aphasia, and a history of cognitive decline were excluded from this study, which may underestimate the real PSCI rate and lead to a selection bias. Thirdly, the cross-sectional design of this study prevents causal inference. Further prospective studies with larger samples are needed to establish causality. In conclusion, the current study revealed that the higher SII index at baseline was significantly associated with 3-month PSCI in patients with acute ischemic stroke. Further clinical research is warranted to examine whether managing inflammation reactions provides a potential preventive or therapeutic target for PSCI. ## References 1. Wang Y.J., Li Z.X., Gu H.Q., Zhai Y., Jiang Y., Zhao X.Q., Zhao J.Z.. China Stroke Statistics 2019: A Report from the National Center for Healthcare Quality Management in Neurological Diseases, China National Clinical Research Center for Neurological Diseases, the Chinese Stroke Association, National Center for Chronic and Non-communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention and Institute for Global Neuroscience and Stroke Collaborations. Stroke Vasc. Neurol. (2020) 5 211-239. PMID: 32826385 2. Global, regional, and national age-sex specific mortality for 264 causes of death, 1980–2016: A systematic analysis for the Global Burden of Disease Study 2016. Lancet (2017) 390 1151-1210. DOI: 10.1016/S0140-6736(17)32152-9 3. Wu S., Wu B.O., Liu M., Chen Z., Wang W., Anderson C.S., Zhang S.. Stroke in China: Advances and challenges in epidemiology, prevention, and management. Lancet Neurol. (2019) 18 394-405. DOI: 10.1016/S1474-4422(18)30500-3 4. Soto-Camara R., Gonzalez-Santos J., González-Berna J., Trejo-Gabriel-Galan J.M.. Factors associated with a rapid call for assistance for patients with ischemic stroke. Emergencias (2020) 32 33-39. PMID: 31909910 5. Tu W.J., Hua Y., Yan F., Bian H., Yang Y., Lou M., Wang L.. Prevalence of stroke in China, 2013–2019: A population-based study. Lancet Reg. Health W. Pac. (2022) 28 100550. DOI: 10.1016/j.lanwpc.2022.100550 6. Sun J., Tan L., Yu J.. Post-stroke cognitive impairment: Epidemiology, mechanisms and management. Ann. Transl. Med. (2014) 2 80. PMID: 25333055 7. Pendlebury S., Rothwell P.. Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: A systematic review and meta-analysis. Lancet Neurol. (2009) 8 1006-1018. DOI: 10.1016/S1474-4422(09)70236-4 8. Pendlebury S., Rothwell P.. Incidence and prevalence of dementia associated with transient ischaemic attack and stroke: Analysis of the population-based Oxford Vascular Study. Lancet Neurol. (2019) 18 248-258. DOI: 10.1016/S1474-4422(18)30442-3 9. Brainin M., Tuomilehto J., Heiss W.D., Bornstein N.M., Bath P.M., Teuschl Y.. Post Stroke Cognition Study Group. Post-stroke cognitive decline: An update and perspectives for clinical research. Eur. J. Neurol. (2015) 22 229-e16. DOI: 10.1111/ene.12626 10. Kim J.Y., Kawabori M., Yenari M.A.. Innate inflammatory responses in stroke: Mechanisms and potential therapeutic targets. Curr. Med. Chem. (2014) 21 2076-2097. DOI: 10.2174/0929867321666131228205146 11. Pagram H., Bivard A., Lincz L.F., Levi C.. Peripheral Immune Cell Counts and Advanced Imaging as Biomarkers of Stroke Outcome. Cerebrovasc. Dis. Extra. (2016) 6 120-128. DOI: 10.1159/000450620 12. Kim J., Park J., Chang J., Kim S., Lee J.. Inflammation after Ischemic Stroke: The Role of Leukocytes and Glial Cells. Exp. Neurobiol. (2016) 25 241-251. DOI: 10.5607/en.2016.25.5.241 13. Chamorro Á., Meisel A., Planas A., Urra X., Van de Beek D., Veltkamp R.. The immunology of acute stroke. Nat. Rev. Neurol. (2012) 8 401-410. DOI: 10.1038/nrneurol.2012.98 14. Chen G., Nuñez G.. Sterile inflammation: Sensing and reacting to damage. Nat. Rev. Immunol. (2010) 10 826-837. DOI: 10.1038/nri2873 15. Zhang X., Bi X.. Post-Stroke Cognitive Impairment: A Review Focusing on Molecular Biomarkers. J. Mol. Neurosci. (2020) 70 1244-1254. DOI: 10.1007/s12031-020-01533-8 16. Hu B., Yang X.R., Xu Y., Sun Y.F., Sun C., Guo W., Fan J.. Systemic immune-inflammation index predicts prognosis of patients after curative resection for hepatocellular carcinoma. Clin. Cancer Res. (2014) 20 6212-6622. DOI: 10.1158/1078-0432.CCR-14-0442 17. Wang B.L., Tian L., Gao X.H., Ma X.L., Wu J., Zhang C.Y., Yang X.R.. Dynamic change of the systemic immune inflammation index predicts the prognosis of patients with hepatocellular carcinoma after curative resection. Clin. Chem. Lab. Med. (2016) 54 1963-1969. DOI: 10.1515/cclm-2015-1191 18. Zhang B., Yao W.. Prognostic role of the systemic immune-inflammation index in biliary tract cancers: A meta-analysis of 3515 patients. World J. Surg. Oncol. (2022) 20 320. DOI: 10.1186/s12957-022-02783-z 19. Liang R., Chen N., Li M., Wang X., Mao Q., Liu Y.. Significance of systemic immune-inflammation index in the differential diagnosis of high- and low-grade gliomas. Clin. Neurol. Neurosurg. (2018) 164 50-52. DOI: 10.1016/j.clineuro.2017.11.011 20. Hu J., Wang L., Fan K., Ren W., Wang Q., Ruan Y., He J.. The Association Between Systemic Inflammatory Markers and Post-Stroke Depression: A Prospective Stroke Cohort. Clin. Interv./Aging. (2021) 16 1231-1239. DOI: 10.2147/CIA.S314131 21. Çırakoğlu Ö., Yılmaz A.. Systemic immune-inflammation index is associated with increased carotid intima-media thickness in hypertensive patients. Clin. Exp. Hypertens. (2021) 43 565-571. DOI: 10.1080/10641963.2021.1916944 22. Weng Y., Zeng T., Huang H., Ren J., Wang J., Yang C., Chen G.. Systemic Immune-Inflammation Index Predicts 3-Month Functional Outcome in Acute Ischemic Stroke Patients Treated with Intravenous Thrombolysis. Clin. Interv. Aging. (2021) 16 877-886. DOI: 10.2147/CIA.S311047 23. Yi H.J., Sung J.H., Lee D.H.. Systemic Inflammation Response Index and Systemic Immune-Inflammation Index Are Associated with Clinical Outcomes in Patients Treated with Mechanical Thrombectomy for Large Artery Occlusion. World Neurosurg. (2021) 153 e282-e289. DOI: 10.1016/j.wneu.2021.06.113 24. Muir K., Weir C., Murray G., Povey C., Lees K.. Comparison of neurological scales and scoring systems for acute stroke prognosis. Stroke (1996) 27 1817-1820. DOI: 10.1161/01.STR.27.10.1817 25. Adams H.J., Bendixen B., Kappelle L., Biller J., Love B., Gordon D.. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke (1993) 24 35-41. DOI: 10.1161/01.STR.24.1.35 26. Luo F., Li Y., Zhao Y., Sun M., He Q., Wen R., Xie Z.. Systemic immune-inflammation index predicts the outcome after aneurysmal subarachnoid hemorrhage. Neurosurg. Rev. (2022) 45 1607-1615. DOI: 10.1007/s10143-021-01681-4 27. Xu M., Chen R., Liu L., Liu X., Hou J., Liao J., Hu P.. Systemic immune-inflammation index and incident cardiovascular diseases among middle-aged and elderly Chinese adults: The Dongfeng-Tongji cohort study. Atherosclerosis (2021) 323 20-29. DOI: 10.1016/j.atherosclerosis.2021.02.012 28. Nasreddine Z., Phillips N., Bédirian V., Charbonneau S., Whitehead V., Collin I., Cummings J.L., Chertkow H.. The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J. Am. Geriatr. Soc. (2005) 53 695-699. DOI: 10.1111/j.1532-5415.2005.53221.x 29. Webb A., Pendlebury S., Li L.. Validation of the Montreal cognitive assessment versus mini-mental state examination against hypertension and hypertensive arteriopathy after transient ischemic attack or minor stroke. Stroke (2014) 45 3337-3342. DOI: 10.1161/STROKEAHA.114.006309 30. Zhong C., Bu X., Xu T., Guo L., Wang X., Zhang J., He J.. Serum Matrix Metalloproteinase-9 and Cognitive Impairment After Acute Ischemic Stroke. J. Am. Heart Assoc. (2018) 7 e007776. DOI: 10.1161/JAHA.117.007776 31. Alves H.C., Treurniet K.M., Dutra B.G., Jansen I.G., Boers A.M., Santos E.M., Majoie C.B.. Associations Between Collateral Status and Thrombus Characteristics and Their Impact in Anterior Circulation Stroke. Stroke (2018) 49 391-396. DOI: 10.1161/STROKEAHA.117.019509 32. Mohd Z.M., Ghazali S., Che D.N., Singh D., Subramaniam P.. A Review of Risk Factors for Cognitive Impairment in Stroke Survivors. Sci. World J. (2016) 2016 3456943 33. Duda B., Puente A., Miller L.. Cognitive reserve moderates relation between global cognition and functional status in older adults. J. Clin. Exp. Neuropsychol. (2014) 36 368-378. DOI: 10.1080/13803395.2014.892916 34. Ahmad W.. Overlapped metabolic and therapeutic links between Alzheimer and diabetes. Mol. Neurobiol. (2013) 47 399-424. DOI: 10.1007/s12035-012-8352-z 35. Yun S., Yi H., Lee D., Sung J.. Systemic Inflammation Response Index and Systemic Immune-inflammation Index for Predicting the Prognosis of Patients with Aneurysmal Subarachnoid Hemorrhage. J. Stroke Cerebrovasc. Dis. (2021) 30 105861. DOI: 10.1016/j.jstrokecerebrovasdis.2021.105861 36. Swardfager W., Winer D., Herrmann N., Winer S., Lanctôt K.. Interleukin-17 in post-stroke neurodegeneration. Neurosci. Biobehav. Rev. (2013) 37 436-447. DOI: 10.1016/j.neubiorev.2013.01.021 37. Ong C.W., Pabisiak P.J., Brilha S., Singh P., Roncaroli F., Elkington P.T., Friedland J.S.. Complex regulation of neutrophil-derived MMP-9 secretion in central nervous system tuberculosis. J. Neuroinflammation. (2017) 14 31. DOI: 10.1186/s12974-017-0801-1 38. Winterbourn C., Kettle A., Hampton M.. Reactive Oxygen Species and Neutrophil Function. Annu. Rev. Biochem. (2016) 85 765-792. DOI: 10.1146/annurev-biochem-060815-014442 39. Gill D., Veltkamp R.. Dynamics of T cell responses after stroke. Curr. Opin. Pharmacol. (2016) 26 26-32. DOI: 10.1016/j.coph.2015.09.009 40. Rosenberg G., Sullivan N., Esiri M.. White matter damage is associated with matrix metalloproteinases in vascular dementia. Stroke (2001) 32 1162-1168. DOI: 10.1161/01.STR.32.5.1162 41. Maugeri N., Rovere-Querini P., Evangelista V., Covino C., Capobianco A., Bertilaccio M.T., Manfredi A.A.. Neutrophils phagocytose activated platelets in vivo: A phosphatidylserine, P-selectin, and {beta}2 integrin-dependent cell clearance program. Blood (2009) 113 5254-5265. DOI: 10.1182/blood-2008-09-180794 42. Miranda M., Morici J., Zanoni M., Bekinschtein P.. Brain-Derived Neurotrophic Factor: A Key Molecule for Memory in the Healthy and the Pathological Brain. Front. Cell. Neurosci. (2019) 13 363. DOI: 10.3389/fncel.2019.00363
--- title: Co-Culture of Glomerular Endothelial Cells and Podocytes in a Custom-Designed Glomerulus-on-a-Chip Model Improves the Filtration Barrier Integrity and Affects the Glomerular Cell Phenotype authors: - Daan C. ‘t Hart - Dilemin Yildiz - Valentina Palacio-Castañeda - Lanhui Li - Burcu Gumuscu - Roland Brock - Wouter P. R. Verdurmen - Johan van der Vlag - Tom Nijenhuis journal: Biosensors year: 2023 pmcid: PMC10046631 doi: 10.3390/bios13030339 license: CC BY 4.0 --- # Co-Culture of Glomerular Endothelial Cells and Podocytes in a Custom-Designed Glomerulus-on-a-Chip Model Improves the Filtration Barrier Integrity and Affects the Glomerular Cell Phenotype ## Abstract Crosstalk between glomerular endothelial cells and glomerular epithelial cells (podocytes) is increasingly becoming apparent as a crucial mechanism to maintain the integrity of the glomerular filtration barrier. However, in vitro studies directly investigating the effect of this crosstalk on the glomerular filtration barrier are scarce because of the lack of suitable experimental models. Therefore, we developed a custom-made glomerulus-on-a-chip model recapitulating the glomerular filtration barrier, in which we investigated the effects of co-culture of glomerular endothelial cells and podocytes on filtration barrier function and the phenotype of these respective cell types. The custom-made glomerulus-on-a-chip model was designed using soft lithography. The chip consisted of two parallel microfluidic channels separated by a semi-permeable polycarbonate membrane. The glycocalyx was visualized by wheat germ agglutinin staining and the barrier integrity of the glomerulus-on-a-chip model was determined by measuring the transport rate of fluorescently labelled dextran from the top to the bottom channel. The effect of crosstalk on the transcriptome of glomerular endothelial cells and podocytes was investigated via RNA-sequencing. Glomerular endothelial cells and podocytes were successfully cultured on opposite sides of the membrane in our glomerulus-on-a-chip model using a polydopamine and collagen A double coating. Barrier integrity of the chip model was significantly improved when glomerular endothelial cells were co-cultured with podocytes compared to monocultures of either glomerular endothelial cells or podocytes. Co-culture enlarged the surface area of podocyte foot processes and increased the thickness of the glycocalyx. RNA-sequencing analysis revealed the regulation of cellular pathways involved in cellular differentiation and cellular adhesion as a result of the interaction between glomerular endothelial cells and podocytes. We present a novel custom-made glomerulus-on-a-chip co-culture model and demonstrated for the first time using a glomerulus-on-a-chip model that co-culture affects the morphology and transcriptional phenotype of glomerular endothelial cells and podocytes. Moreover, we showed that co-culture improves barrier function as a relevant functional readout for clinical translation. This model can be used in future studies to investigate specific glomerular paracrine pathways and unravel the role of glomerular crosstalk in glomerular (patho) physiology. ## 1. Introduction The glomerulus is crucial for normal renal function by filtering the blood across the glomerular filtration barrier (GFB). The GFB comprises two cell types: fenestrated glomerular endothelial cells (GEnC) and visceral epithelial cells (podocytes). These two cell types are separated by the glomerular basement membrane, which mainly consists of collagen IV and laminin [1]. GEnC are covered by glycocalyx, a thick, negatively charged layer of carbohydrates. The glycocalyx plays an important role in glomerular function, for example by preserving glomerular structural integrity and affecting charge-selective glomerular permselectivity [2,3,4]. GFB injury is a crucial aspect of the pathogenesis of both acquired and hereditary forms of renal diseases, such as diabetic nephropathy and focal segmental glomerulosclerosis (FSGS) [5]. The aforementioned glomerulopathies are characterized by GEnC and podocyte injury, podocyte foot process effacement- and depletion, proteinuria, reduced glomerular filtration and eventually loss of kidney function [6,7,8]. A crucial aspect of in vivo GFB physiology is the crosstalk between GEnC and podocytes. Various studies have highlighted that GEnC and podocyte crosstalk is fundamental for the GFB to function as a filtration barrier [9,10]. For example, the secretion of vascular endothelial growth factor (VEGF) by podocytes affects GEnCs and is an important mechanism to prevent albuminuria [11]. In addition, secretion of the endothelial growth factor Angiopoietin-1 by podocytes stabilizes the glomerular capillaries [12,13]. Furthermore, GEnC have been shown to interact with podocytes via the secretion of exosomes [14]. Moreover, activation of protein kinase C by endothelium-derived thrombomodulin, protects against the development of podocyte injury [15]. In addition to these known examples, more knowledge about the paracrine crosstalk between GEnC and podocytes is likely to lead to a further understanding of in vivo GFB (patho)physiology. Currently, our understanding of the interaction between GEnC and podocytes is limited due to the small number of studies using in vitro models that truly recapitulate the in vivo physiology of the glomerulus [16,17,18,19,20]. Until recently, in vitro studies investigating the pathogenic mechanisms of glomerular cell injury used mainly 2D monocultures of e.g., GEnC or podocytes. Organ-on-a-chip technology has been widely used to successfully study elements of the physiology of the heart, lung, liver, and kidney [16,17,18,19,20,21,22,23,24]. Organ-on-a-chip technology has a high potential to create biologically-relevant and complex in vitro models of organs and tissues; in our case the glomerulus, [25,26]. In these models, complex glomerular biology can be mimicked by co-culturing GEnC and podocytes in spatial arrangements that recapitulate the physiological tissue architecture. In addition, an important advantage of organ-on-a-chip technology compared to simpler models such as Transwell inserts, is the possibility to eventually implement important conditions like laminar flow and trans-glomerular filter pressure gradients. A recent publication described a glomerulus-on-a-chip model using amniotic fluid-derived podocytes and cultured podocytes and GEnC on top of each other in a single channel [17]. Of note, however, amniotic fluid-derived podocytes are not routinely applicable for high-throughput use. Moreover, this set-up also precludes stimulation or treatment of one specific cell-type, as well as easy experimental separation and subsequent individual analysis of the two cell types. Importantly, whether podocytes and GEnCs affect each other by crosstalk mechanisms was actually not investigated in the previously developed glomerulus-on-a-chip models [16,17,18,19,20]. Given the deficits of the systems presented so far, the current study aimed to develop a glomerulus-on-a-chip model that can be used to investigate the effects of crosstalk between GEnC and podocytes on the GFB as a whole, as well as the effects of crosstalk on the phenotype (e.g., gene expression) of the individual glomerular cell types separately. For this purpose, we designed a glomerulus-on-a-chip model with functional filtration capacity by co-culturing conditionally immortalized mouse GEnC and podocyte cell lines on opposite sides of a semi-permeable membrane. The use of conditionally immortalized cell lines in a tailored microfluidic device gives future possibility for high-throughput application. We demonstrate that culturing of podocytes together with GEnC affects gene expression profiles and the morphology and functional phenotype of both GEnC and podocytes, and also improves the filtration barrier of the in vitro GFB. ## 2.1. Microfluidic Chip Design and Assemblance Figure 1a. shows a schematic overview of the chip design for the co-culture of GEnC and podocytes. In our chip model, the endothelial and podocyte compartment are separated using a track-etched polycarbonate membrane (8 µm pore size, porosity (void volume) min-max: 4–$20\%$) (WHA155846, Sigma-Aldrich, Zwijndrecht, The Netherlands). The SU-8 master mold was fabricated by patterning a negative photoresist (SU-8 2150, Microchemicals GmbH) on a 4-inch silicon wafer (Si-Mat Silicon Materials, Germany) using conventional photolithography. The resulting SU-8 pattern was about 180 ± 10 µm measured using DektakXT® stylus profilometer (Bruker, Billerica, MA, USA). After fabrication, a silanization process was performed by placing the mold in Trichloro(1H,1H,2H,2H-perfluorooctyl)silane (Sigma-Aldrich, Zwijndrecht, The Netherlands) vapor overnight, leaving the mold ready for PDMS casting. Microfluidic organ-on-a-chip devices were subsequently produced as described previously [27]. In brief, endothelial and podocyte compartments were cast with polydimethylsiloxane (PDMS) (Sylgard 184, Sigma-Aldrich, Zwijndrecht, The Netherlands) in a 10:1 ratio (w/w) to curing agent. The PDMS mixture was subsequently degassed using a vacuum pump for 10 min and rested for 5 min at room temperature (RT). This step was repeated three times, whereafter the PDMS was cured for 2.5 h at 65 °C. After curing, the PDMS was cut to size and the inlets were punched using a 1.2 mm sized puncher (Harris Uni-Core, Sigma-Aldrich, Zwijndrecht, The Netherlands). When assembling the two-layer chips, a modified version of the protocol provided by Sip and Folch was used [28]. The experimental steps to fabricate the SU-8 master mold and coat the polycarbonate membrane are visualized in Figure 1b. In order to create chemical reactivity, the polycarbonate membranes were first oxygen plasma treated using a PDC-32-G-2 plasma cleaner (Harrick Plasma, 1 min, power high). The membranes were subsequently treated using a $2\%$ bis-amino-silane (413,356, Sigma-Aldrich, Zwijndrecht, The Netherlands) solution in isopropanol with $1\%$ H2O for 20 min at 80 °C to bind reactive silane groups to the polycarbonate membrane. The membranes were afterwards washed with isopropanol, cured for 30 min at 65 °C and immersed for 30 min in $70\%$ EtOH at RT in small aluminium trays (57 × 16 mm, 40.8 mL) (Avantor, Arnhem, The Netherlands). The endothelial and podocyte PDMS compartments were subsequently oxygen plasma treated to activate the reactive groups of the PDMS (1 min, power high) and directly bonded to the polycarbonate membrane. To ensure an irreversible bond between the reactive silane groups on the polycarbonate membrane and the activated reactive groups of the PDMS compartments, the bond was cured overnight at 65 °C. Finally, the PDMS device was plasma-bonded to a 24 × 50 mm glass cover slide (Avantor, Allentown, PA, USA) (1 min, power high) for 1 h at 65 °C. To create a hydrophilic PDMS surface, the microfluidic device was subsequently oxygen plasma treated (1 min, power high). To ensure a clean cell culturing environment, the microfluidic device and its channels were washed with $70\%$ (v/v) ethanol. As the podocyte compartment was blocked by the polycarbonate membrane after assembly, the corresponding two inlets were punctured with a sterile p20 pipet tip. After washing, the ethanol was removed and the microfluidic device was washed with sterile PBS. Hereafter, a p20 pipette was used to reach the inlets of the microfluidic channels when handling the microfluidic device. ## 2.2. Cell Culture Conditionally immortalized mouse podocytes (MPC-5) and conditionally immortalized mouse glomerular endothelial cells (mGEnC) were cultured as described previously [29,30]. In brief, MPC-5 and mGEnC were grown at permissive conditions at 33 °C to allow proliferation. To induce differentiation in the microfluidic device, cells were transferred to 37 °C with removal of IFNγ from the growth medium. For the experiments to determine the optimal coating for the cells to adhere and differentiate on the polycarbonate membrane outside the chip, three different coating strategies were tested; [1] 3 h at 37 °C with 1 µg/cm2 bovine fibronectin (Thermofisher Scientific, Breda, The Netherlands), [2] 3 h at 37 °C with 1 mg/mL collagen A (i.e., functionally the same as Collagen I) (Merck, Schiphol-Rijk, The Netherlands) or [3] 1 h using 2 mg/mL polydopamine (H8502, Sigma-Aldrich, Zwijndrecht, The Netherlands) in 10 mM Tris-HCl (pH 8.5) at RT, washed with MilliQ, dried for 1 h at 65 °C and subsequently coated for 1 h using 1 mg/mL collagen A at 37 °C. When MPC-5 and mGEnC were seeded in the microfluidic device, the polycarbonate membrane was coated using coating strategy 3, as described above. Two microfluidic devices were stored in a 100 × 20 mm petri dish during culturing (Corning, Sigma-Aldrich, Zwijndrecht, The Netherlands). MPC-5 was seeded into the bottom channel of the chip with a density of 1 × 106 cells/mL. Chips were immediately flipped 180° for 3 h to allow the MPC-5 to adhere to the polycarbonate membrane. After 3 h, chips were flipped back, the medium was placed on top of the microfluidic device to prevent evaporation, and MPC-5 were grown in the device for 14 days to ensure complete differentiation. Every 2 days, the medium was refreshed by adding fresh medium to the channels and fresh medium was placed on top of the microfluidic device. In case of a slanted microfluidic device, which resulted in the inability of the medium to create a surface tension, a custom designed 3D printed lid (Supplementary Figure S2) was inserted in the podocyte compartment inlets. MPC-5 need 14 days to ensure complete differentiation, whereas mGEnC need only 7 days. Therefore, seven days after seeding of MPC-5, mGEnC were seeded with a density of 2 × 107 cells/mL in the top channel of the chip device. Following seeding of mGEnC in the chips, a 1:1 mixture of mGEnC and MPC-5 medium was used to culture the cells. MPC-5 and mGEnC were co-cultured for the remaining 7 days at 37 °C to ensure complete differentiation of both cell types. Every 2 days, the medium was refreshed by adding fresh medium to the channels and fresh medium was replaced on top of the microfluidic device. For the experiments with CellTracker, MPC-5 were stained for 30 min at 37 °C with 12.5 µM CMPTX CellTracker Red and mGEnC with 12.5 µM CMFDA CellTracker Green in serum-free media. 72 h upon seeding of MPC-5, confocal images were acquired as described below. ## 2.3. Immunofluorescent Stainings Cells were fixed for 15 min at RT by replacing the medium in the channels with $2\%$ paraformaldehyde (PFA) with $4\%$ sucrose in PBS to stain against von Willebrand Factor (vWF) and synaptopodin. For staining with wheat germ agglutinin (WGA), the cells were fixed for 10 min at RT by replacing the medium with $90\%$ ice-cold acetone. When cells were fixed with PFA, the cells were washed three times with PBS and cells were subsequently permeabilized for 10 min at RT using a $0.3\%$ Triton X-100 solution in 1× PBS. Thereafter, cells were washed three times with PBS and blocked for 30 min at RT by replacing the PBS in the microfluidic channel with blocking solution ($2\%$ BSA, $2\%$ FBS and $0.2\%$ fish gelatin in PBS). In case of acetone fixation, cells were washed three times with PBS upon fixation and PBS was replaced by a $1\%$ BSA solution as blocking solution for 30 min at RT. Biotinylated WGA lectin (1:1000 dilution, Vectorlabs, Burlingame, CA, USA) and primary antibodies for vWF (1:25 dilution, DAKO, A0082, rabbit polyclonal) and synaptopodin (1:5 dilution, Progen, 65,194, mouse monoclonal) were diluted in blocking solution and incubated for 1 h at RT by replacing the respective blocking solutions in the microfluidic channels. Streptavidin-conjugated Alexa 594, goat anti-rabbit Alexa 488 and goat anti-mouse IgG1 Alexa 488 (both from Thermofisher Scientific, Breda, The Netherlands) were diluted in blocking solution (1:200) and incubated for 45 min in the dark at RT upon washing the cells three times with PBS. Actin Green 488 or Actin Red 555 (2 drops per 1 mL blocking solution, Thermofisher Scientific, Breda, The Netherlands) were used to stain the actin cytoskeleton and were incubated for 60 min in the dark at RT. Cell nuclei were stained using 2 µg/mL Hoechst 33,342 (Thermofisher Scientific, Breda, The Netherlands) in blocking solution for 5 min at RT. ## 2.4. Microscopy and Image Analysis All fluorescent images were acquired using a Zeiss LSM900 inverted confocal laser scanning microscope with Airyscan 2 with a EC Plan-Neofluar 10x/0.45 or a Plan-Apochromat 40x/1.2 water long-distance objective. The surface area of the podocyte foot processes was quantified using the FiloQuant plug-in of FIJI (version 1.53c). In short, a threshold was applied to the images, wherein FiloQuant was used to shave off the protrusions. After, the shaved image was subtracted from the thresholded image and the area of the resulting picture was measured (Supplementary Figure S3). For measuring the total fluorescent intensity of the WGA glycocalyx, Z-stacks were made with a slice interval of 1 µm using a Plan-Apochromat 40x/1.2 water objective. Fluorescent intensity of the WGA Z-stack was measured per slice using FIJI. Three-dimensional reconstructions were made using the 3D viewer plug-in of FIJI. ## 2.5. Barrier Integrity Assay To validate the barrier integrity of the chip, 0.5 mg/mL 40-kDA sulphated dextran-FITC and 155-kDA dextran-TRITC (Sigma-Aldrich, Zwijndrecht, The Netherlands) was added to the endothelial compartment and collected from the podocyte compartment after 30 min incubation at 37 °C. Fluorescent intensity was measured at a fluorescent plate reader (Tecan, Infinite Pro2000) and used as a measure of barrier permeability from the endothelial to the podocyte compartment. ## 2.6. RNA Isolation To collect mGEnC and MPC-5 from the chips, cells were washed once with PBS and subsequently incubated with trypsin for 5 min in the chip. To collect the mGEnC, the top endothelial channel was flushed via the top channel inlet with 1 mL medium, while simultaneously collecting the medium from the top channel outlet. Afterwards, the same procedure was repeated for the bottom channel to collect the MPC-5. Per experimental condition, cells isolated from six chips were pooled into one sample. The following four experimental conditions were tested; [1] mGEnC from monoculture chips, [2] MPC-5 from monoculture chips, [3] mGEnC from co-culture chips, and [4] MPC-5 from co-culture chips. Total RNA isolation was performed using a RNeasy microkit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. RNA concentration and quality were assessed on a Qubit Fluorometer (Thermofischer Scientific, Breda, The Netherlands). ## 2.7. RNA Sequencing and Analysis Bulk RNA sequencing was performed using Single Cell Discoveries (Utrecht, The Netherlands). RNA extraction and library preparation followed the CEL-seq2 protocol with a sequencing depth of 20 million reads per sample. R version 3.5.2 was used to analyse the RNA sequencing data. Bulk RNA sequencing count normalization and differential gene expression analysis was performed using the DESeq2 package v3.15 [31]. Significant differentially expressed genes between experimental groups (i.e., co-culture samples vs mono-culture samples, per cell-type) were selected using a log2 fold change threshold of <−1.5 and >1.5. Gene set enrichment analysis (GSEA) was performed using the GSEA v4.1.0 software package [32,33]. Gene functional classification was performed using DAVID tools and functional annotation clustering was performed using GOTERM_BP_DIRECT. Figures were created using BioRender.com. ## 2.8. Statistical Analyses Data are presented as mean ± SEM. Statistical analysis was conducted with a two-tailed student’s t-test when two experimental groups were compared or a one-way ANOVA with Tukey’s Post-Hoc test when three or more experimental conditions were analysed. All statistical analyses were performed using GraphPad Prism version 5.03 (GraphPad Software, Inc., San Diego, CA, USA). A p-value of < 0.05 was considered statistically significant. ## 3.1. Polydopamine and Collagen A Double Coating Is Most Optimal for Glomerular Cell Growth on Polycarbonate Membrane First, we optimized the coating which could allow for the growth of mGEnC and MPC-5 on the polycarbonate membrane outside the chip. Notably, over the last decade, we have gained extensive experience in 2D monoculture with these two cell lines and exactly know how these cells behave morphologically and functionally [29,34,35]. The following four experimental conditions were investigated: [1] uncoated membranes, [2] membranes coated with fibronectin, [3] membranes coated with collagen A and [4] membranes coated with both polydopamine and collagen A. We found that coating the membrane first with polydopamine and then with collagen A was the optimal membrane coating, as this combination resulted in complete monolayer growth and differentiation for both cell types after their respective differentiation period (i.e., 7 days for mGEnC and 14 days for MPC-5) (Figure 2a,b). In comparison, neither mGEnC nor MPC-5 adhered properly on uncoated membranes. On membranes solely coated with fibronectin, only a small number of MPC-5 were able to grow, whereas mGEnC were completely unable to grow. On membranes coated with collagen A, the growth of mGEnC was improved profoundly but was still insufficient to obtain a complete monolayer of mGEnC. Collagen A coating of the membrane also improved the growth of MPC-5 compared to the fibronectin-coated membrane. Coating of the membranes with both polydopamine and collagen A yielded complete monolayers for both mGEnC and MPC-5. Moreover, MPC-5 displayed a large cell surface area and expressed synaptopodin (data not shown), thereby indicating their complete differentiation [36]. Therefore, the double coating of polycarbonate membranes with polydopamine and collagen A was selected to be used in this study to culture mGEnC and MPC-5 in the glomerulus-on-a-chip device. ## 3.2. Development of a Co-Culture in the Microfluidic Device After identifying the optimal membrane coating for culturing mGEnC and MPC-5 on the polycarbonate membrane outside the chip, we investigated the behavior of mGEnC and MPC-5 when co-cultured in our glomerulus-on-a-chip device. We found that mGEnC were seeded into the top channel of the chip, whereas MPC-5 were seeded into the bottom channel of the chip to grow on the bottom side of the membrane. We confirmed that mGEnC in monoculture also formed a monolayer on the polydopamine and collagen A double coated membrane in the chip (Figure 3a). Furthermore, mGEnC displayed high levels of vWF expression in Weibel-Palade bodies, which confirmed the complete differentiation of mGEnC in the chip (Figure 3a, second panel from the left, indicated by the arrows) [29]. MPC-5 also formed a complete monolayer when cultured in monoculture in the chip on the bottom side of the membrane coated with both polydopamine and collagen A (Figure 3b). Moreover, MPC-5 showed high expression levels of the podocyte-maturity marker synaptopodin, which validated the correct differentiation of the podocytes in our chip device [36]. Upon verifying the separate growth and differentiation of mGEnC and MPC-5 in our chip device, we also confirmed that mGEnC and MPC-5 simultaneously formed intact monolayers on opposite sides of the membrane in our chip device using fluorescent CellTracker dyes (Figure 3c–e). These findings validated the successful co-culture of mGEnC and MPC-5 in our glomerulus-on-a-chip model. ## 3.3. Co-Culture Affects Podocyte Morphology and Increases Glycocalyx Thickness After ascertaining that mGEnC and MPC-5 could be co-cultured in our chip device and showed a differentiated endothelial and podocyte phenotype, we investigated the effect of co-culture on the morphology and functional phenotype of mGEnC and MPC-5. First, we observed an altered podocyte morphology in co-culture, which was characterized by extended podocyte protrusions and an increased number of filopodia (Figure 4a,b). Moreover, the surface area of podocyte filopodia was increased when cultured together with mGEnC (Figure 4c, $p \leq 0.05$). We subsequently used the lectin WGA to study the effect of co-culture on the formation of the endothelial glycocalyx in our chip model. WGA detects both hyaluronic acid (HA) and unsulfated domains within heparan sulfate (HS). HA is the main structural component of the glycocalyx and responsible for maintaining the gel-like structure of the glycocalyx. In addition, HS is the main functional component of the glycocalyx and prevents for example the binding of leukocytes and cytokines to the healthy endothelium, whereas it facilitates binding of leukocytes and cytokines under inflammatory conditions [2]. Notably, the formation of the endothelial glycocalyx increased when mGEnC were co-cultured with MPC-5 compared to mGEnC cultured in monoculture (Figure 4d,e). In addition, the intensity of the WGA staining was significantly increased when mGEnC were cultured together with MPC-5 compared to mGEnC cultured in monoculture, suggesting a denser glycocalyx in co-culture (Figure 4f,g, $p \leq 0.05$). ## 3.4. Co-Culture of GEnC and Podocytes Improves Functional Barrier Integrity of the Glomerulus-on-a-Chip Model Maintaining the integrity of the charge and permselective barrier is an important characteristic for the correct functioning of the glomerulus in vivo, and glomerular injury is associated with increased GFB permeability and eventually albuminuria and proteinuria. Thus, after investigating the effect of co-culture on the morphology of the individual cell types, we investigated the effect of co-culture on the functional integrity of the filtration barrier in our chip model. To determine the functional barrier integrity, we measured the passage rate of 155-kDA dextran-TRITC from the top to the bottom channel of the microfluidic device, i.e., from endothelial channel to podocyte channel. The monoculture of mGEnC and MPC-5 resulted in a ~$50\%$ decreased passage of 155-kDA dextran-TRITC across the barrier compared to chips without cells (Figure 4h,i, $p \leq 0.001$ and $p \leq 0.01$, respectively). Importantly, the passage rate of 155-kDA dextran-TRITC across the barrier was further decreased in chips with a co-culture of mGEnC and MPC-5 compared to chips with a monoculture of either mGEnC or MPC-5 (Figure 4j,k, $$p \leq 0.05$$ and $p \leq 0.05$). We obtained similar findings using the negatively charged 40-kDA sulphated dextran-FITC dye (Supplementary Figure S1). These findings highlight the importance of both podocytes and GEnC as a functional part of the filtration barrier in our chip model. ## 3.5. Co-Culture of GEnC and Podocytes Regulate Pathways Involved in Cell Differentiation and Cell Adhesion Finally, we studied the effect of co-culture in the glomerulus-on-a-chip device on gene expression of both GEnC and podocytes by performing whole transcriptome bulk RNA-seq analysis. The experimental setup is visualized in Figure 5a. Principal component analysis (PCA) revealed that co-culture profoundly affected the RNA expression profile of both MPC-5 and mGEnC (Figure 5b). Upon selecting genes with a log fold change in gene expression of <−1.5 or >1.5, we observed that 1054 on a total of the 15,485 sequenced genes were significantly upregulated in mGEnC and 925 genes out of the 15,485 sequenced genes were significantly downregulated when cultured together with MPC-5 (Figure 5c). In addition, 730 out of the 15,485 sequenced genes were significantly upregulated and 675 out of the 15,485 sequenced genes were significantly downregulated in MPC-5 in co-culture with mGEnC compared to monoculture. We performed functional gene classification to investigate which biological processes appeared regulated in MPC-5 and mGEnC as a result of co-culture. Notably, pathways involved in cellular differentiation, cell adhesion and kidney development were regulated in mGEnC as a result of co-culture with MPC-5 (Figure 5d,f). In addition, pathways involved in extracellular matrix organization, cell differentiation and cell adhesion were affected in MPC-5 when cultured together with mGEnC (Figure 5e,g). Taken together, co-culture of GEnC and podocytes in our novel glomerulus-on-a-chip system profoundly affected the transcriptome of both GEnCs as well as podocytes. ## 4. Discussion In this study, we report the design and application of a novel microfluidic glomerulus-on-a-chip model, designed to enable studying cross-talk between cell types. In this glomerulus-on-a-chip model glomerular endothelial cells and podocytes were cultured on different sides of a semi-permeable membrane in separate channels in a microfluidic chip design, thereby reconstituting the in vivo glomerular architecture. We were able for the first time to successfully reconstitute the filtration barrier function of the GFB in a microfluidic device by culturing two conditionally immortalized glomerular cell lines. In order to mirror the in vivo situation, we chose to use a very thin poly-carbonate membrane as physical support to mimic the glomerular basement membrane and establish the glomerular filtration barrier. The microfluidic setup was optimized using a poly-carbonate membrane, ensuring proper chemical reactivity of the three layers (PDMS-membrane-PDMS) to induce bonding, leakage-free membrane integration and practical assembly steps (Figure 1). We showed that co-culture of differentiated GEnCs and podocytes in this device alters both their morphology and transcriptome, the thickness of the glycocalyx as well as cross-membrane barrier integrity, in comparison to monocultures. Crosstalk between GEnC and podocytes is known to be crucial to maintain the integrity of the GFB [9,10]. However, studies using advanced in vitro models to accurately study these mechanisms are still scarce. The GFB is often considered as an unchanging biological entity. However, over the last decade, an increasing number of studies have suggested that the dynamic interaction between the different components of the GFB plays a crucial role in glomerular health and glomerular disease [13,37,38]. By developing a microfluidic glomerulus-on-a-chip model, we can demonstrate the importance of co-culture of GEnC and podocytes on the morphology and functional phenotype of both cell types. In addition, we showed that co-culture resulted in a thicker glycocalyx. The glycocalyx is a key structural element of the GFB and plays a role in immune cell adhesion, growth factor binding, glomerular structural integrity, specific endothelial function as well as overall glomerular barrier function [2,3,4]. Interestingly, although structurally different from endothelial cells, podocytes also produce a glycocalyx, the thickness of which is. affected by diabetic conditions, although the contribution of podocyte glycocalyx to the charge-dependent barrier function of GFB remains elusive [3,39,40]. Generally, a decreased thickness of the endothelial glycocalyx is associated with albuminuria and has been previously observed in the pathogenesis of kidney pathology such as diabetic nephropathy and chronic kidney disease [41,42]. The findings presented in this study reveal that co-culture of glomerular endothelial cells and podocytes, and thus probably glomerular crosstalk, is important to maintain a thicker glycocalyx and subsequently to prevent the development of albuminuria during renal health and disease. In addition to the effect of glomerular crosstalk on the glycocalyx, our experiments have also revealed the effect of co-culture on cellular differentiation in general. Our findings following bulk RNA sequencing experiments provide a further insight, that co-culture led to profound alterations of biological pathways in both cell types. Moreover, the effect on cellular differentiation was further substantiated by differences in the surface area of podocyte filopodia. Of note, the retraction and shrinkage of podocyte foot processes, known as podocyte foot process effacement, is a pathological characteristic observed in a wide variety of glomerular diseases [43,44]. In addition, podocyte foot process effacement is often interpreted as evidence that podocyte injury is the initial step in the pathogenesis of glomerular diseases. However, the data shown in this study might suggest that podocyte foot process effacement could also be the result of pathogenetic mechanisms originating in the glomerular endothelium instead. Clearly, the aim of using organ-on-a-chip technology in this study was not to recapitulate a completely functional kidney outside the human body. It was rather a strategy to recapitulate a crucial and complex substructure of the kidney, namely the filtration barrier, which cannot be addressed with conventional 2D monoculture strategies or cannot be readily investigated in animal models. In the current study, we primarily wanted to demonstrate that the interaction between GEnC and podocytes could be studied in our custom glomerulus-on-a-chip device, and that we could measure differences in the clinically relevant outcome parameter like filtration barrier integrity. Whereas podocytes form the architectural backbone and provide mechanical stability to the barrier, glomerular endothelial cell functions are regulated by podocytes with respect to e.g., growth and differentiation via paracrine factors such as VEGF [45,46]. This perfectly illustrates the interplay between physical and functional properties of podocytes and GEnC in the GFB. Additional features and improvements may be added to the current chip model in future studies depending on the research question, e.g., addition of a channel containing mesangial cells. Over the past few years, different glomerulus-on-a-chip models have been developed via distinct methodologies [16,17,18,19,20]. For example, one study used a commercially available organ-on-a-chip platform [17], while other studies used soft lithography to develop a custom-made glomerulus-on-a-chip model [16,18,19,20]. Whether podocytes and GEnCs separated by a membrane affect each other by crosstalk mechanisms was not investigated in the previously developed glomerulus-on-a-chip models. In the current study, we decided to use soft lithography to design our glomerulus-on-a-chip model, as this allowed us to tailor the design of the microfluidic device to answer our research question: the effect of co-culture on the phenotype of GEnC and podocytes. Designing a glomerulus-on-a-chip model consisting of two parallel microfluidic channels, separated by a polycarbonate membrane, allowed us to study the two cell types separately and also enabled the easy separation of the two cell types for RNA-seq analysis after co-culture. Whereas some studies cultured isolated glomeruli in their microfluidic device, other studies cultured stem cell-derived podocytes or amniotic fluid-derived podocytes. In the current study, we were able for the first time to successfully reconstitute the filtration barrier function of the GFB in a microfluidic device by culturing two conditionally immortalized glomerular cell lines. We chose these two cell lines because we already had comprehensive knowledge about both cell lines and therefore knew precisely how they should grow and differentiate [29,34,35]. Importantly, demonstrating that conditionally immortalized glomerular cell lines can be used to construct a glomerulus-on-a-chip opens the possibility for a high through-put application, which is challenging when culturing amniotic fluid-derived podocytes or isolated glomeruli. In conclusion, the glomerulus-on-a-chip model we developed can be used in future studies to further investigate the crosstalk between GEnC and podocytes in vitro. An improved understanding of cellular interactions in the glomerulus will greatly advance our understanding of the molecular mechanisms in GFB health and disease, and might eventually lead to the discovery of new therapeutic strategies for the treatment of glomerulopathies. In addition, the model can be used to perform disease-modelling and subsequent drug screening studies to reduce the use of animal studies in drug development. For example, plasma from patients with primary FSGS or membranous nephropathy (MN) could be added to our glomerulus-on-a-chip model to recapitulate the pathogenesis of FSGS and MN [47]. Furthermore, the podocyte injury inducing compounds Adriamycin or puromycin aminonucleoside (PAN) could be used in future studies to mimic the pathophysiology of FSGS [48,49]. Moreover, an alternative strategy to use the glomerulus-on-a-chip model for disease modelling would be the culture of GEnC and podocytes derived from stem cells from patients, e.g., those with Alport syndrome or other inherited glomerulopathies, even enabling personalized approaches [50]. Finally, the glomerulus-on-a-chip model developed in this study can also be used to study the molecular mechanisms underlying the migration of immune cells through the GFB by introducing immune cells to the endothelial compartment. As glomerular infiltration of immune cells is an important pathogenic mechanism of immune-related glomerular diseases, the model developed in the current study might provide new insights in the pathogenesis of e.g., diabetic nephropathy and glomerulonephritis [51,52]. ## References 1. Jarad G., Miner J.H.. Update on the glomerular filtration barrier. Curr. Opin. Nephrol. Hypertens. (2009) 18 226-232. DOI: 10.1097/MNH.0b013e3283296044 2. Rops A.L., Loeven M.A., van Gemst J.J., Eversen I., Van Wijk X.M., Dijkman H.B., van Kuppevelt T.H., Berden J.H., Rabelink T.J., Esko J.D.. Modulation of heparan sulfate in the glomerular endothelial glycocalyx decreases leukocyte influx during experimental glomerulonephritis. Kidney Int. (2014) 86 932-942. DOI: 10.1038/ki.2014.115 3. Garsen M., Lenoir O., Rops A.L., Dijkman H.B., Willemsen B., van Kuppevelt T.H., Rabelink T.J., Berden J.H., Tharaux P.L., van der Vlag J.. Endothelin-1 Induces Proteinuria by Heparanase-Mediated Disruption of the Glomerular Glycocalyx. J. Am. Soc. Nephrol. (2016) 27 3545-3551. DOI: 10.1681/ASN.2015091070 4. Van den Berg B.M., Wang G., Boels M.G.S., Avramut M.C., Jansen E., Sol W., Lebrin F., van Zonneveld A.J., de Koning E.J.P., Vink H.. Glomerular Function and Structural Integrity Depend on Hyaluronan Synthesis by Glomerular Endothelium. J. Am. Soc. Nephrol. (2019) 30 1886-1897. DOI: 10.1681/ASN.2019020192 5. Hebert L.A., Parikh S., Prosek J., Nadasdy T., Rovin B.H.. Differential diagnosis of glomerular disease: A systematic and inclusive approach. Am. J. Nephrol. (2013) 38 253-266. DOI: 10.1159/000354390 6. Lu C.C., Wang G.H., Lu J., Chen P.P., Zhang Y., Hu Z.B., Ma K.L.. Role of Podocyte Injury in Glomerulosclerosis. Adv. Exp. Med. Biol. (2019) 1165 195-232. DOI: 10.1007/978-981-13-8871-2_10 7. Trimarchi H.. Mechanisms of Podocyte Detachment, Podocyturia, and Risk of Progression of Glomerulopathies. Kidney Dis. (2020) 6 324-329. DOI: 10.1159/000507997 8. Ziyadeh F.N., Wolf G.. Pathogenesis of the podocytopathy and proteinuria in diabetic glomerulopathy. Curr. Diabetes Rev. (2008) 4 39-45. DOI: 10.2174/157339908783502370 9. Fogo A.B.. Talking back: The podocytes and endothelial cells duke it out. Kidney Int. (2016) 90 1157-1159. DOI: 10.1016/j.kint.2016.08.031 10. Fu J., Lee K., Chuang P.Y., Liu Z., He J.C.. Glomerular endothelial cell injury and cross talk in diabetic kidney disease. Am. J. Physiol.-Renal Physiol. (2015) 308 F287-F297. DOI: 10.1152/ajprenal.00533.2014 11. Eremina V., Sood M., Haigh J., Nagy A., Lajoie G., Ferrara N., Gerber H.P., Kikkawa Y., Miner J.H., Quaggin S.E.. Glomerular-specific alterations of VEGF-A expression lead to distinct congenital and acquired renal diseases. J. Clin. Investig. (2003) 111 707-716. DOI: 10.1172/JCI17423 12. Satchell S.C., Harper S.J., Tooke J.E., Kerjaschki D., Saleem M.A., Mathieson P.W.. Human podocytes express angiopoietin 1, a potential regulator of glomerular vascular endothelial growth factor. J. Am. Soc. Nephrol. (2002) 13 544-550. DOI: 10.1681/ASN.V132544 13. Mahtal N., Lenoir O., Tharaux P.L.. Glomerular Endothelial Cell Crosstalk With Podocytes in Diabetic Kidney Disease. Front. Med. (2021) 8 659013. DOI: 10.3389/fmed.2021.659013 14. Wu X., Gao Y., Xu L., Dang W., Yan H., Zou D., Zhu Z., Luo L., Tian N., Wang X.. Exosomes from high glucose-treated glomerular endothelial cells trigger the epithelial-mesenchymal transition and dysfunction of podocytes. Sci. Rep. (2017) 7 9371. DOI: 10.1038/s41598-017-09907-6 15. Isermann B., Vinnikov I.A., Madhusudhan T., Herzog S., Kashif M., Blautzik J., Corat M.A., Zeier M., Blessing E., Oh J.. Activated protein C protects against diabetic nephropathy by inhibiting endothelial and podocyte apoptosis. Nat. Med. (2007) 13 1349-1358. DOI: 10.1038/nm1667 16. Musah S., Mammoto A., Ferrante T.C., Jeanty S.S.F., Hirano-Kobayashi M., Mammoto T., Roberts K., Chung S., Novak R., Ingram M.. Mature induced-pluripotent-stem-cell-derived human podocytes reconstitute kidney glomerular-capillary-wall function on a chip. Nat. Biomed. Eng. (2017) 1 0069. DOI: 10.1038/s41551-017-0069 17. Petrosyan A., Cravedi P., Villani V., Angeletti A., Manrique J., Renieri A., De Filippo R.E., Perin L., Da Sacco S.. A glomerulus-on-a-chip to recapitulate the human glomerular filtration barrier. Nat. Commun. (2019) 10 3656. DOI: 10.1038/s41467-019-11577-z 18. Wang L., Tao T., Su W., Yu H., Yu Y., Qin J.. A disease model of diabetic nephropathy in a glomerulus-on-a-chip microdevice. Lab Chip (2017) 17 1749-1760. DOI: 10.1039/C7LC00134G 19. Zhou M., Zhang X., Wen X., Wu T., Wang W., Yang M., Wang J., Fang M., Lin B., Lin H.. Development of a Functional Glomerulus at the Organ Level on a Chip to Mimic Hypertensive Nephropathy. Sci. Rep. (2016) 6 31771. DOI: 10.1038/srep31771 20. Roye Y., Bhattacharya R., Mou X., Zhou Y., Burt M.A., Musah S.. A Personalized Glomerulus Chip Engineered from Stem Cell-Derived Epithelium and Vascular Endothelium. Micromachines (2021) 12. DOI: 10.3390/mi12080967 21. Beckwitt C.H., Clark A.M., Wheeler S., Taylor D.L., Stolz D.B., Griffith L., Wells A.. Liver ‘organ on a chip’. Exp. Cell Res. (2018) 363 15-25. DOI: 10.1016/j.yexcr.2017.12.023 22. Lee J., Kim S.. Kidney-on-a-Chip: A New Technology for Predicting Drug Efficacy, Interactions, and Drug-induced Nephrotoxicity. Curr. Drug Metab. (2018) 19 577-583. DOI: 10.2174/1389200219666180309101844 23. Sakamiya M., Fang Y., Mo X., Shen J., Zhang T.. A heart-on-a-chip platform for online monitoring of contractile behavior via digital image processing and piezoelectric sensing technique. Med. Eng. Phys. (2020) 75 36-44. DOI: 10.1016/j.medengphy.2019.10.001 24. Kızılkurtlu A.A., Polat T., Aydın G.B., Akpek A.. Lung on a Chip for Drug Screening and Design. Curr. Pharm. Des. (2018) 24 5386-5396. DOI: 10.2174/1381612825666190208122204 25. Azizipour N., Avazpour R., Rosenzweig D.H., Sawan M., Ajji A.. Evolution of Biochip Technology: A Review from Lab-on-a-Chip to Organ-on-a-Chip. Micromachines (2020) 11. DOI: 10.3390/mi11060599 26. Wu Q., Liu J., Wang X., Feng L., Wu J., Zhu X., Wen W., Gong X.. Organ-on-a-chip: Recent breakthroughs and future prospects. Biomed. Eng. Online (2020) 19 9. DOI: 10.1186/s12938-020-0752-0 27. Chernyavska M., Schmid M., Freitag P.C., Palacio-Castañeda V., Piruska A., Huck W.T.S., Plückthun A., Verdurmen W.P.R.. Unravelling Receptor and RGD Motif Dependence of Retargeted Adenoviral Vectors using Advanced Tumor Model Systems. Sci. Rep. (2019) 9 18568. DOI: 10.1038/s41598-019-54939-9 28. Sip C.G., Folch A.. Stable chemical bonding of porous membranes and poly(dimethylsiloxane) devices for long-term cell culture. Biomicrofluidics (2014) 8 036504. DOI: 10.1063/1.4883075 29. Rops A.L., van der Vlag J., Jacobs C.W., Dijkman H.B., Lensen J.F., Wijnhoven T.J., van den Heuvel L.P., van Kuppevelt T.H., Berden J.H.. Isolation and characterization of conditionally immortalized mouse glomerular endothelial cell lines. Kidney Int. (2004) 66 2193-2201. DOI: 10.1111/j.1523-1755.2004.66009.x 30. Shankland S.J., Pippin J.W., Reiser J., Mundel P.. Podocytes in culture: Past, present, and future. Kidney Int. (2007) 72 26-36. DOI: 10.1038/sj.ki.5002291 31. Love M.I., Huber W., Anders S.. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. (2014) 15 550. DOI: 10.1186/s13059-014-0550-8 32. Mootha V.K., Lindgren C.M., Eriksson K.F., Subramanian A., Sihag S., Lehar J., Puigserver P., Carlsson E., Ridderstråle M., Laurila E.. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. (2003) 34 267-273. DOI: 10.1038/ng1180 33. Subramanian A., Tamayo P., Mootha V.K., Mukherjee S., Ebert B.L., Gillette M.A., Paulovich A., Pomeroy S.L., Golub T.R., Lander E.S.. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA (2005) 102 15545-15550. DOI: 10.1073/pnas.0506580102 34. Sonneveld R., Hoenderop J.G., Isidori A.M., Henique C., Dijkman H.B., Berden J.H., Tharaux P.L., van der Vlag J., Nijenhuis T.. Sildenafil Prevents Podocyte Injury via PPAR-γ-Mediated TRPC6 Inhibition. J. Am. Soc. Nephrol. (2017) 28 1491-1505. DOI: 10.1681/ASN.2015080885 35. ‘t Hart D.C., van der Vlag J., Nijenhuis T.. Laminar flow substantially affects the morphology and functional phenotype of glomerular endothelial cells. PLoS ONE (2021) 16. DOI: 10.1371/journal.pone.0251129 36. Mundel P., Reiser J., Zúñiga Mejía Borja A., Pavenstädt H., Davidson G.R., Kriz W., Zeller R.. Rearrangements of the cytoskeleton and cell contacts induce process formation during differentiation of conditionally immortalized mouse podocyte cell lines. Exp. Cell Res. (1997) 236 248-258. DOI: 10.1006/excr.1997.3739 37. Gil C.L., Hooker E., Larrivée B.. Diabetic Kidney Disease, Endothelial Damage, and Podocyte-Endothelial Crosstalk. Kidney Med. (2021) 3 105-115. DOI: 10.1016/j.xkme.2020.10.005 38. Siddiqi F.S., Advani A.. Endothelial-podocyte crosstalk: The missing link between endothelial dysfunction and albuminuria in diabetes. Diabetes (2013) 62 3647-3655. DOI: 10.2337/db13-0795 39. Unnersjö-Jess D., Butt L., Höhne M., Witasp A., Kühne L., Hoyer P.F., Patrakka J., Brinkkötter P.T., Wernerson A., Schermer B.. A fast and simple clearing and swelling protocol for 3D in-situ imaging of the kidney across scales. Kidney Int. (2021) 99 1010-1020. DOI: 10.1016/j.kint.2020.10.039 40. Reiser J., Altintas M.. Podocytes [version 1; peer review: 2 approved]. F1000Research (2016) 5 114. DOI: 10.12688/f1000research.7255.1 41. Nieuwdorp M., Mooij H.L., Kroon J., Atasever B., Spaan J.A., Ince C., Holleman F., Diamant M., Heine R.J., Hoekstra J.B.. Endothelial glycocalyx damage coincides with microalbuminuria in type 1 diabetes. Diabetes (2006) 55 1127-1132. DOI: 10.2337/diabetes.55.04.06.db05-1619 42. Padberg J.S., Wiesinger A., di Marco G.S., Reuter S., Grabner A., Kentrup D., Lukasz A., Oberleithner H., Pavenstädt H., Brand M.. Damage of the endothelial glycocalyx in chronic kidney disease. Atherosclerosis (2014) 234 335-343. DOI: 10.1016/j.atherosclerosis.2014.03.016 43. Kriz W., Shirato I., Nagata M., LeHir M., Lemley K.V.. The podocyte’s response to stress: The enigma of foot process effacement. Am. J. Physiol.-Renal Physiol. (2013) 304 F333-F347. DOI: 10.1152/ajprenal.00478.2012 44. Kriz W., Lemley K.V.. Mechanical challenges to the glomerular filtration barrier: Adaptations and pathway to sclerosis. Pediatr. Nephrol. (2017) 32 405-417. DOI: 10.1007/s00467-016-3358-9 45. Pozzi A., Jarad G., Moeckel G.W., Coffa S., Zhang X., Gewin L., Eremina V., Hudson B.G., Borza D.B., Harris R.C.. Beta1 integrin expression by podocytes is required to maintain glomerular structural integrity. Dev. Biol. (2008) 316 288-301. DOI: 10.1016/j.ydbio.2008.01.022 46. Sison K., Eremina V., Baelde H., Min W., Hirashima M., Fantus I.G., Quaggin S.E.. Glomerular structure and function require paracrine, not autocrine, VEGF-VEGFR-2 signaling. J. Am. Soc. Nephrol. (2010) 21 1691-1701. DOI: 10.1681/ASN.2010030295 47. Den Braanker D.J.W., Maas R.J., Deegens J.K., Yanginlar C., Wetzels J.F.M., van der Vlag J., Nijenhuis T.. Novel in vitro assays to detect circulating permeability factor(s) in idiopathic focal segmental glomerulosclerosis. Nephrol. Dial. Transplant. (2021) 36 247-256. DOI: 10.1093/ndt/gfaa211 48. Glasser R.J., Velosa J.A., Michael A.F.. Experimental model of focal sclerosis. I. Relationship to protein excretion in aminonucleoside nephrosis. Lab. Investig. (1977) 36 519-526. PMID: 865079 49. Lee V.W., Harris D.C.. Adriamycin nephropathy: A model of focal segmental glomerulosclerosis. Nephrology (2011) 16 30-38. DOI: 10.1111/j.1440-1797.2010.01383.x 50. Ding F., Wickman L., Wang S.Q., Zhang Y., Wang F., Afshinnia F., Hodgin J., Ding J., Wiggins R.C.. Accelerated podocyte detachment and progressive podocyte loss from glomeruli with age in Alport Syndrome. Kidney Int. (2017) 92 1515-1525. DOI: 10.1016/j.kint.2017.05.017 51. Mayadas T.N., Rosetti F., Ernandez T., Sethi S.. Neutrophils: Game changers in glomerulonephritis?. Trends Mol. Med. (2010) 16 368-378. DOI: 10.1016/j.molmed.2010.06.002 52. Lim A.K., Tesch G.H.. Inflammation in diabetic nephropathy. Mediat. Inflamm. (2012) 2012 146154. DOI: 10.1155/2012/146154
--- title: Pretreatment Frequency of Circulating Th17 Cells and FeNO Levels Predicted the Real-World Response after 1 Year of Benralizumab Treatment in Patients with Severe Asthma authors: - Yuuki Sandhu - Norihiro Harada - Hitoshi Sasano - Sonoko Harada - Shoko Ueda - Tomohito Takeshige - Yuki Tanabe - Ayako Ishimori - Kei Matsuno - Sumiko Abe - Tetsutaro Nagaoka - Jun Ito - Asako Chiba - Hisaya Akiba - Ryo Atsuta - Kenji Izuhara - Sachiko Miyake - Kazuhisa Takahashi journal: Biomolecules year: 2023 pmcid: PMC10046637 doi: 10.3390/biom13030538 license: CC BY 4.0 --- # Pretreatment Frequency of Circulating Th17 Cells and FeNO Levels Predicted the Real-World Response after 1 Year of Benralizumab Treatment in Patients with Severe Asthma ## Abstract Benralizumab treatment reduces exacerbations and improves symptom control and quality of life in patients with severe eosinophilic asthma. However, the determination of biomarkers that predict therapeutic effectiveness is required for precision medicine. Herein, we elucidated the dynamics of various parameters before and after treatment as well as patient characteristics predictive of clinical effectiveness after 1 year of benralizumab treatment in severe asthma in a real-world setting. Thirty-six patients with severe asthma were treated with benralizumab for 1 year. Lymphocyte subsets in peripheral blood samples were analyzed using flow cytometry. Treatment effectiveness was determined based on the ACT score, forced expiratory volume in 1 s (FEV1), and the number of exacerbations. Benralizumab provided symptomatic improvement in severe asthma. Benralizumab significantly decreased peripheral blood eosinophil and basophil counts and the frequencies of regulatory T cells (Tregs), and increased the frequencies of Th2 cells. To our knowledge, this is the first study to show benralizumab treatment increasing circulating Th2 cells and decreasing circulating Tregs. Finally, the ROC curve to discriminate patients who achieved clinical effectiveness of benralizumab treatment revealed that the frequency of circulating Th17 cells and FeNO levels might be used as parameters for predicting the real-world response of benralizumab treatment in patients with severe asthma. ## 1. Introduction Asthma affects over 300 million people worldwide, and there are more than 10 million people with asthma in Japan, of which approximately $10\%$ of them have intractable asthma with symptoms that cannot be controlled by existing treatments. The medical costs for these patients account for the majority of the total medical costs associated with asthma. Asthma is one of the most common chronic diseases characterized by variable airflow limitation and bronchial hyperresponsiveness [1,2,3]. Of the various asthma phenotypes/endotypes, eosinophilic asthma affects more than $50\%$ of patients with asthma [4,5]. Eosinophilic asthma responds well to treatment with inhaled corticosteroids (ICS) because steroids induce the apoptosis of eosinophils [6,7]. However, the effect of ICS on asthma is limited because some patients who develop the most severe clinical phenotype of eosinophilic asthma also have steroid-resistant refractory asthma. Biologics that are expected to have an additional effect on existing treatments for intractable asthma have been developed, with four of these agents being approved for use in Japan [8]. Among them, benralizumab, a humanized anti-IL-5 receptor α subunit monoclonal IgG1 antibody, depletes eosinophils by antibody-dependent cellular cytotoxicity. A major study evaluating benralizumab in patients with moderate to severe asthma found reduced exacerbations, improved lung function, and reduced use of oral corticosteroids (OCS) [9,10,11]. Benralizumab therapy reduces eosinophils in the peripheral blood and airway mucosa, especially by depleting eosinophils in the peripheral blood [11,12]. Based on previous studies, benralizumab has been used in patients with refractory asthma whose asthma symptoms cannot be controlled by existing therapies. Although high peripheral blood eosinophil levels are known to be a biomarker for predicting the effects of benralizumab treatment, a certain number of patients with asthma have been shown to respond to benralizumab, even with low levels of peripheral blood eosinophils [13,14]. Therefore, new biomarkers are needed to help predict the effectiveness of benralizumab therapy. Because biologics are expensive, extracting response cases before administration is important from both the perspective of health economics and future stratified medicine. In this study, we investigated whether benralizumab treatment was associated with peripheral blood immunocompetent cells and conducted an extensive study to identify important biomarkers for use in appropriately elucidating the effectiveness of benralizumab treatment. ## 2.1. Study Subjects This study was a prospective observational study that enrolled severe asthma patients with newly prescribed benralizumab from March 2018 to May 2019. Patients who had severe asthma and were aged 20 years or older, whose asthma symptoms and asthma exacerbations requiring OCS could not be controlled by the existing treatment options despite treatment with high-dose ICS plus long-acting β2 agonists with another controller, and who required benralizumab treatment in the insurance medical treatment were recruited from our outpatient clinic at Juntendo University Hospital (Tokyo, Japan). Asthma was diagnosed based on a clinical history of episodic symptoms with airflow limitation and by either a variation in pulmonary function monitored by forced expiratory volume in 1 s (FEV1) or peak expiratory flow according to the Global Initiative for Asthma (GINA) guidelines [15]. Patients with any of the following criteria were excluded: [1] a diagnosis of eosinophilic granulomatosis with polyangiitis, interstitial pneumonia, infectious disease, or cancer; [2] those administering other antibody preparations; [3] cases that were judged as inappropriate by the study investigators; [4] cases under treatment with omalizumab and mepolizumab with <1 month of the last dose and cases under treatment with other biologics. The present study was reviewed and approved by the Juntendo University Research Ethics Committee (Tokyo, Japan). Written informed consent was obtained from each patient prior to participation in the study. The study was registered in the UMIN Clinical Trial Registry (UMIN000031905) on 23 March 2018 (http://www.umin.ac.jp/ (accessed on 25 March 2022)). The asthma control test (ACT), pulmonary function tests, oscillometry (also known as the forced oscillation technique), measurement of fractional exhaled nitric oxide (FeNO) levels, and blood sampling were performed at the date of initial administration of benralizumab, 4 months, 8 months, and 1 year after administration. FeNO levels were measured in accordance with the American Thoracic Society recommendations at a constant flow of 0.05 L/s against an expiratory resistance of 20 cm water with an electrochemical hand-held NO analyzer (NIOX VERO®; Aerocrine AB, Solna, Sweden). ## 2.2. Definition for Responders Patients were classified as responders and super-responders according to changes in ACT score, lung function, and asthma exacerbations with reference to previous studies [16,17,18,19,20,21,22]. A responder with benralizumab treatment was defined as meeting two of the following three criteria after 1 year of treatment with benralizumab without a significant deterioration in any other criterion: [1] Improvement in ACT score of at least 3 points (including patients who achieved an ACT score of 25 points) and an increase in ACT score of at least 3 points, which was previously suggested as the minimal clinically important difference [23,24]. [ 2] Reduction in the number of asthma exacerbations (including patients who had no exacerbations before and after treatment). [ 3] Improvement in FEV1 of at least 100 mL [22,25]. The following criteria were considered to be associated with significant deterioration: (a) a decreased ACT score of at least 3 points; (b) an increase in the number of exacerbations; (c) a decrease in FEV1 of at least 100 mL. A super-responder was also defined as meeting all three of the above criteria without showing any significant deterioration. ## 2.3. Quantification of Circulating Lymphocyte Frequency Flow cytometry analysis was conducted as previously described [26,27]. Briefly, peripheral venous blood samples were collected in heparin-containing tubes, and PBMCs (3 × 106/well) were purified by density-gradient centrifugation using Ficoll–Paque Plus solution (Cytiva, Tokyo, Japan). The cells were stained with different combinations of the appropriate antibodies for 30 min at 4 °C. The following surface marker antibodies were used in this study: anti-CD3-APC-H7, anti-CD4-FITC, anti-CD19-FITC, anti-CD56-PE-CF594, anti-CD117 (c-Kit)-PE-CF594 (BD Biosciences, San Jose, CA, USA), anti-T cell antigen receptor (TCR)-Pan-γδ-FITC, anti-TCR-Pan-γδ-PE (Beckman Coulter, Miami, FL, USA), anti-BDCA2-FITC, anti-CD1a-FITC, anti-CD3-FITC, anti-CD11c-FITC, anti-CD14-FITC, anti-CD25-PE, anti-CD34-FITC, anti-CD123-FITC, anti-CD127 (IL-7Rα)-Brilliant Violet 605, anti-CD161-PerCPCy5.5, anti-CD183 (CXCR3)-APC, anti-CD194 (CCR4)-Brilliant Violet 510, anti-CD196 (CCR6)-PerCPCy5.5, anti-CD294 (CRTH2)-Brilliant Violet 421, anti-FCɛR1-FITC, anti-Vα7.2- Brilliant Violet 605 (BioLegend, San Diego, CA, USA), and anti-hCD1d tetramer loaded with PBS-57-APC (NIH tetramer core facility at Emory University). Negative lineage markers (Lin−) were defined as CD1a−, CD3−, CD11c−, CD14−, CD19−, CD34−, TCRγδ−, CD123−, BDCA2−, and FCɛR1−. Th1 cells were identified as CD3+, CD4+, CCR4−, CCR6−, and CXCR3+ cells; Th2 cells as CD3+, CD4+, CCR4+, CCR6−, and CXCR3− cells; Th17 cells as CD3+, CD4+, CCR4+, CCR6+, and CXCR3− cells; Tregs as CD3+, CD4+, CD25+, and CD127− cells; natural killer (NK) T (NKT) cells as CD3+ and CD1d/PBS-57 tetramer+ cells; γδT cells as CD3+ and TCRγδ+ cells; mucosal-associated invariant T (MAIT) cells as CD3+, Vα7.2 TCR+, and CD161high cells; NK cells as CD3− and CD56+ cells; ILC1 as Lin−, CD127+, CD161+, CD117−, and CRTH2− cells; ILC2 as Lin−, CD127+, CD161+, and CRTH2+ cells; ILC3 as Lin−, CD127+, CD161+, CD117+, and CRTH2− cells. Dead cells were identified by using the Zombie Fixable Viability Kit (BioLegend), followed by doublet exclusion on forward scatter and side scatter. After overnight fixation, the cells were analyzed by using a fluorescence-activated cell sorting (FACS) LSRFortessa cell analyzer (BD Biosciences). The FACS data were evaluated using FlowJo software (version 9; BD Biosciences). ## 2.4. Quantification of Serum Cytokine and Chemokine Levels The sera of patients were collected after density-gradient centrifugation of blood samples, frozen at −80 °C, and assayed using the MILLIPLEX multiplex assay following the manufacturer’s instructions (Merck Millipore, Burlington, MA, USA). The assay-working range was determined between the lower limit of quantification and the upper limit of quantification (Supplementary Table S1). Serum periostin levels were measured using ELISA (Shino test, Kanagawa, Japan) as previously described [28]. Serum tenascin-C and regulated on activation normal T cell expressed and secreted (RANTES) were simultaneously quantified in thawed serum using the human tenascin-C ELISA kit (IBL, Gunma, Japan) and the human RANTES ELISA kit (R&D Systems, Minneapolis, MN, USA), respectively. ## 2.5. Statistical Analysis Sample normality was examined using the D’Agostino–Pearson test. Differences in parameters between populations were analyzed for significance using Welch’s t-test, the paired t test, the Mann–Whitney U test, Wilcoxon’s signed-rank test, and Fisher’s exact test as appropriate. Comparisons between multiple groups were made by Friedman’s test with Dunn’s multiple comparisons test. ROC curve analyses were performed to differentiate between responders and non-responders to benralizumab. For correlation between variables, the Pearson’s correlation coefficient and Spearman’s rank correlation coefficient were used where appropriate. Differences were statistically significant when p values were <0.05. Statistical analyses were performed using GraphPad Prism version 8 software (GraphPad Software, San Diego, CA, USA). ## 3.1. Baseline Characteristics Thirty-six patients with severe asthma that was uncontrolled by existing treatment regimens, including who had an ACT score less than 20 points on conventional therapy or who had at least one exacerbation requiring oral corticosteroids per year, were enrolled and treated with benralizumab. Adverse events following the initial administration of benralizumab were observed in three patients, including a patient who had anaphylaxis, a patient who was suspected of having anaphylaxis, and a patient who had fever. In addition to the exclusion of the three aforementioned patients who developed adverse events, one patient withdrew consent and one patient had to discontinue the study due to exacerbation of allergic bronchial pulmonary aspergillosis and lower respiratory bacterial infection. Consequently, 31 patients were enrolled for the entire duration of the study. The baseline characteristics are shown in Table 1 and Table 2. The mean (±standard deviation) age of the patients was 54.3 ± 13.5 years (Table 1). Female patients, patients on regular OCS, patients who were treated with omalizumab, and those treated with mepolizumab prior to the study were 22 ($71\%$), 3 ($10\%$), 4 ($13\%$), and 14 ($45\%$), respectively (Table 1). The median daily dose of ICS was 1000 µg, inclusive of four patients who could not administer high-dose ICS due to hoarseness side-effects. Of the three patients with regular OCS, one received 1 mg of daily PSL and the other two received 5 mg of daily PSL. The median duration of asthma and duration of prior treatment with biologics (omalizumab or mepolizumab) was 16 years (interquartile range 8–26) and 604.5 days (426.5–690.0), respectively (Table 1). The effectiveness of previous biologics, except for two patients, was evaluated after at least 4 months of biologics treatment. Of the two patients that switched to benralizumab within 4 months of previous biologics treatment, one was discontinued omalizumab due to a side-effect, and the other was switched to benralizumab immediately after marketing because he had a request and did not recognize an effect of mepolizumab treatment. Abbreviations for all tables: ACT, asthma control test; ABPA, allergic bronchopulmonary aspergillosis; AERD, aspirin-exacerbated respiratory disease; BMI, body mass index; FeNO, fractional exhaled nitric oxide; FP, fluticasone propionate; FVC, forced vital capacity; FEV$1\%$, forced expiratory volume in 1 s/forced vital capacity; FEV1, forced expiratory volume in 1 s; ICS, inhaled corticosteroid; IFN-γ, interferon-gamma; IgE, immunoglobulin E; IL, interleukin; ILC, innate lymphoid cell; MAIT, mucosal associated invariant T; MCP, monocyte chemotactic protein; MIP, macrophage inflammatory protein; NK, natural killer; γδT, gamma delta T; NKT, natural killer T; RANTES, regulated on activation normal T cell expressed and secreted; SD, standard deviation; Th, helper T; Treg, regulatory T. The mean FEV$1\%$, which was calculated as FEV1/forced vital capacity (FVC), and the median (interquartile) peripheral blood eosinophil counts were 70.8 ± $17.6\%$ and 80/μL (32–313), respectively (Table 2). ## 3.2. Changes in Each Parameter 1 Year after Benralizumab Treatment After one year of benralizumab treatment, 18 ($58\%$) of the 31 patients showed improved ACT scores of at least 3 points (known as the minimal clinically important difference [24]) or achieved total control. The comparison of initial data with data after 1 year of treatment with benralizumab was analyzed in 30 cases, excluding one case that was discontinued due to the worsening of asthma symptoms 6 months after the initiation of benralizumab treatment. The number of asthma exacerbations and unscheduled visits from worsening asthma decreased significantly, although the number of hospitalizations did not change. Benralizumab treatment significantly increased the ACT score, percent predicted FEV1 (%FEV1), FEV$1\%$ in pulmonary function parameters, and serum eotaxin-1 levels, but significantly decreased peripheral blood eosinophil and basophil counts (Table 3 and Figure 1). Although the serum levels of IL-5 did not change after 1 year of benralizumab treatment, they were significantly increased at 4 months and then significantly decreased at 12 months compared to 4 months (Table 3 and Figure 1). Because serum IL-4, IL-13, and macrophage inflammatory protein (MIP)-1α levels were below the detection limit, they were excluded from the analysis. Next, we examined the frequency of PBMCs in peripheral bloods using flow cytometry. The gating strategy for the PBMCs is shown in Supplementary Figure S1. The frequencies of Th cells, ILCs, and MAIT cells were demonstrated by their ratios to CD3+ and CD4+ cells, Lin− CD127+ and CD161+ cells, and CD3+ cells, respectively. The frequencies of NK cells, NKT cells, and γδT cells were also shown by their ratios to lymphocytes. Flow cytometric analysis of peripheral blood showed that 1 year of benralizumab treatment significantly increased the frequencies of Th2 cells and significantly decreased the frequencies of Tregs and NKT cells (Table 3 and Figure 2). These findings suggest that the 1-year treatment regimen with benralizumab for patients with severe asthma reduces peripheral blood eosinophils, increases Th2 cells and serum eotaxin-1 levels, and transiently increases serum IL-5 levels. Because these changes in each parameter after 1 year of benralizumab treatment may have been influenced by omalizumab and mepolizumab used before benralizumab treatment, we divided, into two groups, 17 patients previously treated with biologics (14 patients with mepolizumab, 3 patients treated with omalizumab) and 13 patients who were not previously treated with biologics. ACT score and FEV$1\%$ were significantly improved only in patients without previous use of biologics, and increased Th2 cells and decreased Tregs and NKT cells were significant only in patients with previous use of biologics (Supplemental Table S2). These findings showed a similar tendency in each other group, suggesting the possibility that the decrease in the number of patients due to the division had an effect. On the other hand, due to the splitting, in patients without previous use of biologics, a significant increase in serum IL-5 (although the number of patients who could be measured was small) and a significant decrease in circulating ILC1 were observed (Supplemental Table S2). These findings, at least, suggested that decreased peripheral blood eosinophils, increased Th2 cells, and increased serum eotaxin-1 levels may not be affected by previous use of biologics. ## 3.3. Parameters for Predicting the Effectiveness of Benralizumab in Patients with Severe Asthma We then divided the 31 patients into two subgroups according to their response to benralizumab treatment (Table 4). The number of responders was 15 ($48\%$) and they had a significantly lower average BMI and frequency of ILC3 in peripheral blood, as well as a higher mean frequency of Th17 cells before benralizumab treatment than non-responders (Table 4). Non-responders had more cases of atopic dermatitis than responders (Table 4). ROC curve analysis was used to determine the optimal cut-off values of the frequency of Th17 cells, ILC3, and BMI to discriminate responders from non-responders, with observed areas under the curve of 0.733 ($$p \leq 0.027$$), 0.713 ($$p \leq 0.044$$), and 0.704 ($$p \leq 0.053$$), respectively (Figure 3A). The frequencies of Th17 cells of $4.57\%$ Th cells (sensitivity, $100\%$; specificity, $56.3\%$) and ILC3 of $11.45\%$ ILCs (sensitivity, $73.3\%$; specificity, $62.5\%$) were the best cut-off values for the optimal potential effectiveness of benralizumab treatment using the Youden index [29]. Nine patients ($29\%$) with super-responders showed the highest FeNO levels, frequency of Th17 cells in peripheral blood, and number of asthma exacerbations and unscheduled visits for worsening asthma, and lowest %FEV1, %PEFR, and %MMF before benralizumab treatment (Table 4). The areas under the ROC curves of FeNO levels and the frequency of Th17 cells were 0.856 ($$p \leq 0.002$$) and 0.768 ($$p \leq 0.021$$), respectively (Figure 3B). FeNO levels of 44.0 ppb (sensitivity, $100\%$; specificity, $72.7\%$) and Th17 cell frequencies of $4.77\%$ Th cells (sensitivity, $100\%$; specificity, $54.6\%$) were the best cutoffs for predicting super-responders to benralizumab treatment. Therefore, we postulated that high FeNO levels and high frequencies of Th17 cells before treatment could be used as biomarkers for predicting the effectiveness of benralizumab in the treatment of patients with severe asthma. Because this study included 18 patients with previous use of biologics, a similar analysis was performed on 13 patients without previous use of biologics. The number of responders and super-responders was 8 ($62\%$) and 5 ($38\%$), respectively. Responders had a significantly older age at asthma onset and no cases of atopic dermatitis than non-responders (Supplemental Table S3). Responders and super-responders showed a significantly higher frequency of Th17 cells in peripheral blood than non-responders (Supplemental Table S3). Super-responders had significantly higher FeNO levels, peripheral blood eosinophils and basophils, and serum total IgE levels (Supplemental Table S3). These findings suggested that our results of biomarkers may be less modulated by previous use of biologics. ## 3.4. Association between Type 2 Biomarkers and Non-Type 2 Biomarker Levels with Subject Characteristics in Patients with Severe Asthma Although Th17 cells do not play a central role in type 2 inflammation, a high frequency of Th17 cells was a candidate for predicting the effects of benralizumab-targeting on eosinophils in this study. Thus, we investigated whether type 2 biomarkers, including FeNO, and non-type 2 biomarker levels, including Th17 cells and ILC3, were associated with clinical parameters of asthma. FeNO levels were positively correlated with serum total IgE levels ($r = 0.498$, $$p \leq 0.004$$) and serum periostin levels ($r = 0.515$, $$p \leq 0.003$$), but were negatively correlated with the number of hospitalizations (r = −0.476, $$p \leq 0.008$$), %FEV1 (r = −0.406, $$p \leq 0.023$$), FEV$1\%$ (r = −0.532, $$p \leq 0.002$$), MMF (r = −0.481, $$p \leq 0.006$$), and %MMF (r = −0.561, $$p \leq 0.001$$) (Table 5). The frequency of Th17 cells was positively correlated with the number of unscheduled visits for worsening asthma ($r = 0.416$, $$p \leq 0.020$$) and the frequency of Th2 cells ($r = 0.686$, $p \leq 0.001$), and was negatively correlated with serum IFN-γ levels (r = −0.615, $$p \leq 0.037$$) (Table 5). The frequency of ILC3 was negatively correlated with the frequency of ILC1 (r = −0.458, $$p \leq 0.010$$) (Table 5). Peripheral blood eosinophil counts were positively correlated with basophil counts ($r = 0.743$, $p \leq 0.001$), lymphocyte counts ($r = 0.377$, $$p \leq 0.037$$), serum periostin levels ($r = 0.422$, $$p \leq 0.018$$), serum IFN-γ levels ($r = 0.615$, $$p \leq 0.037$$), and serum MIP-1β levels ($r = 0.438$, $$p \leq 0.014$$), and negatively correlated with the frequency of neutrophils (r = −0.445, $$p \leq 0.012$$) and serum eotaxin-1 levels (r = −0.375, $$p \leq 0.038$$) (Table 5). Serum periostin levels were positively correlated with FeNO levels, eosinophil counts, basophil counts ($r = 0.493$, $$p \leq 0.005$$), and serum MIP-1β levels ($r = 0.373$, $$p \leq 0.039$$), and negatively correlated with FEV$1\%$ (r = −0.500, $$p \leq 0.004$$), MMF (r = −0.459, $$p \leq 0.009$$), %MMF (r = −0.501, $$p \leq 0.004$$), and the frequency of neutrophils (r = −0.394, $$p \leq 0.028$$) (Table 5). These findings suggest that Th17 cells and type 2 biomarkers are not correlated; however, the positive correlation between Th17 and Th2 cells suggested that Th17 cells may be involved in Th2 inflammation. ## 4. Discussion One year of benralizumab treatment in patients with severe asthma improved their ACT scores and FEV$1\%$ values, and reduced the number of asthma exacerbations as well as unscheduled hospital visits. Benralizumab treatment also decreased peripheral blood eosinophil and basophil counts, and increased serum eotaxin-1 levels over the 1-year period and transient serum IL-5 levels up to 4 months. To our knowledge, this is the first study to show that benralizumab treatment increases circulating Th2 cells and decreases circulating Tregs after 1 year, and that the high frequency of circulating Th17 cells and high FeNO levels might predict the real-world response of benralizumab treatment in patients with severe asthma. Similar to this real-world study, previous reports have shown that benralizumab treatment suppresses asthma symptoms and exacerbations of asthma for a year and improves airflow limitation, including FEV1 [9,10,11]. The decrease in basophils and eosinophils with benralizumab treatment has already been explained by the expression of IL-5Rα in basophils [30]. In the aforementioned study, peripheral blood eosinophils were absent in all patients after benralizumab treatment, while basophils were still present, even though both eosinophils and basophils expressed IL-5Rα. However, in all nine super-responders herein, peripheral blood basophils were absent after 1 year of benralizumab treatment. Elevated serum eotaxin-1 levels, a transient elevation of serum IL-5 levels, and increased peripheral blood Th2 cells were assumed to be the result of feedback due to peripheral blood eosinophil depletion. Although the involvement of IL-4 was suspected, which can induce Th2 cells and eotaxin production, serum IL-4 and IL-13 were undetectable in this study, and the feedback mechanism of how depleted peripheral blood eosinophils increased Th2 cells remains unclear. Regarding eotaxin-1, similar findings have been suggested in two previous reports that showed an increase in serum levels of eotaxin-1 and eotaxin-2 in patients with asthma following benralizumab (100 mg or 200 mg) treatment for 8 weeks or benralizumab (200 mg) treatment for 52 weeks [31,32]. The decrease in peripheral blood Tregs and NKT cells was thought to be due to the stabilization of airway inflammation during benralizumab treatment, which reduced the need for Tregs to suppress inflammation, and which might reduce CD1d-expressed dendritic cells that activate NKT cells, respectively. The findings in this study at least suggest that Tregs may not be involved in the effectiveness of benralizumab treatment. A previous real-world study in Italy showed that the peripheral percentages of NKT-like cells significantly decreased in 20 patients after 6 months of mepolizumab treatment, but not in 8 patients after benralizumab treatment [33]. This difference from our study may be attributed to the different numbers between 8 and 31 patients with benralizumab treatment. However, further case accumulation and further studies are required to confirm these assumptions. When responders to benralizumab treatment were defined as meeting 2 of the 3 criteria, which were the improvement in ACT score, FEV1, and the number of exacerbations, and super-responders were defined as meeting 3 of the 3 above criteria, a high frequency of Th17 cells and low frequency of ILC3 before benralizumab treatment in responders and high FeNO levels and high frequency of Th17 cells in super-responders could be biomarkers for predicting the effectiveness after 1 year of benralizumab treatment in patients with severe asthma. Eighteen patients, more than half of whom participated in the study, had received biologics as pretreatment, which is one of the limitations of this study. Indeed, super-responders in biologics-naïve patients had high levels of type 2 inflammatory markers including higher peripheral blood eosinophil and basophil counts, serum total IgE levels, and FeNO levels, similar to previous reports. However, it was suggested that the high frequency of circulating Th17 cells before benralizumab treatment in responders and high FeNO levels and a high frequency of circulating Th17 cells in super-responders may be biomarkers for predicting the effectiveness after 1 year of benralizumab treatment in the 13 patients without biologics as pretreatment, as in the whole population of this study. A retrospective analysis of the real-world setting in the United Kingdom (UK) defined either a $50\%$ or greater reduction in asthma exacerbation rate or a $50\%$ reduction in OCS dose as a benralizumab responder, responders were 112 of 130 patients ($86\%$), and the FeNO value of the responder was significantly higher than that of the non-responders, suggesting that FeNO predicts the responder [14]. Additionally, Watanabe et al. reported that benralizumab responders were 16 of 21 patients ($76\%$) after the 24-week treatment with benralizumab, and that baseline peripheral blood eosinophil counts of 100/μL and FeNO levels of 40 ppb were the best cutoffs for predicting responders for benralizumab treatment [34]. Taken together, our results suggested that FeNO is an emerging candidate for effect prediction biomarkers other than eosinophils at least. On the other hand, parameters using flow cytometry such as the high frequency of circulating Th17 cells are not available as clinical biomarkers in practice. Although peripheral blood Th17 cells are thought to play a central role in non-type 2 inflammation, they have been shown to up-regulate Th2-cell-mediated eosinophilic airway inflammation in a mouse model [35]. Furthermore, IL-4- and IL-17-producing Th2/Th17 cells have been reported to be associated with the co-expression of GATA-binding protein 3 and retinoic acid receptor-related orphan receptor γt, which may be carriers of type 2 inflammation [36,37]. Therefore, it is speculated that high FeNO and Th17 levels as biomarkers probably reflect type 2 inflammation. In this study, eosinophils did not function as biomarkers for predicting the effectiveness of benralizumab treatment, because many of the enrolled patients might have had low peripheral blood eosinophils due to previous treatment with OCS or biologics [38,39]. Although previous reports have shown that eosinophils are useful biomarkers, there are a certain number of asthma patients who respond to benralizumab even with low peripheral blood eosinophil counts [13,14]. Nonetheless, further studies are needed to investigate whether Th17 can be used as a type 2 biomarker to predict the therapeutic effectiveness of benralizumab in patients with severe asthma because the types of cytokines produced by peripheral blood Th17 cells analyzed in this study were unclear. Furthermore, non-responders had a higher frequency of ILC3 and higher BMI than responders. ILC3 has been shown to be associated with obesity and asthma in a mouse model [40,41]. In addition, it has been shown that CD69+ ILC3 in peripheral blood correlates with BMI in patients with asthma [26]. These findings suggest that benralizumab is less effective for the treatment of patients who are obese and have asthma. There were also more patients with atopic dermatitis in non-responders than responders to benralizumab, suggesting that benralizumab is less effective for the treatment of asthmatics with atopic dermatitis. In this study, two patients were suspected of having anaphylaxis and one had anaphylaxis as adverse events; however, the patient with anaphylaxis improved immediately and was switched from mepolizumab treatment. Nonetheless, after anaphylaxis due to benralizumab, the patient developed anaphylaxis after being re-initiated on mepolizumab [42]. It is, therefore, necessary to keep in mind the risk of anaphylaxis from the use of biologics. Although peripheral blood eosinophils were depleted to zero in all patients 4 months after benralizumab treatment in this study, eosinophils were subsequently elevated in two patients ($6.5\%$). One patient had worsening asthma during the 1 year of observation, and the other had worsening asthma after the 1 year of benralizumab treatment. Drug-neutralizing antibodies to benralizumab may have developed in these patients; however, these antibodies were not characterized. A real-world study in the UK also demonstrated the development of detectable blood eosinophil counts in keeping with the presumed development of drug-neutralizing antibodies in five patients ($3.8\%$) [14]. Although anti-drug antibodies could work without impacting the treatment efficacy, especially if they are not neutralizing, the characterization of drug-neutralizing antibodies is an issue that needs to be considered in the future. Finally, the limitation of this study was that it was a single-center, single-arm, open-label, observational study with a small sample size. In particular, the small sample size was one of the important limitations, as it can affect the numerical values of various parameters. Moreover, the components in the expert consensus-based criteria of the clinical responders and super-responders in this study have not yet had a consensus, did not remain completely consistent with previous reports, and will likely evolve and change over time. Super-responders had a lower FEV1 compared to the other patients, suggesting the possibility that patients with a lower FEV1 and more likely to be evaluated as super-responders were a limitation of this study. This study in a real-world setting, unlike a phase III clinical trial conducted by a pharmaceutical company, included patients who had asthma symptoms that could not be controlled with existing treatments, but some patients had no exacerbation before benralizumab treatment. One of the limitations of this study is that even such patients were included in the responders if they had no asthma exacerbations during the 1 year of benralizumab treatment. Furthermore, although patients with low eosinophil count immediately before benralizumab treatment were also included in this study, all patients had at least a maximum value of eosinophil counts of 150 or more during all outpatient visits. However, all patients in this study underwent extensive assessment, including treatment options other than biologics, before initiation of benralizumab. Therefore, we feel confident that the clinical improvements observed following the addition of benralizumab to the regimen of patients with uncontrolled severe asthma reflects the effectiveness of benralizumab and does not reflect any optimization of previous background therapy. This study also showed some emerging candidates for effect prediction biomarkers for benralizumab treatment, but the identification of prognostic biomarkers needs to be confirmed in further studies for validation. ## 5. Conclusions This study showed that benralizumab treatment increased ACT scores, FEV$1\%$ levels, peripheral blood Th2 cells, and eotaxin-1 levels over 1 year, as well as transiently increased IL-5 levels up to 4 months, and decreased the number of asthma exacerbations, unscheduled visits, peripheral blood eosinophils, basophils, and Tregs in patients with uncontrolled severe asthma. We have provided the first report in a real-world setting that showed that 1 year of benralizumab treatment increased circulating Th2 cells and decreased circulating Tregs, and that circulating Th17 cells, ILC3, and FeNO levels might predict the effectiveness of benralizumab in the treatment of patients with severe asthma. Nevertheless, additional studies are needed to investigate whether these parameters have a broader purpose for use in the pathophysiology and treatment or management of asthma. ## References 1. Papi A., Brightling C., Pedersen S.E., Reddel H.K.. Asthma. Lancet (2018) 391 783-800. DOI: 10.1016/S0140-6736(17)33311-1 2. Papi A., Ryan D., Soriano J.B., Chrystyn H., Bjermer L., Rodríguez-Roisin R., Dolovich M.B., Harris M., Wood L., Batsiou M.. Relationship of Inhaled Corticosteroid Adherence to Asthma Exacerbations in Patients with Moderate-to-Severe Asthma. J. Allergy Clin. Immunol. Pract. (2018) 6 1989-1998.e3. DOI: 10.1016/j.jaip.2018.03.008 3. Borish L., Culp J.A.. Asthma: A syndrome composed of heterogeneous diseases. Ann. Allergy Asthma Immunol. (2008) 101 1-8. DOI: 10.1016/S1081-1206(10)60826-5 4. Bousquet J., Chanez P., Lacoste J.Y., Barnéon G., Ghavanian N., Enander I., Venge P., Ahlstedt S., Simony-Lafontaine J., Godard P.. Eosinophilic inflammation in asthma. N. Engl. J. Med. (1990) 323 1033-1039. DOI: 10.1056/NEJM199010113231505 5. Schleich F., Brusselle G., Louis R., Vandenplas O., Michils A., Pilette C., Peche R., Manise M., Joos G.. Heterogeneity of phenotypes in severe asthmatics. The Belgian Severe Asthma Registry (BSAR). Respir. Med. (2014) 108 1723-1732. DOI: 10.1016/j.rmed.2014.10.007 6. Woodruff P.G., Modrek B., Choy D.F., Jia G., Abbas A.R., Ellwanger A., Arron J.R., Koth L.L., Fahy J.V.. T-helper type 2-driven inflammation defines major subphenotypes of asthma. Am. J. Respir. Crit. Care Med. (2009) 180 388-395. DOI: 10.1164/rccm.200903-0392OC 7. Zhang X., Moilanen E., Kankaanranta H.. Enhancement of human eosinophil apoptosis by fluticasone propionate, budesonide, and beclomethasone. Eur. J. Pharmacol. (2000) 406 325-332. DOI: 10.1016/S0014-2999(00)00690-7 8. Harada N., Ito J., Takahashi K.. Clinical effects and immune modulation of biologics in asthma. Respir. Investig. (2021) 59 389-396. DOI: 10.1016/j.resinv.2021.03.003 9. Bleeker E.R., FitzGerald J.M., Chanez P., Papi A., Weinstein S.F., Barker P.. Efficacy and safety of benralizumab for patients with severe asthma uncontrolled with high-dosage inhaled corticosteroids and long-acting beta2-agonists (SIROCCO): A randomised, multicentre, placebo-controlled phase 3 trial. Lancet (2016) 388 2115-2127. DOI: 10.1016/S0140-6736(16)31324-1 10. FitzGerald J.M., Bleecker E.R., Nair P., Korn S., Ohta K., Lommatzsch M., Ferguson G.T., Busse W.W., Barker P., Sproule S.. Benralizumab, an anti-interleukin-5 receptor alpha monoclonal antibody, as add-on treatment for patients with severe, uncontrolled, eosinophilic asthma (CALIMA): A randomised, double-blind, placebo-controlled phase 3 trial. Lancet (2016) 388 2128-2141. DOI: 10.1016/S0140-6736(16)31322-8 11. Nair P., Wenzel S., Rabe K.F., Bourdin A., Lugogo N.L., Kuna P., Barker P., Sproule S., Ponnarambil S., Goldman M.. Oral Glucocorticoid-Sparing Effect of Benralizumab in Severe Asthma. N. Engl. J. Med. (2017) 376 2448-2458. DOI: 10.1056/NEJMoa1703501 12. Sehmi R., Lim H.F., Mukherjee M., Huang C., Radford K., Newbold P., Boulet L.-P., Dorscheid D., Martin J.G., Nair P.. Benralizumab attenuates airway eosinophilia in prednisone-dependent asthma. J. Allergy Clin. Immunol. (2018) 141 1529-1532.e8. DOI: 10.1016/j.jaci.2018.01.008 13. Goldman M., Hirsch I., Zangrilli J.G., Newbold P., Xu X.. The association between blood eosinophil count and benralizumab efficacy for patients with severe, uncontrolled asthma: Subanalyses of the Phase III SIROCCO and CALIMA studies. Curr. Med. Res. Opin. (2017) 33 1605-1613. PMID: 28644104 14. Kavanagh J.E., Hearn A.P., Dhariwal J., D’Ancona G., Douiri A., Roxas C., Fernandes M., Green L., Thomson L., Nanzer A.M.. Real-World Effectiveness of Benralizumab in Severe Eosinophilic Asthma. Chest (2021) 159 496-506. DOI: 10.1016/j.chest.2020.08.2083 15. Global Strategy for Asthma Management and Prevention. (2006) 16. Abdo M., Watz H., Veith V., Kirsten A.-M., Biller H., Pedersen F., Von Mutius E., Kopp M.V., Hansen G., Waschki B.. Small airway dysfunction as predictor and marker for clinical response to biological therapy in severe eosinophilic asthma: A longitudinal observational study. Respir. Res. (2020) 21 278. DOI: 10.1186/s12931-020-01543-5 17. Drick N., Seeliger B., Welte T., Fuge J., Suhling H.. Anti-IL-5 therapy in patients with severe eosinophilic asthma-clinical efficacy and possible criteria for treatment response. BMC Pulm Med. (2018) 18. DOI: 10.1186/s12890-018-0689-2 18. Eger K., Kroes J.A., Ten Brinke A., Bel E.H.. Long-Term Therapy Response to Anti-IL-5 Biologics in Severe Asthma-A Real-Life Evaluation. J. Allergy Clin. Immunol. Pract. (2021) 9 1194-1200. DOI: 10.1016/j.jaip.2020.10.010 19. Hamada K., Oishi K., Murata Y., Hirano T., Matsunaga K.. Feasibility of Discontinuing Biologics in Severe Asthma: An Algorithmic Approach. J. Asthma Allergy (2021) 14 1463-1471. DOI: 10.2147/JAA.S340684 20. Mümmler C., Munker D., Barnikel M., Veit T., Kayser M.Z., Welte T., Behr J., Kneidinger N., Suhling H., Milger K.. Dupilumab Improves Asthma Control and Lung Function in Patients with Insufficient Outcome During Previous Antibody Therapy. J. Allergy Clin. Immunol. Pract. (2021) 9 1177-1185.e4. DOI: 10.1016/j.jaip.2020.09.014 21. Kallieri M., Zervas E., Fouka E., Porpodis K., Mitrova M.H., Tzortzaki E., Makris M., Ntakoula M., Papaioannou A.I., Lyberopoulos P.. RELIght: A two-year REal-LIfe study of mepolizumab in patients with severe eosinophilic asTHma in Greece: Evaluating the multiple components of response. Allergy (2022) 77 2848-2852. DOI: 10.1111/all.15382 22. Liu M.C., Chipps B., Munoz X., Devouassoux G., Bergna M., Smith S.G., Price R.G., Galkin D.V., Azmi J., Mouneimne D.. Benefit of switching to mepolizumab from omalizumab in severe eosinophilic asthma based on patient characteristics. Respir. Res. (2021) 22 144. DOI: 10.1186/s12931-021-01733-9 23. Nathan R.A., Sorkness C.A., Kosinski M., Schatz M., Li J.T., Marcus P., Murray J.J., Pendergraft T.B.. Development of the asthma control test: A survey for assessing asthma control. J. Allergy Clin. Immunol. (2004) 113 59-65. DOI: 10.1016/j.jaci.2003.09.008 24. Schatz M., Kosinski M., Yarlas A.S., Hanlon J., Watson M.E., Jhingran P.. The minimally important difference of the Asthma Control Test. J. Allergy Clin. Immunol. (2009) 124 719-723.e1. DOI: 10.1016/j.jaci.2009.06.053 25. Tepper R.S., Wise R.S., Covar R., Irvin C.G., Kercsmar C.M., Kraft M., Liu M.C., O’Connor G.T., Peters S.P., Sorkness R.. Asthma outcomes: Pulmonary physiology. J. Allergy Clin. Immunol. (2012) 129 S65-S87. DOI: 10.1016/j.jaci.2011.12.986 26. Ishimori A., Harada N., Chiba A., Harada S., Matsuno K., Makino F., Ito J., Ohta S., Ono J., Atsuta R.. Circulating activated innate lymphoid cells and mucosal-associated invariant T cells are associated with airflow limitation in patients with asthma. Allergol. Int. (2017) 66 302-309. DOI: 10.1016/j.alit.2016.07.005 27. Basdeo S.A., Cluxton D., Sulaimani J., Moran B., Canavan M., Orr C., Veale D.J., Fearon U., Fletcher J.M.. Ex-Th17 (Nonclassical Th1) Cells Are Functionally Distinct from Classical Th1 and Th17 Cells and Are Not Constrained by Regulatory T Cells. J. Immunol. (2017) 198 2249-2259. DOI: 10.4049/jimmunol.1600737 28. Okamoto M., Hoshino T., Kitasato Y., Sakazaki Y., Kawayama T., Fujimoto K., Ohshima K., Shiraishi H., Uchida M., Ono J.. Periostin, a matrix protein, is a novel biomarker for idiopathic interstitial pneumonias. Eur. Respir. J. (2011) 37 1119-1127. DOI: 10.1183/09031936.00059810 29. Youden W.J.. Index for rating diagnostic tests. Cancer (1950) 3 32-35. DOI: 10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3 30. Kolbeck R., Kozhich A., Koike M., Peng L., Andersson C.K., Damschroder M.M., Reed J.L., Woods R., Dall’Acqua W.W., Stephens G.L.. MEDI-563, a humanized anti-IL-5 receptor alpha mAb with enhanced antibody-dependent cell-mediated cytotoxicity function. J. Allergy Clin. Immunol. (2010) 125 1344-1353.e2. DOI: 10.1016/j.jaci.2010.04.004 31. Pham T.H., Damera G., Newbold P., Ranade K.. Reductions in eosinophil biomarkers by benralizumab in patients with asthma. Respir Med. (2016) 111 21-29. DOI: 10.1016/j.rmed.2016.01.003 32. Sridhar S., Liu H., Pham T.H., Damera G., Newbold P.. Modulation of blood inflammatory markers by benralizumab in patients with eosinophilic airway diseases. Respir. Res. (2019) 20 14. DOI: 10.1186/s12931-018-0968-8 33. Bergantini L., D’Alessandro M., Cameli P., Bianchi F., Sestini P., Bargagli E., Refini R.M.. Personalized Approach of Severe Eosinophilic Asthma Patients Treated with Mepolizumab and Benralizumab. Int. Arch. Allergy Immunol. (2020) 181 746-753. DOI: 10.1159/000508936 34. Watanabe H., Shirai T., Hirai K., Akamatsu T., Nakayasu H., Tamura K., Masuda T., Takahashi S., Tanaka Y., Kishimoto Y.. Blood eosinophil count and FeNO to predict benralizumab effectiveness in real-life severe asthma patients. J. Asthma (2021) 59 1796-1804. DOI: 10.1080/02770903.2021.1963769 35. Wakashin H., Hirose K., Maezawa Y., Kagami S.-I., Suto A., Watanabe N., Saito Y., Hatano M., Tokuhisa T., Iwakura Y.. IL-23 and Th17 cells enhance Th2-cell-mediated eosinophilic airway inflammation in mice. Am. J. Respir. Crit. Care Med. (2008) 178 1023-1032. DOI: 10.1164/rccm.200801-086OC 36. Irvin C., Zafar I., Good J., Rollins D., Christianson C., Gorska M.M., Martin R.J., Alam R.. Increased frequency of dual-positive TH2/TH17 cells in bronchoalveolar lavage fluid characterizes a population of patients with severe asthma. J. Allergy Clin. Immunol. (2014) 134 1175-1186.e7. DOI: 10.1016/j.jaci.2014.05.038 37. Wang Y.H., Voo K.S., Liu B., Chen C.Y., Uygungil B., Spoede W., Bernstein J.A., Huston D.P., Liu Y.-J.. A novel subset of CD4(+) T(H)2 memory/effector cells that produce inflammatory IL-17 cytokine and promote the exacerbation of chronic allergic asthma. J. Exp. Med. (2010) 207 2479-2491. DOI: 10.1084/jem.20101376 38. Busse W.W.. Biological treatments for severe asthma: A major advance in asthma care. Allergol. Int. (2019) 68 158-166. DOI: 10.1016/j.alit.2019.01.004 39. McGregor M.C., Krings J.G., Nair P., Castro M.. Role of Biologics in Asthma. Am. J. Respir. Crit. Care Med. (2019) 199 433-445. DOI: 10.1164/rccm.201810-1944CI 40. Everaere L., Ait-Yahia S., Molendi-Coste O., Vorng H., Quemener S., LeVu P., Fleury S., Bouchaert E., Fan Y., Duez C.. Innate lymphoid cells contribute to allergic airway disease exacerbation by obesity. J. Allergy Clin. Immunol. (2016) 138 1309-1318.e11. DOI: 10.1016/j.jaci.2016.03.019 41. Kim H.Y., Lee H.J., Chang Y.J., Pichavant M., Shore S.A., Fitzgerald K.A., Iwakura Y., Israel E., Bolger K., Faul J.. Interleukin-17-producing innate lymphoid cells and the NLRP3 inflammasome facilitate obesity-associated airway hyperreactivity. Nat. Med. (2014) 20 54-61. DOI: 10.1038/nm.3423 42. Jingo K., Harada N., Nishioki T., Torasawa M., Yamada T., Asao T., Takagi H., Takeshige T., Ito J., Takahashi K.. Anaphylaxis to three humanized antibodies for severe asthma: A case study. Allergy Asthma Clin. Immunol. (2020) 16 46. DOI: 10.1186/s13223-020-00446-w
--- title: Prognostic Value of Simple Non-Invasive Tests for the Risk Stratification of Incident Hepatocellular Carcinoma in Cirrhotic Individuals with Non-Alcoholic Fatty Liver Disease authors: - Angelo Armandi - Gian Paolo Caviglia - Amina Abdulle - Chiara Rosso - Kamela Gjini - Gabriele Castelnuovo - Marta Guariglia - Nuria Perez Diaz del Campo - Daphne D’Amato - Davide Giuseppe Ribaldone - Giorgio Maria Saracco - Elisabetta Bugianesi journal: Cancers year: 2023 pmcid: PMC10046647 doi: 10.3390/cancers15061659 license: CC BY 4.0 --- # Prognostic Value of Simple Non-Invasive Tests for the Risk Stratification of Incident Hepatocellular Carcinoma in Cirrhotic Individuals with Non-Alcoholic Fatty Liver Disease ## Abstract ### Simple Summary In this cohort study of cirrhotic patients with non-alcoholic fatty liver disease, we found that fibrosis-4 score (FIB-4) was associated with incident hepatocellular carcinoma (HCC) over a median follow up of 6 years, independently from metabolic co-factors (obesity and type 2 diabetes). The lowest cut-off of 1.45 to rule out and the highest cut-off of 3.25 to rule in allow for the optimal risk stratification of HCC in this population. ### Abstract Hepatocellular carcinoma (HCC) represents a relevant disease burden in cirrhotic patients with non-alcoholic fatty liver disease (NAFLD). We aimed to investigate the prognostic value of simple non-invasive tests (NITs) (AAR, APRI, BARD, FIB-4) for the stratification of HCC risk development in a cohort of 122 consecutive cirrhotic individuals with NAFLD. Over a median follow up of 5.9 (3.2–9.3) years, 13 ($10.7\%$) developed HCC. Only FIB-4 was associated with HCC risk (HR = 1.27, $95\%$ CI 1.03–1.58, $$p \leq 0.027$$). After evaluating different established FIB-4 cut-offs, the lowest cut-off of 1.45 allowed the ruling out of a greater number of patients with a minimal risk of HCC than the 1.3 cut-off (23 vs. 18 patients). Conversely, the cumulative incidence of HCC using the highest cut-off of 3.25 (rule in) was distinctly higher than the 2.67 cut-off ($19.4\%$ vs. $13.3\%$). After multivariate Cox regression analysis, these cut-offs were independently associated with HCC after adjusting for sex, BMI and T2DM (HR = 6.40, $95\%$ CI 1.71–24.00, $$p \leq 0.006$$). In conclusion, FIB-4 values of <1.3 and >3.25 could allow for the optimal stratification of long-term HCC risk in cirrhotic individuals with NAFLD. ## 1. Introduction Non-alcoholic fatty liver disease (NAFLD) represents the most common form of liver disease worldwide, with an estimated prevalence of $25\%$ among the adult population, growing in parallel with the increase in obesity and type 2 diabetes mellitus (T2DM) [1]. NAFLD encompasses different forms of liver injury, from simple intrahepatic steatosis to a progressive form of chronic inflammation (named non-alcoholic steatohepatitis, NASH), which can lead to advanced liver disease, including cirrhosis and its complications [2,3]. In particular, the incidence of hepatocellular carcinoma (HCC) represents a relevant disease burden in this population, with considerable rates of morbidity and mortality [4,5]. In the NAFLD landscape, multiple metabolic risk factors may synergistically promote tumorigenesis through enhanced oxidative stress and cell metabolic pathway derangements [6]. According to the American Association for the Study of the Liver (AASLD) and the European Association for the Study of the Liver (EASL) guidelines, cirrhotic patients should undergo standardized screening modalities for HCC surveillance via six-month abdominal ultrasound [7,8]. However, this approach does not take into account the heterogeneity in cirrhotic individuals, which may imply different degrees of HCC risk according to either clinical features, or the impact of different etiologies as well as the removal of the etiologic agent [9]. In the context of NAFLD, the natural history is highly unpredictable due to multiple environmental factors, hence requiring a more tailored approach for a better risk stratification, with the aim of employing personalized, cost-effective procedures [10]. Non-invasive tests (NITs) have been introduced in clinical practice to overcome the limitation of invasive procedures for diagnostic/stratification purposes (e.g., liver biopsy to assess fibrosis stage or hepatic vein portal gradient (HVPG) to assess portal hypertension). In addition, some NITs have provided strong evidence for prognostication in cirrhotic patients of any etiology at risk for HCC [11]. Most NITs consist of simple scores using readily available clinical–biochemical variables, including fibrosis-4 (FIB-4) score that has been widely validated for the cross-sectional identification of advanced liver fibrosis in the hepatologist referral pathway [12]. In addition, some evidence has been so far provided for a longitudinal assessment of NITs with regard to long-term outcomes in NAFLD [13,14], including liver decompensation and overall mortality. However, less strong evidence exists with regard to HCC incidence in NAFLD using NITs at univocal cut-offs. In one study, only patients with high FIB-4 values, which persisted across repetitive follow up measurements, had a relevant risk of incident HCC [15]. Based on these premises, the aim of the present study was to investigate the prognostic value of simple NITs for the stratification of HCC risk development in NAFLD cirrhotic patients on long-term follow up. ## 2.1. Study Design and Study Population This retrospective cohort study included consecutive patients with cirrhosis due to NAFLD at their first referral at the outpatient clinic of the Unit of Gastroenterology and Hepatology of A.O.U. Città della Salute e della Scienza di Torino—Molinette Hospital, Turin, Italy, between January 2010 and April 2022. NAFLD was assessed by either histologic evaluation (macrovescicular steatosis > $5\%$) or by liver steatosis at abdominal ultrasound, in the absence of other known causes of liver damage (including viral hepatitis, cholesteric/autoimmune liver disease, drug-induced liver injury, use of steatogenic medications) along with the presence of metabolic risk factors (including obesity, T2DM, arterial hypertension, dyslipidemia) [16]. All available clinical, biochemical and anthropometric variables were retrieved at the time of the cirrhosis diagnosis. Liver stiffness was measured via vibration-controlled transient elastography (VCTE) (FibroScan®, Echosens™, Paris, France) in a fasting condition, performed by an expert operator, and using the M or XL probe as appropriate. All measurements were considered technically reliable with an IQR/med ratio of <$30\%$. A flow chart of study is provided in Figure 1. Inclusion criteria were as follows: aged 18 years or older and a clinical follow up of at least 6 months. Diagnosis of NAFLD cirrhosis was made via histology, or instrumental evidence (including ultrasound or computed tomography imaging), or through indirect signs of portal hypertension (including abdominal collateral circles, platelet count < 150 × 109/L, esophageal varices) [17,18]. Exclusion criteria were as follows: a previous decompensation event (including ascites, gastrointestinal bleeding, hepatic encephalopathy), previous occurrence of HCC, unavailable biochemical data for NIT calculation, and clinical follow up of less than 6 months. HCC was diagnosed via histologic examination or via suggestive signs at second-level imaging (contrast-enhanced computed tomography or magnetic resonance imaging) according to EASL guidelines [8]. ## 2.2. Non-Invasive Tests The following NITs were calculated at the time of the first referral for each patient according to the original formula: Aspartate aminotransferase (AST) to alanine aminotransferase (ALT) ratio (AAR) [19]: AST/ALT ratio; AST to platelets ratio index (APRI) [20]: (AST/AST upper limit normal/Platelets (109/L)) × 100; Body mass index (BMI), AST/ALT ratio, T2DM score (BARD) [21]: BMI ≥ 28 kg/m2 = 1 point, AST/ALT ratio ≥ 0.8 = 2 points, T2DM = 1 point; FIB-4 [22]: (Age (years) × AST (U/L))/(Platelets (109/L) × √ALT (U/L)). ## 2.3. Statistical Analysis Continuous variables were reported as median and interquartile ranges (IQR) according to their distribution. Data normality was assessed using the D’Agostino–Pearson test. Categorical variables were reported as number (n) and percentage (%). The association between baseline NIT values and HCC occurrence during the follow up was assessed using Cox proportional hazards regression analysis; results were reported as hazard ratio (HR) with the corresponding $95\%$ confidence interval (CI). Survival curves were analyzed with the Kaplan–Meier method; the corresponding p values were calculated via a log-rank test. Patients that did not develop HCC were censored at liver transplant or at the last follow up. A two-tailed p value of < 0.05 was considered statistically significant. All the statistical analyses were performed using MedCalc software, v.18.9.1 (MedCalc bvba, Ostend, Belgium). ## 3.1. Baseline Charcateristics of the Study Cohort A total of 122 patients with NAFLD cirrhosis were retrospectively included in this study based on the criteria depicted in Figure 1. Demographic, clinical and biochemical characteristics are reported in Table 1. The median age was 62 (51–67) years of age and the male to female ratio was $\frac{64}{58.}$ A total of $56.6\%$ ($$n = 69$$) of patients were obese (BMI ≥ 30.0 Kg/m2) and $57.4\%$ ($$n = 70$$) had a diagnosis of T2DM. Liver cirrhosis was detected via liver biopsy in 49 ($40.2\%$) patients; in the remaining 73 ($59.8\%$) patients, cirrhosis diagnosis was achieved from instrumental findings and/or clinical evidence of portal hypertension. Elevated liver enzymes (ALT and/or AST and/or γGT) were found in 105 ($86.1\%$) patients, with ALT > upper limit normal (ULN) in 76 ($62.3\%$) patients, AST > ULN in 37 ($30.3\%$) patients, and γGT > ULN in 96 ($78.7\%$) patients. Liver stiffness was available in nearly the half of the study cohort; in these patients, median stiffness was 20.5 (14.3–27.3) kPa. ## 3.2. Baseline NITs Values and Association with HCC Developemnt Baseline NITs values were calculated for all the 122 patients; median AAR values were 0.93 (0.74–1.30), median APRI values were 0.63 (0.79–0.88), median FIB-4 values were 2.37 (1.63–3.33), and median BARD values were 3 (2–4) (Figure 2). A total of 122 patients were followed for a median of 5.9 (3.2–9.3) years; during the follow up, 13 (cumulative incidence = $10.7\%$) patients developed HCC. In our population, the incidence rate was 1.5 per 100 person/year. Among the investigated NITs, only FIB-4 values resulted in a significant association with an increased risk of HCC development during follow up (HR = 1.27, $95\%$ CI 1.03–1.58, $$p \leq 0.027$$). No association was observed for AAR (HR = 1.78, $95\%$CI 0.51–6.31, $$p \leq 0.369$$), APRI (HR = 1.50, $95\%$CI 0.91–2.48, $$p \leq 0.115$$) or BARD (HR = 1.29, $95\%$CI 0.72–2.34, $$p \leq 0.392$$). ## 3.3. Stratifcation of the Risk of HCC According to FIB-4 Baseline NITs values were categorized according to the most widely adopted cut-offs in order to investigate which of them were the most appropriate for the stratification of the risk of HCC development in patients with NAFLD cirrhosis [15,22,23]. After Kaplan–Meier analysis, no statistical significance was observed for AAR, APRI and BARD (Supplementary Table S1 and Supplementary Figure S1); only FIB-4 showed different survival curves. Remarkably, no HCC occurred in patients with FIB-4 values below 1.3 and 1.45 (low-risk category); however, a cut-off value of 1.45 allowed us to rule out a greater number of patients with minimal risk of HCC as compared to a cut-off value of 1.3 (23 vs. 18, respectively). Conversely, the cumulative incidence of HCC in patients with FIB-4 of >3.25 ($\frac{6}{31}$; $19.4\%$) was distinctly higher compared to patients with FIB-4 of > 2.67 ($\frac{6}{45}$; $13.3\%$) (Table 2, Figure 3). According to these findings, a low-risk cut-off of 1.45 and a high-risk cut-off of 3.25 appeared the most appropriate values for the stratification of patients with NAFLD cirrhosis according their individual risk of HCC development during the follow up. Finally, after multivariate Cox regression analysis adjusted for sex, BMI and T2DM, only FIB-4 (<1.45, 1.45–3.25, >3.25) was significantly and independently associated with an increased risk of HCC occurrence (HR = 6.40, $95\%$ CI 1.71–24.00, $$p \leq 0.006$$) (Table 3). ## 4. Discussion In this retrospective cohort study of patients with cirrhosis due to NAFLD, we found that among commonly used NITs, only FIB-4 values were significantly associated with increased incidence of HCC over a median time of 6 years. This significant association persisted after adjusting for major metabolic co-factors including BMI and T2DM. All patients from this cohort had not experienced any liver decompensation event or previous HCC occurrence; hence, this evidence was provided in a homogenous cohort, increasing the plausibility of the findings. HCC represents one of the most common complications of cirrhosis, leading to relevant morbidity and mortality in this population. In particular, NAFLD offers a favorable background for HCC development, giving the persistent chronic inflammation and enhanced oxidative stress caused by the multiple metabolic-dysfunction-related injuries. In fact, up to one third of NAFLD individuals may develop HCC even in a pre-cirrhotic stage. However, since cirrhosis is the most relevant pre-neoplastic condition for HCC development, we included all consecutive cirrhotic patients at their first referral. A wide validation of non-invasive tools for HCC risk stratification and prognostication is still lacking, and cirrhotic patients are univocally monitored over time following standardized protocols. However, a tailored approach, with the aim of achieving a personalized surveillance strategy would be warranted, for either a cost-effective use of resources or for a better risk-based monitoring over time [24]. Consistent with our findings, some studies have assessed the potential role of non-invasive scores for longitudinal purposes, and FIB-4 showed the best accuracy to predict long-term hard outcomes. In fact, FIB-4 has been associated with liver-related events and overall mortality in NAFLD [13,14,25,26,27,28]. Additionally, one study investigated the FIB-4 changes over time in NAFLD patients, finding that persistently high FIB-4 values were associated with increased incidence of cirrhosis and HCC using a cut-off of 1.45 [15]. Similarly, in our cohort, we found that using the lowest FIB-4 cut-off of 1.45 to rule out HCC allows for a better selection of patients with a minimal risk of developing HCC. In addition, in the present study, the cut-off of 3.25 seemed optimal to identify the highest cumulative incidence of HCC. In cross-sectional studies, Sterling et al. first assessed the presence of liver fibrosis in HCV/HIV patients using the same cut-offs of 1.45 and 3.25 to rule out and rule in, respectively [22]. On the other hand, another study assessed the risk of advanced fibrosis using 1.30 and 2.67 cut-offs, which did not provide the same evidence in our longitudinal setting ($$p \leq 0.052$$) [29]. Consistent with our evidence, a 5-year prospective study on biopsy-proven NAFLD patients assessed a higher risk of death and liver transplantation with FIB-4 of >3.25 (HR = 6.33) [25]. The majority of other studies assessing the risk for long-term outcomes in NAFLD has provided stronger evidence using the 1.30 and 2.67 cut-offs [14,23,26,27,28]. These discrepancies may be partly be explained by the biological variability in the study cohorts and in the different degree of liver disease severity that is translated into the FIB-4 score, with regard to age, portal hypertension (and derivative platelet count) and intrahepatic ongoing inflammation as mirrored by transaminases. In the landscape of NAFLD, where the natural history may be unpredictable, being shaped by multiple environmental factors, a careful stratification of patients is advisable. Based on our findings and in the context of the existing literature, the follow-up evaluation of FIB-4 values would allow for the best characterization of patients into risk categories, which may change over time according to diverse variables [30]. The strength of this study is the well-characterized cohort of NAFLD patients with cirrhosis according to defined criteria and with the long-term follow up that allowed for the longitudinal evaluation. However, some limitations have to be outlined. The retrospective nature of the study did not allow for a complete clinical–biochemical evaluation at the time of diagnosis, leading to a relevant lack of data in some cases (including CHILD class definition and availability of liver stiffness values for the whole cohort) that prevented for an optimal cohort stratification. In addition, the lack of comprehensive baseline biochemical data did not allow for the calculation of other well-established scores and biomarkers (i.e., alpha-fetoprotein (AFP)) in the HCC landscape. Indeed, AFP data availability was even lower ($$n = 27$$), since the measure of circulating biomarkers is not recommended for the surveillance of patients at risk of HCC in European guidelines [8]. Overall, these limitations are also explained in the view that our aim was to explore the incidence of HCC starting from the cirrhosis diagnosis, which may have occurred prior to our first observation, significantly affecting the availability of the data. ## 5. Conclusions In conclusion, in this cohort study of NAFLD cirrhotic patients, FIB-4 values are significantly and independently associated with HCC incidence over a median time of 5 years. Among the most validated FIB-4 cut-offs, we found that using the low cut-off of 1.45 (ruling out) and the high cut-off of 3.25 (ruling in) would allow for the optimal risk stratification in this population. ## References 1. Younossi Z.M., Koenig A.B., Abdelatif D., Fazel Y., Henry L., Wymer M.. Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology (2016) 64 73-84. DOI: 10.1002/hep.28431 2. Bugianesi E., Leone N., Vanni E., Marchesini G., Brunello F., Carucci P., Musso A., Paolis P.D., Capussotti L., Salizzoni M.. Expanding the natural history of nonalcoholic steatohepatitis: From cryptogenic cirrhosis to hepatocellular carcinoma. Gastroenterology (2002) 123 134-140. DOI: 10.1053/gast.2002.34168 3. Armandi A., Bugianesi E.. Natural history of NASH. Liver Int. (2021) 41 78-82. DOI: 10.1111/liv.14910 4. Reddy S.K., Steel J.L., Chen H.W., DeMateo D.J., Cardinal J., Behari J., Humar A., Marsh J.W., Geller D.A., Tsung A.. Outcomes of curative treatment for hepatocellular cancer in nonalcoholic steatohepatitis versus hepatitis C and alcoholic liver disease. Hepatology (2012) 55 1809-1819. DOI: 10.1002/hep.25536 5. Huang D.Q., El-Serag H.B., Loomba R.. Global epidemiology of NAFLD-related HCC: Trends, predictions, risk factors and prevention. Nat. Rev. Gastroenterol. Hepatol. (2021) 18 223-238. DOI: 10.1038/s41575-020-00381-6 6. Marengo A., Rosso C., Bugianesi E.. Liver Cancer: Connections with Obesity, Fatty Liver, and Cirrhosis. Annu. Rev. Med. (2016) 67 103-117. DOI: 10.1146/annurev-med-090514-013832 7. Marrero J.A., Kulik L.M., Sirlin C.B., Zhu A.X., Finn R.S., Abecassis M.M., Roberts L.R., Heimbach J.K.. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology (2018) 68 723-750. DOI: 10.1002/hep.29913 8. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J. Hepatol. (2018) 69 182-236. DOI: 10.1016/j.jhep.2018.03.019 9. Ciancio A., Ribaldone D.G., Spertino M., Risso A., Ferrarotti D., Caviglia G.P., Carucci P., Gaia S., Rolle E., Sacco M.. Who Should Not Be Surveilled for HCC Development after Successful Therapy with DAAS in Advanced Chronic Hepatitis C? Results of a Long-Term Prospective Study. Biomedicines (2023) 11. DOI: 10.3390/biomedicines11010166 10. Singal A.G., Lok A.S., Feng Z., Kanwal F., Parikh N.D.. Conceptual Model for the Hepatocellular Carcinoma Screening Continuum: Current Status and Research Agenda. Clin. Gastroenterol. Hepatol. (2022) 20 9-18. DOI: 10.1016/j.cgh.2020.09.036 11. Caviglia G.P., Troshina G., Santaniello U., Rosati G., Bombaci F., Birolo G., Nicolosi A., Saracco G.M., Ciancio A.. Long-Term Hepatocellular Carcinoma Development and Predictive Ability of Non-Invasive Scoring Systems in Patients with HCV-Related Cirrhosis Treated with Direct-Acting Antivirals. Cancers (2022) 14. DOI: 10.3390/cancers14030828 12. EASL Clinical Practice Guidelines on non-invasive tests for evaluation of liver disease severity and prognosis-2021 update. J. Hepatol. (2021) 75 659-689. DOI: 10.1016/j.jhep.2021.05.025 13. Younes R., Caviglia G.P., Govaere O., Rosso C., Armandi A., Sanavia T., Pennisi G., Liguori A., Francione P., Gallego-Duran R.. Long-term outcomes and predictive ability of non-invasive scoring systems in patients with non-alcoholic fatty liver disease. J. Hepatol. (2021) 75 786-794. DOI: 10.1016/j.jhep.2021.05.008 14. Hagström H., Nasr P., Ekstedt M., Stål P., Hultcrantz R., Kechagias S.. Accuracy of Noninvasive Scoring Systems in Assessing Risk of Death and Liver-Related Endpoints in Patients With Nonalcoholic Fatty Liver Disease. Clin. Gastroenterol. Hepatol. (2019) 17 1148-1156. DOI: 10.1016/j.cgh.2018.11.030 15. Cholankeril G., Kramer J.R., Chu J., Yu X., Balakrishnan M., Li L., El-Serag H.B., Kanwal F.. Longitudinal changes in fibrosis markers are associated with risk of cirrhosis and hepatocellular carcinoma in non-alcoholic fatty liver disease. J. Hepatol. (2022) 78 493-500. DOI: 10.1016/j.jhep.2022.10.035 16. Boyle M., Masson S., Anstee Q.M.. The bidirectional impacts of alcohol consumption and the metabolic syndrome: Cofactors for progressive fatty liver disease. J. Hepatol. (2018) 68 251-267. DOI: 10.1016/j.jhep.2017.11.006 17. Gaia S., Campion D., Evangelista A., Spandre M., Cosso L., Brunello F., Ciccone G., Bugianesi E., Rizzeto M.. Non-invasive score system for fibrosis in chronic hepatitis: Proposal for a model based on biochemical, FibroScan and ultrasound data. Liver Int. (2015) 35 2027-2035. DOI: 10.1111/liv.12761 18. Caviglia G.P., Armandi A., Rosso C., Gaia S., Aneli S., Rolle E., Abate M.L., Olivero A., Nicolosi A., Guariglia M.. Biomarkers of Oncogenesis, Adipose Tissue Dysfunction and Systemic Inflammation for the Detection of Hepatocellular Carcinoma in Patients with Nonalcoholic Fatty Liver Disease. Cancers (2021) 13. DOI: 10.3390/cancers13102305 19. De Ritis F., Coltorti M., Giusti G.. An enzymic test for the diagnosis of viral hepatitis; the transaminase serum activities. Clin. Chim. Acta. (1957) 2 70-74. DOI: 10.1016/0009-8981(57)90027-X 20. Wai C.T., Greenson J.K., Fontana R.J., Kalbfleisch J.D., Marrero J.A., Conjeevaram H.S., Lok A.S.. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology (2003) 38 518-526. DOI: 10.1053/jhep.2003.50346 21. Harrison S.A., Oliver D., Arnold H.L., Gogia S., Neuschwander-Tetri B.A.. Development and validation of a simple NAFLD clinical scoring system for identifying patients without advanced disease. Gut (2008) 57 1441-1447. DOI: 10.1136/gut.2007.146019 22. Sterling R.K., Lissen E., Clumeck N., Sola R., Correa M.C., Montaner J., Sulkowski M.S., Torriani F.J., Dieterich D.T., Thomas D.L.. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology (2006) 43 1317-1325. DOI: 10.1002/hep.21178 23. Angulo P., Bugianesi E., Bjornsson E.S., Charatcharoenwitthaya P., Mills P.R., Barrera F., Haflidadottir S., Day C.P., George J.. Simple noninvasive systems predict long-term outcomes of patients with nonalcoholic fatty liver disease. Gastroenterology (2013) 145 782-789. DOI: 10.1053/j.gastro.2013.06.057 24. Ioannou G.N.. HCC surveillance after SVR in patients with F3/F4 fibrosis. J. Hepatol. (2021) 74 458-465. DOI: 10.1016/j.jhep.2020.10.016 25. Sebastiani G., Alshaalan R., Wong P., Rubino M., Salman A., Metrakos P., Deschenes M., Ghali P.. Prognostic Value of Non-Invasive Fibrosis and Steatosis Tools, Hepatic Venous Pressure Gradient (HVPG) and Histology in Nonalcoholic Steatohepatitis. PLoS ONE (2015) 10. DOI: 10.1371/journal.pone.0128774 26. Kim D., Kim W.R., Kim H.J., Therneau T.M.. Association between noninvasive fibrosis markers and mortality among adults with nonalcoholic fatty liver disease in the United States. Hepatology (2013) 57 1357-1365. DOI: 10.1002/hep.26156 27. Vieira Barbosa J., Milligan S., Frick A., Broestl J., Younossi Z., Afdhal N.H., Lai M.. Fibrosis-4 Index as an Independent Predictor of Mortality and Liver-Related Outcomes in NAFLD. Hepatol. Commun. (2022) 6 765-779. DOI: 10.1002/hep4.1841 28. Önnerhag K., Hartman H., Nilsson P.M., Lindgren S.. Non-invasive fibrosis scoring systems can predict future metabolic complications and overall mortality in non-alcoholic fatty liver disease (NAFLD). Scand. J. Gastroenterol. (2019) 54 328-334. DOI: 10.1080/00365521.2019.1583366 29. Shah A.G., Lydecker A., Murray K., Tetri B.N., Contos M.J., Sanyal A.J., Network N.C.R.. Comparison of noninvasive markers of fibrosis in patients with nonalcoholic fatty liver disease. Clin. Gastroenterol. Hepatol. (2009) 7 1104-1112. DOI: 10.1016/j.cgh.2009.05.033 30. Ishiba H., Sumida Y., Tanaka S., Yoneda M., Hyogo H., Ono M., Fujii H., Eguchi Y., Suzuki Y., Yoneda M.. The novel cutoff points for the FIB4 index categorized by age increase the diagnostic accuracy in NAFLD: A multi-center study. J. Gastroenterol. (2018) 53 1216-1224. DOI: 10.1007/s00535-018-1474-y
--- title: Ultrasensitive Photoelectrochemical Immunoassay Strategy Based on Bi2S3/Ag2S for the Detection of the Inflammation Marker Procalcitonin authors: - Guanhui Zhao - Yingying Wang - Huixin Wang - Guozhen Bai - Nuo Zhang - Yaoguang Wang - Qin Wei journal: Biosensors year: 2023 pmcid: PMC10046654 doi: 10.3390/bios13030366 license: CC BY 4.0 --- # Ultrasensitive Photoelectrochemical Immunoassay Strategy Based on Bi2S3/Ag2S for the Detection of the Inflammation Marker Procalcitonin ## Abstract As an inflammatory marker, procalcitonin (PCT) is more representative than other traditional inflammatory markers. In this work, a highly efficient photoelectrochemical (PEC) immunosensor was constructed based on the photoactive material Bi2S3/Ag2S to realize the sensitive detection of PCT. Bi2S3 was prepared by a hydrothermal method, and Ag2S quantum dots were deposited on the ITO/Bi2S3 surface via in situ reduction. Bi2S3 is a kind of admirable photoelectric semiconductor nanomaterial on account of its moderate bandgap width and low binding rate of photogenerated electron holes, which can effectively convert light energy into electrical energy. Therefore, based on the energy level matching principle of Bi2S3 and Ag2S, a labeled Bi2S3/Ag2S PEC immunosensor was constructed, and the sensitive detection of PCT was successfully established. The linear detection range of the PEC immunosensor was 0.50 pg∙mL−1 to 50 ng∙mL−1, and the minimum detection limit was 0.18 pg∙mL−1. Compared with the traditional PEC strategy, the proposed PEC immunosensor is simple, convenient, and has good anti-interference, sensitivity, and specificity, which could provide a meaningful theoretical basis and reference value for the clinical detection of PCT. ## 1. Introduction Procalcitonin (PCT) is an effective marker of blood infection that can be used to guide antibiotic treatment and disease evaluation in patients with respiratory system infections and blood flow infections [1,2]. The PCT levels increase when the patient has inflammation and infection, since the endocrine cells in the lungs and intestinal tissue synthesize and secrete large amounts of PCT [3,4]. The more serious the infection, the higher the PCT content [5,6]. In addition, the level of PCT concentration also plays a guiding role in the diagnosis of sepsis [7,8]. When the concentration of PCT in human serum exceeds 2 ng mL−1, it indicates that the blood has an infection that manifests as septicemia [9,10]. In summary, accurate detection of PCT content in serum has important significance for the early prevention and further treatment of inflammation [11,12]. To date, many PCT detection strategies have been reported, including colorimetric immunoassay [13,14], chemiluminescence immunoassay [15], electrochemical immunoassay [16], microfluidic immunoassay, and fluorescence immunoassay [17,18]. Here, we studied the application of PEC immunosensor technology to detect PCT. PEC immunosensors are a kind of sensing technology that combines photoelectric materials and specific recognition of biomolecules [19]. At present, many biomolecules (e.g., disease markers, gene sequences, specific recognition cells, etc.) have been used as detection targets [20]. The development prospects of immunosensor technology are promising because of its advantages, such as sensitive recognition, high selectivity, simple and easy operation process, and low cost. It is widely used in clinical medicine, food detection, and environmental monitoring [21]. PEC immunosensor is a kind of high-efficiency sensor that combines the advantages of photoelectrochemistry and biochemistry [22]. It has the advantages of miniaturization, simple operation, high sensitivity, and high specificity [23]. Although it has shown many advantages in various fields compared with traditional chemical detection methods, there are still many problems to be solved in the process of PEC technology’s development [24]. Firstly, a more stringent testing environment is required. Secondly, its stability and photoelectric conversion efficiency need to be improved [25]. Therefore, PEC immunosensors not only need bioactive materials with high photoelectric conversion efficiency, a large absorption range, and good stability, but also need to increase the diversity of the detection environment and the tolerance of the sensors [26]. In this work, two kinds of semiconductor materials with excellent photoelectric activity were selected to prepare a photoelectric chemical immunosensor that possessed the advantages of high photoelectric conversion efficiency, high stability, and simple determination [27]. Bi2S3 belongs to a family of metal chalcogenides in a class of non-toxic semiconductor materials [28], whose importance in photovoltaic and thermoelectric applications is well recognized [29]. Bi2S3 is an excellent photoelectric material because of its moderate bandgap width and low binding rate of photogenerated electron holes, which can effectively convert light energy into electrical energy [30]. Therefore, Bi2S3 was selected as the base material of the immunosensor in this work. In addition, Bi2S3 was sensitized by in situ deposition of Ag2S quantum dots via an immersion method. The preparation process was simple, universal, green, and excellent. The Bi2S3/Ag2S modified electrode significantly improved the photocurrent signal intensity of the sensor substrate, laying a good foundation for the construction of the PEC immunosensor. In this work, rod-like Bi2S3 was synthesized by a hydrothermal method, and Ag2S quantum dots were deposited in situ via an immersion strategy. Based on this, Bi2S3/Ag2S was used as the base material of the electrode, and anti-PCT, bovine serum albumin (BSA), and PCT were modified layer by layer; thus, a new and high-efficiency PEC immunosensor was constructed to realize the sensitive detection of PCT, and the limit of detection (LOD) was 0.18 pg mL−1. Compared with traditional sensors, the preparation process was simple and had fewer interference factors. The sensitivity, selectivity, and stability were excellent. This work could provide a theoretical basis for the use of semiconductor materials in sensing analysis. ## 2.1. Materials Anti-PCT and PCT were obtained from Lingchao Biotechnology Co., Ltd. (Shanghai, China). The other materials and apparatus used are discussed in the Supplementary Materials. ## 2.2. Synthesis Procedure of Bi2S3 The Bi2S3 was synthesized via a one-step hydrothermal method as reported in the literature [31]. First, 0.2254 g of thioacetamide and 0.7902 g of anhydrous bismuth nitrate were mixed and dissolved in 40 mL of ethanol, which was stirred to completely dissolve the mixture and then transferred to high-pressure Teflon-lined stainless-steel autoclave for 12 h at 180 °C. The product was washed with ultrapure water and ethanol and dried at 70 °C for 12 h in a vacuum-drying oven to obtain Bi2S3 as a black powder. ## 2.3. Steps of Synthesis of Bi2S3/Ag2S On the basis of the previous literature, Ag2S was deposited in situ via an immersion method. First, the ITO modified with Bi2S3 was immersed in 0.1 mol L−1 AgNO3 solution prepared with ethanol for 3 min, and then washed gently with ethanol. After natural drying at room temperature, it was immersed in 0.1 mol L−1 Na2S prepared by blending methanol and ultrapure water (volume ratio 1:1) for 3 min. Then, it was washed with a mixture of methanol and water (volume ratio 1:1). After drying at room temperature, Ag2S was successfully deposited on the surface of the Bi2S3/ITO electrode. ## 2.4. The Establishment Process of the Proposed PEC Immunosensor Firstly, the ITO glass electrode was cut to 2.5 × 1.0 cm2. Then, the ITO was washed ultrasonically with acetone, ethanol, and ultrapure water solution successively for 30 min. Then, the ITO was dried at 70 °C in the vacuum oven for 2 h. Next, we took 6 μL of Bi2S3 suspension solution (concentration = 5 mol L−1), applied it to the surface of the clean ITO electrode, and allowed it to dry naturally at room temperature until slightly wet. Then, the Bi2S3/ITO electrode was immersed in 0.1 mol L−1 AgNO3 solution for 3 min and then gently washed once with ethanol after being taken out with a clip. After it was dried naturally at room temperature, the ITO was immersed in 0.1 mol L−1 Na2S solution for 3 min. Then, it was taken out with a clip and washed twice with a mixture of methanol and ultrapure water (1:1). Thus, the synthetic deposition of Ag2S quantum dots was successful, and the Ag2S/Bi2S3/ITO modified electrode was successfully prepared. Then, we added 6 μL of 3 mmol L−1 TGA solution onto the prepared Ag2S/Bi2S3/ITO surface, and the carboxyl group in TGA and the amino group in the anti-PCT could undergo a condensation reaction and connect with one another. Subsequently, 6 μL of EDC/NHS (1:1) mixture was dropped onto the ITO to connect the anti-PCT with the base material and dried at 4 °C in a refrigerator until the surface was slightly wet. Then, 6 μL of PBS solution containing $0.1\%$ BSA was dropped to block non-specific binding sites on the surface of the Ag2S/Bi2S3/ITO. Finally, we added different concentrations of PCT and dried them at 4 °C in the refrigerator until the surface was slightly wet. It is worth noting that in each step of the electrode modification process, PBS solution was used once to wash away the excess substances not participating in the reaction when the electrode surface was dried to slightly wet. At this point, we had completed the construction of the PCT/BSA/anti-PCT/EDC/NHS/Ag2S/Bi2S3/ITO immunosensor, and it was then stored at 4 °C for standby. The construction diagram of the sensor is shown in Scheme 1. ## 2.5. PEC Analysis of PCT The photocurrent signal of this PEC sensor was measured with a three-electrode system on the photoelectrochemical workstation. The three-electrode system included a saturated calomel electrode (reference electrode), a counter electrode, and a working electrode. The calomel electrode was easily affected by the concentration and temperature of potassium chloride, so it was necessary to ensure that the potassium chloride solution was saturated. The counter electrode used in this workstation was a platinum counter electrode. As the electronic conductor of the other two electrodes, it did not participate in the electrode reaction. The Bi2S3/Ag2S modified ITO glass electrode was used as the working electrode. We placed the three-electrode system in PBS (pH = 7.4) solution containing 0.14 mol L−1 ascorbic acid (AA), and the light source was an LED lamp. The photocurrent signal of the sensor was detected by chronoamperometry, and the working curve was constructed according to the relationship between the photocurrent signal and the logarithm of the concentration of PCT antigen, so as to achieve the detection of the concentration level of the inflammatory marker PCT. ## 3.1. The Characteristics of Bi2S3 and Bi2S3/Ag2S Field-emission scanning electron microscopy (SEM) was used to observe the nanoscale morphology and structure of Bi2S3 and the Bi2S3/Ag2S composites to determine whether the synthesized materials met the requirements and whether Ag2S was successfully loaded onto the Bi2S3 materials, as shown in Figure 1. Figure 1A,B show the SEM images of Bi2S3 at different magnifications. It can be seen that Bi2S3 was a block composed of nanorods of different lengths and thicknesses crisscrossed together. Figure S1 shows the X-ray diffraction of Bi2S3, where it can be seen that, compared with the standard card, the XRD spectrum covered the most of the characteristic peaks of Bi2S3. Figure 1B shows that Bi2S3 was actually scattered nanorods with a diameter of 17~26 nm. The cylindrical side of the rod-shaped structure could provide a greater specific surface area and more active sites for the loading of Ag2S. Figure 1C,D show the SEM images of Bi2S3/Ag2S material under two detectors (InLens and BSE, respectively). The BSE detector was mainly used for signal scanning of backscattered electrons. The higher the atomic number, the stronger the signal, and the brighter the image. By comparing the brightness and size in Figure 1C,D, it can be seen that the Ag2S quantum dots were uniformly attached to the surface of the Bi2S3. At the same time, the elements contained in the material were analyzed with an energy spectrometer, as shown in Figure 1E. It could be seen that the base material contained the elements Bi, S, and Ag, and the Ag was uniformly distributed on the surface of the Bi2S3, further proving that Bi2S3 and Ag2S were successfully modified on the surface of the ITO electrode. In addition, the EDS mapping images of the Bi2S3/Ag2S composites are shown in Figure S2. In order to further prove that the Ag2S quantum dots were successfully deposited on the Bi2S3 nanorods, the UV–vis absorption spectra of Bi2S3 and Bi2S3/Ag2S were measured, as shown in Figure 1F. It could be observed that when the Ag2S quantum dots were deposited on the Bi2S3, the absorbance of Bi2S3/Ag2S (b curve) increased significantly and showed slight redshifts, indicating that Ag2S has a certain sensitization effect on Bi2S3. The satisfying photoelectric performance of the substrate material was dependent on its electron transfer mechanism. As shown in Figure 2, Bi2S3 has a bandgap energy of about 1.38 eV. Under the illumination of the light source, the absorbed light energy of electrons in the valence band of Bi2S3 transitioned to the conduction band. Thus, the holes were generated in the valence band and the electrons were increased in the conduction band. The electron holes moved and formed the current circuit, which successfully converted light energy into electric energy. AA, as an electron donor, could provide electrons and neutralize excessive photogenerated holes in the valence band, inhibiting the recombination of electron–hole pairs, enhancing the continuity of the current circuit, and improving the output efficiency of the photocurrent. Thus, the photoelectric activity of Bi2S3 could be improved. In addition, the generation principle of the electron–hole pairs was the same as described above when the light source was irradiated on the Ag2S quantum dots. The electrons in the conduction band of the Ag2S quantum dots were transferred to the conduction band of the Bi2S3 and then transferred to the electrode surface, forming an electron gradient flow. The energy band matching of the Bi2S3 and the Ag2S quantum dots effectively prevented the electron–hole recombination of Ag2S. Therefore, the electrons on the Ag2S could be efficiently transferred to the electrode surface to increase the photocurrent. ## 3.2. The Performance Characterization of the Proposed PEC Immunosensor Figure 3 shows the electrochemical impedance spectroscopy (EIS) characteristics of the immunosensor in PBS (pH = 7.4) solution containing 2.5 mmol L−1 [Fe(CN)6]3−/4− and 5 mmol L−1 KNO3. As shown in Figure 3A, with the layer-by-layer modification of the electrode surface, the semicircle’s diameter changed accordingly. Curve (a) shows the bare ITO glass electrode, and its semicircle diameter is very small, which indicates that its electron transfer resistance was very low. When the Bi2S3 (curve b) nanorod was modified to the electrode surface, it could be seen that it had a certain impedance because Bi2S3 is a nano-semiconductor material. However, the impedance value decreased instead (curve c) when Ag2S was deposited on the Bi2S3/ITO surface. Curves (d) and (e) show the impedance changes after successful modification of the coupling agents TGA and EDC/NHS, respectively. The impedance value decreased after the addition of TGA (curve d), which may have been due to the acidic nature of TGA. This could make the base liquid around the tested electrode present a strong electrolyte solution, resulting in an increase in the action of free electron transfer. After dropping with EDC/NHS (curve e), the diameter of the semicircle increased. We hypothesized that the EDC/NHS activated the carboxyl groups and reduced the acidity of TGA, so the free electron transfer decreased and the impedance value increased. The impedance value increased significantly (curve f) when the antibody, BSA, and PCT were successfully deposited in sequence, because they are biological protein molecules. As shown in Figure 3B, we studied the timing currents of the electrodes at different modification steps. It can be seen from curve (a) that the timing current of the bare ITO electrode was almost zero. The photocurrent signal increased significantly to 16 μA when the electrode surface was modified with Bi2S3 (curve b), because Bi2S3 is an efficient photoelectric semiconductor material. After depositing Ag2S on the surface of Bi2S3 (curve c), the current signal increased significantly to about 130 µA because of the energy band matching principle of Bi2S3 and Ag2S. After the coupling agents TGA (curve d) and EDC/NHS (curve e) were deposited onto the Ag2S/Bi2S3/ITO surface in sequence, the photocurrent change was inversely proportional to the EIS. When the antibody (curve f), BSA (curve g), and PCT were continuously deposited on the Ag2S/Bi2S3/ITO surface, the current signal decreased successively, because all of these substances are biological protein molecules that could block the electronic transmission. ## 3.3. Optimal Conditions for Analysis The different concentrations of Bi2S3 suspensions were prepared to drop onto the electrode. The photocurrent detection results are shown in Figure 4A. It can be seen that excessive concentrations of Bi2S3 suspension would make the resistance greater than the conductivity. Therefore, 5 mg mL−1 of Bi2S3 suspension was selected. As shown in Figure 4B, the thick material layer formed by the surplus Ag2S nanoparticles could reduce the conductivity and electron transfer rate and increase the photogenerated electron–hole binding rate. Therefore, 0.1 mol L−1 of AgNO3 solution was finally selected as the best concentration. Figure 4C shows that the photocurrent value reached its maximum when the pH was 7.4. Thus, PBS buffer with pH 7.4 was selected in this experiment. It can be seen from Figure 4D that the photocurrent response was the strongest when the concentration of AA was 0.14 mol L−1. The decrease in the photocurrent signal at higher concentrations of AA could be because of the quenching absorption of the electrolyte solution, which reduced the formation efficiency and light intensity of the excited electron–hole center. As shown in Figure 4E, the photocurrent signal intensity reached about 127 µA and then remained essentially unchanged when the working current was 3.0 A. In order to avoid the impact of strong light on photoelectric materials and biomolecules, 3.0 A was selected as the best working current. ## 3.4. PCT Detection Under the optimal experimental conditions, the photocurrent response curves were as shown in Figure 5A. Because PCT and PCT-antibody are not conductive, they could hinder the transfer of electrons between the electrode surface and the substrate solution. Thus, the response current continued to decrease with the increase in the PCT concentration and presented a good linear relationship, as shown in Figure 5B. The linear equation was I (μA) = 35.7555–12.6381 log c; linear correlation coefficient R2 = 0.9952. The linear detection range of PCT for this proposed immunosensor was 0.5 pg mL−1~50 ng mL−1, and the detection limit was as low as 0.18 pg mL−1. ( S/$$n = 3$$). In order to verify the accuracy of the constructed PEC immunosensor, four samples with known PCT concentrations were mixed and dropped onto the constructed immunosensor. In Figure 5C, curve (a) shows the photocurrent curve of the mixture of two known PCT concentrations (0.1 ng mL−1 and 0.5 ng mL−1). According to the linear fitting formula, the theoretical photocurrent intensity was 42.0 μA. The actual photocurrent intensity was 44.65 μA. Curve (b) shows the photocurrent curve of the mixture of two known PCT concentrations (0.01 ng mL−1 and 0.005 ng mL−1). According to the linear fitting formula, the theoretical photocurrent intensity was 62.0 μA. The actual photocurrent intensity was 63.75 μA, and the relative deviation was $2.8\%$ and $5.9\%$, respectively, showing that the accuracy of this PEC immunosensor was high and met the experimental needs. Comparing this sensor with other sensors (Table S1), it can be concluded that this PEC sensor has a wide linear range and a low detection limit, indicating that it has relatively excellent performance in detecting PCT, so this work is valuable. ## 3.5. Specificity, Stability, and Application of the PEC Immunosensor In order to study the specificity of the PEC immunosensor, we selected the interfering substance B-type natriuretic peptide (BNP). In Figure 5D, curve (a) shows the photoelectric signal response of the blank sample. Curve (b) shows the photoelectric signal response of a mixture of BNP (100 ng mL−1) and blank. Curve (c) shows the photoelectric signal response of the PCT sample (0.50 ng mL−1). Curve (d) shows the photoelectric signal response of the sample containing PCT (0.50 ng mL−1) and BNP (100 ng mL−1). The prepared sensor was determined according to the experimental method. Adding BNP had no obvious effect on the current signal of the PEC immunosensor, indicating that it had good specificity. To test the stability of the PEC immunosensor, the light was repeatedly switched on and off 18 times in 400 s to detect the photocurrent. The experimental results are shown in Figure 5E. The relative standard deviation (RSD) was 0.047, and the coefficient of variation (CV) was $4.7\%$ relative to the stability experiment. The difference between the timing current after 18 on/off cycles and the initial timing current was very small, indicating that the PEC immunosensor has good operational stability. In addition, seven sensors were stored in a refrigerator at 4 °C for five days, and the results showed that the current value of the immunosensor changed by less than $4\%$, indicating that the sensor has good storage stability. To study the potential application significance of the proposed PEC immunosensor for PCT determination, the standard addition recovery tests were implemented. Before that, we conducted pre-treatment for serum samples, including low-speed centrifugation at 4 °C, removing sediments, and using the supernatant for testing. As shown in Table S2, the recoveries were in the range of 86~$101\%$, and the RSD was in the range of 3.1~$4.1\%$, indicating great reference significance for clinical PCT detection. ## 4. Conclusions In this work, since the complex of Bi2S3/Ag2S has excellent photoelectric performance, it was used as the base material to construct the PEC immunosensor. In addition, anti-PCT, BSA, and PCT were successively deposited on the Bi2S3/Ag2S/ITO surface. The novel unmarked PEC immunosensor was constructed successfully and achieved the highly sensitive detection of PCT in human serum. Under the optimal experimental conditions, the linear range of the PEC immunosensor for PCT detection was 0.5 pg mL−1~50.0 ng mL−1, and the detection limit was as low as 0.18 pg mL−1. Moreover, the PEC immunosensor showed admirable stability, selectivity, and reproducibility. This could provide a theoretical basis and reference value for the clinical application of semiconductor materials in the field of disease marker detection. ## Figures and Scheme Scheme 1: Schematic diagram of the PEC sensor’s construction. Figure 1: SEM images (A,B) of Bi2S3 with different magnification; SEM diagrams (C,D) of Bi2S3/Ag2S under InLens and BSE detectors, respectively; Energy spectrum analysis (EDS) diagram (E) of Bi2S3/Ag2S; Ultraviolet–visible (UV–vis) near-infrared absorption spectra (F) of Bi2S3 (a) and Ag2S (b). Figure 2: The electron transfer mechanism of the PEC immunosensor. Figure 3: The electrochemical impedance spectroscopy (EIS) (A) and the chronoamperogram (B) of the photoelectrochemical immunosensor: (a) ITO; (b) Bi2S3/ITO; (c) Ag2S/Bi2S3/ITO; (d) TGA/Ag2S/Bi2S3/ITO; (e) EDC/NHS/TGA/Ag2S/Bi2S3/ITO; (f) anti-PCT/EDC/NHS/TGA/Ag2S/Bi2S3/ITO; (g) BSA/anti-PCT/EDC/NHS/TGA/Ag2S/Bi2S3/ITO; (h) PCT/BSA/anti-PCT/EDC/NHS/TGA/Ag2S/Bi2S3/ITO. Figure 4: Optimal concentrations of Bi2S3 (A), AgNO3 (B), and AA (D); pH (C); photocurrent intensity (E). Error bar = SD (n = 5). Figure 5: Photocurrent response curves (A) and the linearity correction curve (B) of the PEC immunosensor at different PCT concentrations: (a) 0.0005, (b) 0.001, (c) 0.005, (d) 0.01, (e) 0.1, (f) 0.5, (g) 1.0, (h) 10, (i) 30, (j) 50 ng mL−1. Photocurrent curve of unknown PCT concentration (C): the mixture of PCT with concentrations of (a) 0.1 ng mL−1 and 0.5 ng mL−1, (b) 0.01 ng mL−1 and 0.005 ng mL−1. The specificity of the constructed immunosensor (D): (a) blank, (b) blank + 100 ng mL−1 BNP, (c) 0.50 ng mL−1 PCT, (d) 0.50 ng mL−1 + 100 ng mL−1 BNP. Stability evaluation of the photoelectric chemical immunosensor under multiple on/off irradiation cycles for 400 s (E). Error bars = SD (n = 5). ## References 1. Wussler D., Kozhuharov N., Oliveira M.T., Bossa A., Sabti Z., Nowak A., Murray K., de Lavallaz J.F., Badertscher P., Twerenbold R.. Clinical utility of procalcitonin in the diagnosis of pneumonia. Clin. Chem. (2019) 65 1532-1542. DOI: 10.1373/clinchem.2019.306787 2. Brabenec L., Hellenthal K., Müller M., Kintrup S., Zurek-Leffers F., Kardell M., Otto M., Wagner N.-M.. Procalcitonin mediates vascular dysfunction in obesity. Life Sci. (2022) 307 120889. DOI: 10.1016/j.lfs.2022.120889 3. Wolfisberg S., Gregoriano C., Schuetz P.. Procalcitonin for individualizing antibiotic treatment: An update with a focus on COVID-19. Crit. Rev. Clin. Lab. Sci. (2022) 59 54-65. DOI: 10.1080/10408363.2021.1975637 4. Battaglia F., Baldoneschi V., Meucci V., Intorre L., Minunni M., Scarano S.. Detection of canine and equine procalcitonin for sepsis diagnosis in veterinary clinic by the development of novel MIP-based SPR biosensors. Talanta (2021) 230 122347. DOI: 10.1016/j.talanta.2021.122347 5. Jia Y., Yang L., Xue J., Zhang N., Fan D., Ma H., Ren X., Hu L., Wei Q.. Bioactivity-protected electrochemiluminescence biosensor using gold nanoclusters as the low-potential luminophor and Cu. ACS Sens. (2019) 4 1909-1916. DOI: 10.1021/acssensors.9b00870 6. Zhu K., Chen J., Hu J., Xiong S., Zeng L., Huang X., Xiong Y.. Low-sample-consumption and ultrasensitive detection of procalcitonin by boronate affinity recognition-enhanced dynamic light scattering biosensor. Biosens. Bioelectron. (2022) 200 113914. DOI: 10.1016/j.bios.2021.113914 7. Ge X., Zhang J., Feng Y., Wang A., Mei L., Feng J.J.. Label-free electrochemical biosensor for determination of procalcitonin based on graphene-wrapped Co nanoparticles encapsulated in carbon nanobrushes coupled with AuPtCu nanodendrites. Microchim. Acta (2022) 189 110. DOI: 10.1007/s00604-022-05179-8 8. Miao J., Du K., Li X., Xu X., Dong X., Fang J., Cao W., Wei Q.. Ratiometric electrochemical immunosensor for the detection of procalcitonin based on the ratios of SiO. Biosens. Bioelectron. (2021) 171 112713. DOI: 10.1016/j.bios.2020.112713 9. Zhao L., Song X., Ren X., Wang H., Fan D., Wu D., Wei Q.. Ultrasensitive near-infrared electrochemiluminescence biosensor derived from Eu-MOF with antenna effect and high efficiency catalysis of specific CoS. Biosens. Bioelectron. (2021) 191 113409. DOI: 10.1016/j.bios.2021.113409 10. Song X., Zhao L., Zhang N., Liu L., Ren X., Ma H., Kuang X., Li Y., Luo C., Wei Q.. Ultrasensitive electrochemiluminescence biosensor with silver nanoclusters as a novel signal probe and α-Fe. Anal. Chem. (2023) 95 1582-1588. DOI: 10.1021/acs.analchem.2c04673 11. Zhang N., Feng J., Zhao G., Duan X., Wang Y., Zhang D., Wei Q.. Ultrasensitive photochemical immunosensor based on flowerlike SnO. Biosensors (2021) 11. DOI: 10.3390/bios11110421 12. Gupta Y., Pandey C.M., Ghrera A.S.. Development of conducting cellulose paper for electrochemical sensing of procalcitonin. Microchim. Acta (2023) 190 32. DOI: 10.1007/s00604-022-05596-9 13. Li Y., Liu L., Wang Y., Ren R., Fan D., Wu D., Du Y., Xu K., Ren X., Wei Q.. Enzyme-free colorimetric immunoassay for procalcitonin based on MgFe. Sens. Actuators B-Chem. (2020) 316 128163. DOI: 10.1016/j.snb.2020.128163 14. Zhang Y., Si X., Zhang M., Yang X., Yuan H., Wang X., Zhang Y., Wang H.. Rapid colorimetric determination of procalcitonin using magnetic separation and enzymatic catalysis. Anal. Lett. (2019) 52 602-612. DOI: 10.1080/00032719.2018.1482312 15. Huang E., Huang D., Wang Y., Cai D., Luo Y., Zhong Z., Liu D.. Active droplet-array microfluidics-based chemiluminescence immunoassay for point-of-care detection of procalcitonin. Biosens. Bioelectron. (2022) 195 113684. DOI: 10.1016/j.bios.2021.113684 16. Tanak A.S., Jagannath B., Tamrakar Y., Muthukumar S., Prasad S.. Non-faradaic electrochemical impedimetric profiling of procalcitonin and C-reactive protein as a dual marker biosensor for early sepsis detection. Anal. Chim. Acta X (2019) 3 100029. DOI: 10.1016/j.acax.2019.100029 17. Molinero-Fernández Á., López M., Escarpa A.. An on-chip microfluidic-based electrochemical magneto-immunoassay for the determination of procalcitonin in plasma obtained from sepsis diagnosed preterm neonates. Analyst (2020) 145 5004-5010. DOI: 10.1039/D0AN00624F 18. Molinero-Ferna A.G., Moreno-Guzman M., Arruza L., Lo M.A.N., Escarpa A.. Polymer-based micromotor fluorescence immunoassay for on-the-move sensitive procalcitonin determination in very low birth weight infants’ plasma. ACS Sens. (2020) 5 1336-1344. DOI: 10.1021/acssensors.9b02515 19. Cao Y., Wang L., Wang C., Hu X., Liu Y., Wang G.. Sensitive detection of glyphosate based on a Cu-BTC MOF/g-C. Electrochim. Acta (2019) 317 341-347. DOI: 10.1016/j.electacta.2019.06.004 20. Afzal A., Iqbal M., Dastgeer G., Nazir G., Mumtaz S., Usman M., Eom J.. WS. ACS Appl. Mater. Inter. (2020) 12 39524-39532. DOI: 10.1021/acsami.0c05114 21. Aydin M., Aydin E., Sezgintürk M.. Advances in immunosensor technology. Adv. Clin. Chem. (2021) 102 1-62. PMID: 34044908 22. Zhang J., Xue X., Du Y., Zhao J., Ma H., Ren X., Wei Q., Ju H.. Antigen-down PEC immunosensor for Cyfra21-1 detection based on photocurrent polarity switching strategy. Anal. Chem. (2022) 94 12368-12373. DOI: 10.1021/acs.analchem.2c01478 23. Liu K., Deng H., Wang Y., Cheng S., Xiong X., Li C.. A sandwich-type photoelectrochemical immunosensor based on ReS. Sensor. Actuators B-Chem. (2020) 320 128341. DOI: 10.1016/j.snb.2020.128341 24. Li H., Mu Y., Yan J., Cui D., Ou W., Wan Y., Liu S.. Label-free photoelectrochemical immunosensor for neutrophil gelatinase-associated lipocalin based on the use of nanobodies. Anal. Chem. (2015) 87 2007-2015. DOI: 10.1021/ac504589d 25. Lu Y., Wang H., Shi X., Ding C., Fan G.. Photoanode-supported cathodic immunosensor for sensitive and specific detection of human chorionic gonadotropin. Anal. Chim. Acta (2022) 1199 339560. DOI: 10.1016/j.aca.2022.339560 26. Wang J., Guo Q., Li Q., Zheng L., Yang X., Wang X., Nie G.. A “signal-off” type photoelectrochemical immunosensor for detecting carcinoembryonic antigen based on TiO. Microchem. J. (2022) 182 107888. DOI: 10.1016/j.microc.2022.107888 27. Zhang L., Ran J., Qiao S., Jaroniec M.. Characterization of semiconductor photocatalysts. Chem. Soc. Rev. (2019) 48 5184-5206. DOI: 10.1039/C9CS00172G 28. Jia Y., Liu P., Wang Q., Wu Y., Cao D., Qiao Q.. Construction of Bi. J. Colloid Interf. Sci. (2021) 585 459-469. DOI: 10.1016/j.jcis.2020.10.027 29. Wang B., Cao J., Dong Y., Liu F., Fu X., Ren S., Ma S., Liu Y.. An in situ electron donor consumption strategy for photoelectrochemical biosensing of proteins based on ternary Bi. Chem. Commun. (2018) 54 806-809. DOI: 10.1039/C7CC08132D 30. Xiao W., Wang L., Wei X., Li J.. Chitosan-based molecularly imprinted photoelectric sensor with ZnO/Bi. Microchim. Acta (2022) 189 247. DOI: 10.1007/s00604-022-05326-1 31. Shi H., Zhao Y., Fan J., Tang Z.. Construction of novel Z-scheme flower-like Bi. Appl. Surf. Sci. (2019) 465 212-222. DOI: 10.1016/j.apsusc.2018.09.164 32. Li Y., Liu L., Liu X., Ren Y., Xu K., Zhang N., Sun X., Yang X., Ren X., Wei Q.. A dual-mode PCT electrochem-ical immunosensor with CuCo2S4 bimetallic sulfides as enhancer. Biosens. Bioelectron. (2020) 163 112280. DOI: 10.1016/j.bios.2020.112280 33. Chiang C., Huang T., Wang C., Huang C., Tsai T., Yu S., Chen Y., Hong S., Hsu C., Chang T.. Fiber optic nanogold-linked immunosorbent assay for rapid detection of procalcitonin at femtomolar concentration level. Biosens. Bioelectron. (2020) 151 111871. DOI: 10.1016/j.bios.2019.111871 34. Zhou Y., Shao X., Han Y., Zhang H.. Detection of procalcitonin (PCT) using the double antibody sandwich method based on fluorescence resonance energy transfer between upconversion nanoparticles and quantum dots. Anal. Methods (2018) 10 1015-1022. DOI: 10.1039/C8AY00039E 35. Wang X., Ma L., Hu C., Liu T., Sun S., Liu X., Guan M.. Simultaneous quantitative detection of IL-6 and PCT using SERS magnetic immunoassay with sandwich structure. Nanotechnology (2021) 32 255702. DOI: 10.1088/1361-6528/abee48 36. Xu X., Lei X., Ye L., Song S., Liu L., Xu L., Xu C., Hua K.. Gold-based paper sensor for sensitive detection of procalcitonin in clinical samples. Chinese J. Anal. Chem. (2022) 50 100062. DOI: 10.1016/j.cjac.2022.100062
--- title: Insulin Resistance Is Associated with an Unfavorable Serum Lipoprotein Lipid Profile in Women with Newly Diagnosed Gestational Diabetes authors: - Mikael Huhtala - Tapani Rönnemaa - Kristiina Tertti journal: Biomolecules year: 2023 pmcid: PMC10046655 doi: 10.3390/biom13030470 license: CC BY 4.0 --- # Insulin Resistance Is Associated with an Unfavorable Serum Lipoprotein Lipid Profile in Women with Newly Diagnosed Gestational Diabetes ## Abstract Background: Gestational diabetes (GDM) is associated with various degrees of insulin resistance—a feature related to increased risk of adverse perinatal outcomes. We aimed to determine the previously poorly investigated associations between maternal insulin resistance and serum fasting metabolome at the time of GDM diagnosis. Methods: Serum lipoprotein and amino acid profile was analyzed in 300 subjects with newly diagnosed GDM using a validated nuclear magnetic resonance spectroscopy protocol. Associations between insulin resistance (homeostasis model assessment of insulin resistance, HOMA2-IR) and serum metabolites were examined with linear regression. Results: We found insulin resistance to be associated with a distinct lipid pattern: increased concentration of VLDL triglycerides and phospholipids and total triglycerides. VLDL size was positively related and LDL and HDL sizes were inversely related to insulin resistance. Of fatty acids, increased total fatty acids, relative increase in saturated and monounsaturated fatty acids, and relative decrease in polyunsaturated and omega fatty acids were related to maternal insulin resistance. Conclusions: In newly diagnosed GDM, the association between maternal insulin resistance and serum lipoprotein profile was largely as described in type 2 diabetes. Lifestyle interventions aiming to decrease insulin resistance from early pregnancy could benefit pregnancy outcomes via more advantageous lipid metabolism. ## 1. Introduction Gestational diabetes (GDM) affects every sixth pregnancy globally [1], and it is associated with adverse perinatal outcomes such as macrosomia, shoulder dystocia, cesarean delivery, preeclampsia, neonatal hypoglycemia, hyperbilirubinemia, and increased need of neonatal intensive care [2]. In early pregnancy, maternal pancreatic beta-cell mass and function are increased already before the physiologic increase in insulin resistance [3]. In cases of inadequate insulin action, to compensate for the increased insulin resistance, maternal hyperglycemia, i.e., GDM, ensues. The origin of hyperglycemia is heterogeneous among GDM patients, with some having especially increased insulin resistance, some having mainly impaired beta-cell function, and some having a mixed pathophysiologic subtype [4]. The risk of perinatal adverse outcomes is related to GDM subtype—many of the GDM-associated complications are more common in insulin-resistant patients [4,5,6]. What factors fundamentally drive these associations between insulin resistance and adverse perinatal outcomes is not well elucidated. GDM is characterized not only by hyperglycemia, but also by increased serum lipids, amino acids, and ketone bodies [7,8,9,10,11]. Of these metabolites, lipids and amino acids have previously been shown to associate with fetal growth [12,13,14] and duration of gestation [12]. The associations between maternal insulin resistance and metabolome have previously been studied in the HAPO [15,16] and UPBEAT [17] cohorts as well as at least one smaller prospective cohort [18], but the effects of maternal insulin resistance on lipids in different lipoprotein subfractions are not known. To develop more effective and individualized treatments for GDM, the pathophysiology, also regarding hyperlipidemia and amino-acidemia, needs to be more thoroughly understood. As metformin and insulin treatments have partially diverging effects on maternal serum metabolome [12,13], individualizing treatment according to metabolic profile or pathophysiologic subtype (as described by Powe et al. [ 4]) could possibly yield improved maternal and fetal outcomes. In this study, we characterized the associations between insulin resistance and maternal serum metabolome in patients with newly diagnosed GDM. We hypothesized that maternal insulin resistance would be associated with similar changes in serum metabolomics as previously seen in type 2 diabetes, and that the changes in serum metabolome in GDM would be largely related to insulin resistance. ## 2. Materials and Methods The patients in this cohort were derived from a previous randomized trial comparing metformin and insulin treatments of GDM at Turku University Hospital, Finland, during the years 2006–2010 [19]. Patients who met the same inclusion and exclusion criteria but did not require antihyperglycemic medication were also included. The study population was described previously in detail [12,19,20]. Briefly, women with newly diagnosed GDM and singleton pregnancy were included in the study. GDM diagnosis was based on at least 2 abnormal values in a 2 h 75 g oral glucose tolerance test (OGTT). The diagnostic cutoff values were ≥4.8 (fasting), ≥10.0 (1 h), and ≥8.7 mmol/L (2 h) until the release of Finnish national guidelines in December 2008; thereafter cutoff values were ≥5.3, ≥10.0, and ≥8.6 mmol/L, respectively. The exclusion criteria were cardiac or renal insufficiency, liver disease, metformin use 0–3 months prior to conception or during early pregnancy before OGTT, or glucose above 7.0 mmol/L (fasting) or 11.0 mmol/L (60 min postprandial) in home plasma glucose monitoring. Pharmacological treatment (with metformin or insulin [19]) was initiated in cases of recurrent hyperglycemia (fasting glucose ≥ 5.5 and/or postprandial glucose ≥ 7.8 mmol/L) despite diet and lifestyle changes. Patients were recruited at mean 30 gestational weeks, after a diagnostic OGTT. Follow-up and mode and timing of delivery was decided by the managing clinician according to hospital guidelines. Maternal fasting blood samples were drawn at recruitment after GDM diagnosis and prior to initiation of any pharmacological antihyperglycemic treatment. Fasting plasma glucose, C-peptide, and glycated hemoglobin A1 (HbA1c) were analyzed using routine laboratory methods in the clinical laboratory of Turku University Hospital [19]. Samples for fasting plasma glucose measurement were collected in lithium heparin gel tubes according to local hospital laboratory protocol. All glucose samples were analyzed within 60 min of collection. The homeostasis model assessment of insulin resistance (HOMA2-IR) was calculated using C-peptide and glucose (https://www.dtu.ox.ac.uk/homacalculator, accessed on 3 November 2021) [21]. C-peptide with its longer half-life than insulin is a good measure of insulin secretion and may be used when estimating insulin resistance [22,23]. Additional serum samples were stored at below −70 °C to be used for assessment of targeted maternal metabolome (including detailed lipoprotein profile, fatty acids (FAs), and amino acids) by commercially available high-throughput 1H nuclear magnetic resonance spectroscopy (NMR) method (Nightingale Health Ltd., Helsinki, Finland) [24]. After sample preparation (described in [25,26]), a Bruker AVALANCE III 500 Mhz spectrometer was used to acquire the NMR spectra in two parts. First, a presaturated proton NMR spectrum including resonances from proteins and lipids in different lipoprotein particles, and, second, a T2-filtered spectrum in which the broad macromolecule and lipoprotein lipid signals are mostly suppressed, leading to better detection of low-molecular-weight metabolites [24]. The spectra were measured with 80 k data points and 8 scans with Bruker noesypresat pulse sequence, and 64 k data points and 24 or 16 scans with T2-relaxation filtered Bruker 1D CPMG pulse sequence, respectively. The NMR data were processed using Fourier transformation, phase correction, overall signal check for missing and/or extra peaks, background control, removal of baseline, and spectral area-specific signal alignments [26]. As quality control, the results were compared to control samples and an extensive database of quantitative molecular data. The quantification method was based on Bayesian models [27,28], and the concentrations were calibrated to agree with external standards [26]. This method enables quantification of lipid content in different lipoprotein subclasses and has been calibrated against gel permeation high-performance liquid chromatography [29]. The consistency is also similar to mass spectrometry and gas chromatography [30,31]. The original trial was approved by the Ethics Committee of the Southwest Hospital District of Finland (Dnro $\frac{246}{2005}$); it is registered at Clinicaltrials.gov (3 November 2010, NCT01240785, http://clinicaltrials.gov/ct2/show/NCT01240785), and it is in accordance with the 1964 Helsinki Declaration. New specimens or clinical data were not collected for the present secondary analysis; therefore, separate ethics committee approval was considered unnecessary. All the study participants signed an informed consent for participating in the original trial and for collection and further analysis of serum samples. ## Statistical Analyses All analyses were performed in R statistical software (version 4.0.3). Associations between HOMA2-IR and individual serum metabolites were assessed by linear regression. Prior to analyses, HOMA2-IR was log-transformed, centered, and scaled and the metabolite data centered and scaled. Regression models were run unadjusted and adjusted for a priori selected confounding factors: body mass index (BMI) class and gestational age at sampling. BMI classes were normal weight (18.5–24.9 kg/m2), overweight (25–30 kg/m2), and obese (≥30 kg/m2), according to BMI at the first antenatal visit. Two women were underweight (BMI < 18.5 kg/m2) and were excluded from the analyses. For illustrative purposes, unadjusted univariate regression analyses were performed also without first centering and scaling the data. Confidence intervals (CI) were acquired with the adjusted bootstrap percentile method. To avoid type I error, a p-value below 0.01 was considered statistically significant. ## 3. Results C-peptide, glucose, and serum samples for NMR analysis were available from 300 patients. Because some individual metabolite measures were discarded in the quality control, and there were missing data regarding BMI, 271–300 patients were available for the unadjusted and 268–297 for the adjusted analyses. Clinical characteristics are given in Table 1. The study participants were, on average, 31.6 ± 5.2 years old with mean pre-pregnancy BMI of 29.2 ± 5.3 kg/m2. GDM was diagnosed at mean 26.9 ± 2.4 gestational weeks, and C-peptide and NMR metabolome were measured at mean 30.5 ± 1.8 gestational weeks. Overall, the adjustment for BMI and gestational age had mostly minimal effects on the associations (Table S1). The adjusted associations are depicted in Figure 1 and Figure 2, and the results from both unadjusted and adjusted regression analyses are given in more detail in a supplementary table (Table S1). ## 3.1. Lipoprotein Concentrations and Total Lipids Insulin resistance was positively related to particle concentration and total lipids in large to extremely large very-low-density lipoprotein (VLDL) (Figure 2), and total VLDL lipids (Figure 1). The positive associations were significant between insulin resistance, total lipids in medium VLDL, and concentration of small VLDL (Figure 2). Insulin resistance was inversely related to particle concentration and concentration of total lipids in large to very large high-density lipoprotein (HDL) (Figure 2). ## 3.2. Cholesterol While insulin resistance was positively related to the concentrations of cholesterol in large to extremely large VLDL, the association to relative cholesterol content in every lipoprotein subclass was inverse (Figure 2). Insulin resistance was also inversely related to cholesterol concentration in intermediate-density lipoprotein (IDL), total HDL, and medium to very large HDL (Figure 1 and Figure 2). ## 3.3. Triglycerides Insulin resistance was associated to increased total triglyceride (TG) concentration, increased TG concentration in VLDL and HDL, and increased TG-to-phosphoglycerides ratio (Figure 1). In more detailed lipoprotein subclasses the associations were significant in all VLDL subclasses, small to medium low-density lipoprotein (LDL), and small to medium HDL. Insulin resistance was associated with a notable increase of relative TG content in every lipoprotein subclass (Figure 2). ## 3.4. Phospholipids Insulin resistance was positively related to the concentration of phospholipids in total VLDL (Figure 1) and large to extremely large VLDL, and inversely to phospholipids in large to very large HDL (Figure 2). The relationship between insulin resistance and relative number of phospholipids in various lipoprotein subclasses contained notable variation regarding magnitude and direction. There was an inverse association between insulin resistance and relative phospholipid content in very large VLDL, medium VLDL, small VLDL, large LDL, and medium LDL and a positive association between insulin resistance and relative number of phospholipids in very small VLDL, large HDL, and medium HDL (Figure 2). ## 3.5. Lipoprotein Particle Size Insulin resistance related positively to average VLDL particle size and inversely to LDL and HDL particle sizes (Figure 1 and Figure 3). ## 3.6. Fatty Acids (FAs) Insulin resistance was positively associated with total FAs, saturated FAs (SFAs), and monounsaturated FAs (MUFAs) concentrations (Figure 1). Accordingly, insulin resistance was positively associated with the MUFA-to-total-FA and SFA-to-total-FA ratios, whereas the associations of insulin resistance with the polyunsaturated FA (PUFA)-to-total-FA ratio, PUFA-to-MUFA ratio, and degree of FA unsaturation were inverse. Insulin resistance was inversely related to proportions of linoleic acid, docosahexaenoic acid, omega-3 FAs, and omega-6 FAs (Figure 1). ## 3.7. Amino Acids Of amino acids, alanine, valine, and phenylalanine were positively related to insulin resistance (Figure 1). Total branched-chain amino acid (BCAA) concentration was positively related to insulin resistance in the unadjusted model, but the association was no longer significant after adjusting for confounding factors (Table S1). ## 4. Discussion We found insulin resistance in newly diagnosed GDM patients to be associated with increased serum VLDL concentration, total TG concentration, and increased TG content in VLDL, small to medium LDL, and small to medium HDL. Additionally, insulin resistance was related to increased VLDL particle size and decreased HDL and LDL particle sizes. These observations are mostly similar to differences in lipoprotein profile between nonpregnant patients with type 2 diabetes and normoglycemic controls [32]. Additionally, increased concentrations of total FAs, MUFAs, and SFAs and decreased FA unsaturation, PUFA-to-MUFA ratio, and lower proportions of omega-3 FAs, omega-6 FAs, linoleic acid, and docosahexaenoic acid were related to increased insulin resistance. In the current era of personalized medicine, we are expecting the more detailed classification of GDM to improve outcomes [33], although the exact pathogenesis of GDM remains incompletely understood. The associations between insulin resistance and serum metabolome have been characterized previously outside pregnancy, but to extrapolate these findings into a pregnant population has several pitfalls. In addition, previous studies involving pregnant subjects have been scarce [15,16,17,18], and to our best knowledge the associations between maternal insulin resistance in GDM and detailed lipoprotein lipid profile have not previously been studied. Previously when stratified according to pathophysiologic subtype of GDM (insulin resistance, beta-cell deficiency, or mixed), total TG was higher and HDL cholesterol lower in the insulin-resistant compared to the beta-cell-deficient subgroup [34]—largely in agreement with our findings. Conversely, the only difference between beta-cell-deficient GDM patients and normoglycemic controls was higher free fatty acid concentration in the former subgroup [34]. Altogether, the literature supports the hypothesis that hypertriglyceridemia in GDM is driven by maternal insulin resistance [15,17,34]. Here we provide further evidence that insulin resistance is related to variation in the same lipoprotein lipids that have previously been reported to differ between GDM patients and normoglycemic pregnant controls, such as higher VLDL particle concentration and lipid contents, increased proportion of TG in most lipoprotein classes, increased cholesterol in largest lipoproteins, and decreased cholesterol large HDL particles [10,11]. VLDL, initially rich in TG, are formed in the liver and deliver FAs into the peripheral tissues when VLDL TG is hydrolyzed by lipoprotein lipase (LPL) (Figure 4). VLDL size is decreased by the gradual removal of TG, until VLDL is transformed into IDL. LPL activity is enhanced by insulin and, on the contrary, attenuated by insulin resistance. The observed positive associations between insulin resistance and large VLDL particles likely reflect hepatic insulin resistance and increased VLDL synthesis. Also due to prevailing insulin resistance, the formation of smaller VLDL particles, by LPL, is decreased, and likely hence weaker associations were observed between smaller VLDL lipids and insulin resistance. The positive associations between insulin resistance and VLDL particle size and increased large VLDL concentrations are consistent with the previous literature outside pregnancy [35,36]. The associations to IDL were weaker; only the inverse associations to absolute and relative cholesterol in IDL and positive association to relative amount of TG in IDL were significant. While pregnancy in general is related to increase in total lipids in all VLDL subclasses and IDL and a slight increase in VLDL particle size [37,38], we observed strong positive associations between insulin resistance, lipids in larger VLDL subclasses, and especially VLDL particle size. Excessive maternal TG-rich VLDL concentrations are likely driven by increased production in the liver and decreased clearance in maternal peripheral tissues due to reduced LPL activity, both related to insulin resistance [39]. However, to which extent maternal insulin resistance affects placental transfer of lipids is currently not known. As lipoproteins cannot simply cross the placenta, TG in lipoproteins is hydrolyzed into FAs, which are then transferred into the placenta [39]. Placental LPL (pLPL) and endothelial lipase are among the most studied lipases in the placenta [40], but due to natural limitations in accessing human placenta, placental lipoprotein metabolism remains incompletely understood. Based on limited evidence, insulin and hyperglycemia may affect pLPL activity [41]. However, given the greater abundance of maternal peripheral LPL and higher activity of LPL in maternal adipose tissue compared to pLPL [42], the effect of placental metabolism on maternal circulating VLDL concentration is assumed to be small. We found insulin resistance to associate with a higher concentration of cholesterol and phospholipids in large to extremely large VLDL. This observation likely results from higher overall production and decreased clearance of these particles, as the association between insulin resistance and the relative amount of cholesterol and phospholipids in the corresponding particles was inverse or absent. Why insulin resistance was associated with increased proportion of phospholipids in very small VLDL remains to be clarified. In GDM, increased insulin resistance combined with beta-cell deficiency was associated with higher total cholesterol and LDL cholesterol than sole insulin resistance [34]. We did not find associations between insulin resistance and LDL particle or LDL cholesterol concentrations, but insulin resistance was, however, clearly related to a shift toward smaller LDL particles and increased LDL TG content. This is likely promoted by high VLDL TG, which increases cholesterol ester transfer protein (CETP)-mediated transfer of TG from VLDL to LDL [43]. LDL TG and phospholipids are hydrolyzed by hepatic lipase, leading to a decrease in LDL particle size [44]. Accordingly, we found inverse associations between insulin resistance and relative LDL cholesterol content. LDL particle size is not significantly affected by pregnancy [37], but decreased LDL particle size could be a marker of excessive insulin resistance in pregnancies complicated by GDM as found in the present study. Generally, pregnancy is related to increased HDL particle size [37,38,45], increased HDL TG [37,46], and altered protein composition of HDL [38]. We found insulin resistance to be related to increased relative TG and decreased relative cholesterol in HDL, but also decreased HDL particle size. Insulin resistance was also associated with increased absolute TG in small to medium HDL and decreased phospholipids, cholesterol, total lipids, and particle concentration in large to very large HDL. In line with our findings, large HDL particles have been inversely associated and small HDL particles positively associated with insulin resistance [35,36]. In GDM, HDL particle size is decreased [10,11]. By which mechanisms HDL particle distribution is altered in obesity, insulin resistance, and diabetes is not fully known. TG enrichment of HDL has been contributed to increased activity of CETP [44,46], which transfers cholesterol esters from HDL to LDL and VLDL in exchange for TG (Figure 4). Hydrolysis of HDL TG and phospholipids by hepatic lipase promotes formation of smaller HDL particles [32,44]. While increase in HDL TG is physiologic in pregnancy, TG in small and medium HDL are further increased in obese women with GDM compared to normoglycemic controls [10,11]. Insulin resistance in our study was not related to either apolipoprotein A1 (ApoA1), the major apolipoprotein in HDL particles, or apolipoprotein B (ApoB), the major apolipoprotein in TG-rich lipoprotein VLDL and LDL. One copy of ApoB-100 is found in each VLDL, IDL, and LDL particle. Given that LDL is the most abundant of these, the ApoB measure very closely reflects the concentration of LDL. Accordingly, the associations between insulin resistance, ApoB, and ApoA1 resembled the associations between insulin resistance, LDL particle concentration, and HDL particle concentration, respectively. None of those associations were statistically significant in our study. Outside pregnancy, insulin resistance has been related to LDL particle concentrations [35,47], and in a cross-sectional design the apoB-to-ApoA1 ratio has been associated with insulin resistance [48]. Altogether, our data underscore the relationship between insulin resistance and the particle size and TG content of HDL and LDL, rather than mere particle concentrations. There was a clear association between insulin resistance and FA classes. SFAs and MUFAs were positively related and PUFAs, omega-3 FAs, and omega-6 FAs inversely related to insulin resistance. Similarly, in the first trimester, maternal insulin resistance is associated with lower omega-3 FAs [49]. Although all analyses in the present study were made assuming insulin resistance to be the predictor, the causality might be to some extent vice versa. Circulating free FAs have been shown to promote insulin resistance [50,51]. Based on our data, the overall FA profile and insulin resistance are highly intercorrelated in GDM. However, supplementation with fish oil seems not to improve insulin resistance [52]. In women with GDM, the ratios of omega-6 FAs, linoleic acid, and PUFAs to total FAs were significantly lower and the MUFA-to-total-FA ratio higher compared to normoglycemic pregnant controls [10,11]. The differences in absolute concentrations of the FA classes and in the omega-3-to-total-FA ratio were clearly smaller [10,11]. These associations are similar to our observations in associations between insulin resistance and FA classes. During late pregnancy, lipolysis increases so that the circulating FAs may represent mobilization of maternal FA stores. Additionally, in late pregnancy fetal FAs are obtained increasingly via de novo lipogenesis and less by placental transfer [53], although the essential FAs need to be obtained from the mother. We have previously demonstrated that maternal total FAs, SFAs, and MUFAs are related to higher birth weight [13], but whether the proportion of certain FA classes in maternal circulation are of importance regarding fetal growth or perinatal outcomes requires further studies. Although BCAAs are strongly associated to insulin resistance in general [54] and in pregnancy [15], we found insulin resistance to be related positively to total BCAAs only in the unadjusted model. After accounting for confounding factors, valine was the only BCAA associated with insulin resistance. These differences between our data and previous literature [15] might be explained by the smaller sample size in our study, or that all patients in our study had GDM and most required antihyperglycemic medication. Additionally, blood samples were not collected on a fixed gestational age in our study, which may have produced some degree of variation in the data. However, we found clear positive associations between insulin resistance and alanine and phenylalanine, comparable to what was previously reported [15]. Altogether, these findings in recently diagnosed GDM patients demonstrate that the associations between insulin resistance and serum metabolites are similar in GDM compared to those in nonpregnant subjects with type 2 diabetes [32]. Moreover, insulin resistance could significantly contribute to differences in lipoprotein lipid profile observed between GDM patients and pregnant normoglycemic controls [10,11]. Lifestyle interventions have been capable of ameliorating insulin resistance [55], and in one study such an intervention has even reduced the risk of developing GDM, when deployed early in pregnancy [56]. Therefore, focusing on interventions in early pregnancy, or even before conception, to decrease insulin resistance in high-risk populations could yield not only decreased risk of GDM, but also overall healthier metabolic profile with plausible additional benefits. ## Strengths and Limitations The strengths of our study are reasonably large sample size, prospectively collected data, and the use of a validated and widely used NMR protocol. Still, this was a secondary analysis of previously collected data, and drawing all serum samples at the same gestational age could have reduced some variation in the data. Moreover, other metabolites such as acylcarnitines and all individual FAs were not included in the targeted NMR metabolome. We estimated insulin resistance using HOMA2-IR, which is not as precise as hyperinsulinemic clamp. HOMA estimates of insulin resistance are, however, frequently used in research [4,15,16,18] due to the method being less invasive, less time-consuming, and more applicable for studies with a higher number of participants. Use of lithium heparin gel tubes, compared to fluoride citrate tubes [57], for fasting glucose measurement may have resulted in slightly lower glucose values in our analyses. It is unlikely that this has caused any systematic bias in our regression analyses, as the HOMA2-IR values were centered and scaled prior to analyses. ## 5. Conclusions As expected, we found strong positive associations between insulin resistance and serum TG rich lipoproteins in GDM. Moreover, the FA profile was related to insulin resistance. The associations between insulin resistance and fasting amino acids were weaker than expected. These associations were similar to those previously described in type 2 diabetes, but we demonstrated their relevance also in the pregnant population with GDM. Furthermore, small LDL particle size could be a marker of strong insulin resistance in GDM pregnancy. Lifestyle interventions aiming to lower insulin resistance from early pregnancy or before conception could benefit pregnancy outcomes via more advantageous lipid metabolism. ## References 1. Yuen L., Saeedi P., Riaz M., Karuranga S., Divakar H., Levitt N., Yang X., Simmons D.. Projections of the Prevalence of Hyperglycaemia in Pregnancy in 2019 and beyond: Results from the International Diabetes Federation Diabetes Atlas, 9th Edition. Diabetes Res. Clin. Pract. (2019) 157 107841. DOI: 10.1016/j.diabres.2019.107841 2. Langer O., Yogev Y., Most O., Xenakis E.M.J.. Gestational Diabetes: The Consequences of Not Treating. Am. J. Obstet. Gynecol. (2005) 192 989-997. DOI: 10.1016/j.ajog.2004.11.039 3. Powe C.E., Huston Presley L.P., Locascio J.J., Catalano P.M.. Augmented Insulin Secretory Response in Early Pregnancy. Diabetologia (2019) 62 1445-1452. DOI: 10.1007/s00125-019-4881-6 4. Powe C.E., Allard C., Battista M.C., Doyon M., Bouchard L., Ecker J.L., Perron P., Florez J.C., Thadhani R., Hivert M.F.. Heterogeneous Contribution of Insulin Sensitivity and Secretion Defects to Gestational Diabetes Mellitus. Diabetes Care (2016) 39 1052-1055. DOI: 10.2337/dc15-2672 5. Immanuel J., Simmons D., Harreiter J., Desoye G., Corcoy R., Adelantado J.M., Devlieger R., Lapolla A., Dalfra M.G., Bertolotto A.. Metabolic Phenotypes of Early Gestational Diabetes Mellitus and Their Association with Adverse Pregnancy Outcomes. Diabet. Med. (2021) 38 e14413. DOI: 10.1111/dme.14413 6. Benhalima K., Van Crombrugge P., Moyson C., Verhaeghe J., Vandeginste S., Verlaenen H., Vercammen C., Maes T., Dufraimont E., De Block C.. Characteristics and Pregnancy Outcomes across Gestational Diabetes Mellitus Subtypes Based on Insulin Resistance. Diabetologia (2019) 62 2118-2128. DOI: 10.1007/s00125-019-4961-7 7. Metzger B.E., Phelps R.L., Freinkel N., Navickas I.A.. Effects of Gestational Diabetes on Diurnal Profiles of Plasma Glucose, Lipids, and Individual Amino Acids. Diabetes Care (1980) 3 402-409. DOI: 10.2337/diacare.3.3.402 8. Pappa K.I., Vlachos G., Theodora M., Roubelaki M., Angelidou K., Antsaklis A.. Intermediate Metabolism in Association with the Amino Acid Profile during the Third Trimester of Normal Pregnancy and Diet-Controlled Gestational Diabetes. Am. J. Obstet. Gynecol. (2007) 196 65.e1-65.e5. DOI: 10.1016/j.ajog.2006.06.094 9. Huynh J., Xiong G., Bentley-Lewis R.. A Systematic Review of Metabolite Profiling in Gestational Diabetes Mellitus. Diabetologia (2014) 57 2453-2464. DOI: 10.1007/s00125-014-3371-0 10. Mokkala K., Vahlberg T., Houttu N., Koivuniemi E., Laitinen K.. Distinct Metabolomic Profile Because of Gestational Diabetes and Its Treatment Mode in Women with Overweight and Obesity. Obesity (2020) 28 1637-1644. DOI: 10.1002/oby.22882 11. White S.L., Pasupathy D., Sattar N., Nelson S.M., Lawlor D.A., Briley A.L., Seed P.T., Welsh P., Poston L.. Metabolic Profiling of Gestational Diabetes in Obese Women during Pregnancy. Diabetologia (2017) 60 1903-1912. DOI: 10.1007/s00125-017-4380-6 12. Huhtala M.S., Tertti K., Pellonperä O., Rönnemaa T.. Amino Acid Profile in Women with Gestational Diabetes Mellitus Treated with Metformin or Insulin. Diabetes Res. Clin. Pract. (2018) 146 8-17. DOI: 10.1016/j.diabres.2018.09.014 13. Huhtala M.S., Tertti K., Rönnemaa T.. Serum Lipids and Their Association with Birth Weight in Metformin and Insulin Treated Patients with Gestational Diabetes. Diabetes Res. Clin. Pract. (2020) 170 108456. DOI: 10.1016/j.diabres.2020.108456 14. Kadakia R., Nodzenski M., Talbot O., Kuang A., Bain J.R., Muehlbauer M.J., Stevens R.D., Ilkayeva O.R., O’Neal S.K., Lowe L.P.. Maternal Metabolites during Pregnancy Are Associated with Newborn Outcomes and Hyperinsulinaemia across Ancestries. Diabetologia (2019) 62 473-484. DOI: 10.1007/s00125-018-4781-1 15. Sandler V., Reisetter A.C., Bain J.R., Muehlbauer M.J., Nodzenski M., Stevens R.D., Ilkayeva O., Lowe L.P., Metzger B.E., Newgard C.B.. Associations of Maternal BMI and Insulin Resistance with the Maternal Metabolome and Newborn Outcomes. Diabetologia (2017) 60 518-530. DOI: 10.1007/s00125-016-4182-2 16. Liu Y., Kuang A., Talbot O., Bain J.R., Muehlbauer M.J., Hayes M.G., Ilkayeva O.R., Lowe L.P., Metzger B.E., Newgard C.B.. Metabolomic and Genetic Associations with Insulin Resistance in Pregnancy. Diabetologia (2020) 63 1783-1795. DOI: 10.1007/s00125-020-05198-1 17. Furse S., Koulman A., Ozanne S.E., Poston L., White S.L., Meek C.L.. Altered Lipid Metabolism in Obese Women with Gestational Diabetes and Associations with Offspring Adiposity. J. Clin. Endocrinol. Metab. (2022) 107 E2825-E2832. DOI: 10.1210/clinem/dgac206 18. Hellmuth C., Lindsay K.L., Uhl O., Buss C., Wadhwa P.D., Koletzko B., Entringer S.. Association of Maternal Prepregnancy BMI with Metabolomic Profile across Gestation. Int. J. Obes. (2017) 41 159-169. DOI: 10.1038/ijo.2016.153 19. Tertti K., Ekblad U., Koskinen P., Vahlberg T., Rönnemaa T.. Metformin vs. Insulin in Gestational Diabetes. A Randomized Study Characterizing Metformin Patients Needing Additional Insulin. Diabetes Obes. Metab. (2013) 15 246-251. DOI: 10.1111/dom.12017 20. Pellonperä O., Rönnemaa T., Ekblad U., Vahlberg T., Tertti K.. The Effects of Metformin Treatment of Gestational Diabetes on Maternal Weight and Glucose Tolerance Postpartum—A Prospective Follow-up Study. Acta Obstet. Gynecol. Scand. (2016) 95 79-87. DOI: 10.1111/aogs.12788 21. Levy J.C., Matthews D.R., Hermans M.P.. Correct Homeostasis Model Assessment (HOMA) Evaluation Uses the Computer Program. Diabetes Care (1998) 21 2191-2192. DOI: 10.2337/diacare.21.12.2191 22. Wallace T.M., Levy J.C., Matthews D.R.. Use and Abuse of HOMA Modeling. Diabetes Care (2004) 27 1487-1495. DOI: 10.2337/diacare.27.6.1487 23. Radaelli T., Farrell K.A., Huston-Presley L., Amini S.B., Kirwan J.P., McIntyre H.D., Catalano P.M.. Estimates of Insulin Sensitivity Using Glucose and C-Peptide from the Hyperglycemia and Adverse Pregnancy Outcome Glucose Tolerance Test. Diabetes Care (2010) 33 490-494. DOI: 10.2337/dc09-1463 24. Soininen P., Kangas A.J., Wurtz P., Tukiainen T., Tynkkynen T., Laatikainen R., Jarvelin M.-R., Kahonen M., Lehtimaki T., Viikari J.. High-Throughput Serum NMR Metabonomics for Cost-Effective Holistic Studies on Systemic Metabolism. Analyst (2009) 134 1781-1785. DOI: 10.1039/b910205a 25. Chrysafi P., Perakakis N., Farr O.M., Stefanakis K., Peradze N., Sala-Vila A., Mantzoros C.S.. Leptin Alters Energy Intake and Fat Mass but Not Energy Expenditure in Lean Subjects. Nat. Commun. (2020) 11 1-15. DOI: 10.1038/s41467-020-18885-9 26. Soininen P., Kangas A.J., Würtz P., Suna T., Ala-Korpela M.. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Cardiovascular Epidemiology and Genetics. Circ. Cardiovasc. Genet. (2015) 8 192-206. DOI: 10.1161/CIRCGENETICS.114.000216 27. Vehtari A., Mäkinen V.-P., Soininen P., Ingman P., Mäkelä S.M., Savolainen M.J., Hannuksela M.L., Kaski K., Ala-Korpela M.. A Novel Bayesian Approach to Quantify Clinical Variables and to Determine Their Spectroscopic Counterparts in 1H NMR Metabonomic Data. BMC Bioinform. (2007) 8 S8. DOI: 10.1186/1471-2105-8-S2-S8 28. Würtz P., Kangas A.J., Soininen P., Lawlor D.A., Davey Smith G., Ala-Korpela M.. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Large-Scale Epidemiology: A Primer on-Omic Technologies. Am. J. Epidemiol. (2017) 186 1084-1096. DOI: 10.1093/aje/kwx016 29. Würtz P., Soininen P.. Reply to: “Methodological Issues Regarding: ‘A Third of Nonfasting Plasma Cholesterol Is in Remnant Lipoproteins: Lipoprotein Subclass Profiling in 9293 Individuals’”. Atherosclerosis (2020) 302 59-61. DOI: 10.1016/j.atherosclerosis.2020.03.028 30. Würtz P., Havulinna A.S., Soininen P., Tynkkynen T., Prieto-Merino D., Tillin T., Ghorbani A., Artati A., Wang Q., Tiainen M.. Metabolite Profiling and Cardiovascular Event Risk: A Prospective Study of 3 Population-Based Cohorts. Circulation (2015) 131 774-785. DOI: 10.1161/CIRCULATIONAHA.114.013116 31. Deelen J., Kettunen J., Fischer K., van der Spek A., Trompet S., Kastenmüller G., Boyd A., Zierer J., van den Akker E.B., Ala-Korpela M.. A Metabolic Profile of All-Cause Mortality Risk Identified in an Observational Study of 44,168 Individuals. Nat. Commun. (2019) 10 1-8. DOI: 10.1038/s41467-019-11311-9 32. Vergès B.. Pathophysiology of Diabetic Dyslipidaemia: Where Are We?. Diabetologia (2015) 58 886-899. DOI: 10.1007/s00125-015-3525-8 33. Powe C.E., Hivert M.F., Udler M.S.. Defining Heterogeneity among Women with Gestational Diabetes Mellitus. Diabetes (2020) 69 2064-2074. DOI: 10.2337/dbi20-0004 34. Layton J., Powe C., Allard C., Battista M.C., Doyon M., Bouchard L., Perron P., Wessel J., Hivert M.F.. Maternal Lipid Profile Differs by Gestational Diabetes Physiologic Subtype. Metabolism (2019) 91 39-42. DOI: 10.1016/j.metabol.2018.11.008 35. Garvey W.T., Kwon S., Zheng D., Shaughnessy S., Wallace P., Hutto A., Pugh K., Jenkins A.J., Klein R.L., Liao Y.. Effects of Insulin Resistance and Type 2 Diabetes on Lipoprotein Subclass Particle Size and Concentration Determined by Nuclear Magnetic Resonance. Diabetes (2003) 52 453-462. DOI: 10.2337/diabetes.52.2.453 36. Wang J., Stančáková A., Soininen P., Kangas A.J., Paananen J., Kuusisto J., Ala-Korpela M., Laakso M.. Lipoprotein Subclass Profiles in Individuals with Varying Degrees of Glucose Tolerance: A Population-Based Study of 9399 Finnish Men. J. Intern. Med. (2012) 272 562-572. DOI: 10.1111/j.1365-2796.2012.02562.x 37. Wang Q., Würtz P., Auro K., Mäkinen V.-P., Kangas A.J., Soininen P., Tiainen M., Tynkkynen T., Jokelainen J., Santalahti K.. Metabolic Profiling of Pregnancy: Cross-Sectional and Longitudinal Evidence. BMC Med. (2016) 14. DOI: 10.1186/s12916-016-0733-0 38. Melchior J.T., Swertfeger D.K., Morris J., Street S.E., Warshak C.R., Welge J.A., Remaley A.T., Catov J.M., Davidson W.S., Woollett L.A.. Pregnancy Is Accompanied by Larger High Density Lipoprotein Particles and Compositionally Distinct Subspecies. J. Lipid Res. (2021) 62 100107. DOI: 10.1016/j.jlr.2021.100107 39. Herrera E., Desoye G.. Maternal and Fetal Lipid Metabolism under Normal and Gestational Diabetic Conditions. Horm. Mol. Biol. Clin. Investig. (2016) 26 109-127. DOI: 10.1515/hmbci-2015-0025 40. Barrett H.L., Kubala M.H., Scholz Romero K., Denny K.J., Woodruff T.M., McIntyre H.D., Callaway L.K., Nitert M.D.. Placental Lipases in Pregnancies Complicated by Gestational Diabetes Mellitus (GDM). PLoS ONE (2014) 9 1-8. DOI: 10.1371/journal.pone.0104826 41. Magnusson-Olsson A.L., Hamark B., Ericsson A., Wennergren M., Jansson T., Powell T.L.. Gestational and Hormonal Regulation of Human Placental Lipoprotein Lipase. J. Lipid Res. (2006) 47 2551-2561. DOI: 10.1194/jlr.M600098-JLR200 42. Heerwagen M.J.R., Gumina D.L., Hernandez T.L., Van Pelt R.E., Kramer A.W., Janssen R.C., Jensen D.R., Powell T.L., Friedman J.E., Winn V.D.. Placental Lipoprotein Lipase Activity Is Positively Associated with Newborn Adiposity. Placenta (2018) 64 53-60. DOI: 10.1016/j.placenta.2018.03.001 43. Lahdenperä S., Syvänne M., Kahri J., Taskinen M.R.. Regulation of Low-Density Lipoprotein Particle Size Distribution in NIDDM and Coronary Disease: Importance of Serum Triglycerides. Diabetologia (1996) 39 453-461. DOI: 10.1007/BF00400677 44. Barter P.J., Brewer H.B., Chapman M.J., Hennekens C.H., Rader D.J., Tall A.R.. Cholesteryl Ester Transfer Protein: A Novel Target for Raising HDL and Inhibiting Atherosclerosis. Arterioscler. Thromb. Vasc. Biol. (2003) 23 160-167. DOI: 10.1161/01.ATV.0000054658.91146.64 45. Silliman K., Tall A.R., Kretchmer N., Forte T.M.. Unusual High-Density Lipoprotein Subclass Distribution during Late Pregnancy. Metabolism (1993) 42 1592-1599. DOI: 10.1016/0026-0495(93)90156-I 46. Iglesias A., Montelongo A., Herrera E., Lasunción M.A.. Changes in Cholesteryl Ester Transfer Protein Activity during Normal Gestation and Postpartum. Clin. Biochem. (1994) 27 63-68. DOI: 10.1016/0009-9120(94)90013-2 47. Goff D.C., D’Agostino R.B., Haffner S.M., Otvos J.D.. Insulin Resistance and Adiposity Influence Lipoprotein Size and Subclass Concentrations. Results from the Insulin Resistance Atherosclerosis Study. Metabolism (2005) 54 264-270. DOI: 10.1016/j.metabol.2004.09.002 48. Sierra-Johnson J., Romero-Corral A., Somers V.K., Lopez-Jimenez F., Walldius G., Hamsten A., Hellénius M.L., Fisher R.M.. ApoB/ApoA-I Ratio: An Independent Predictor of Insulin Resistance in US Non-Diabetic Subjects. Eur. Heart J. (2007) 28 2637-2643. DOI: 10.1093/eurheartj/ehm360 49. Bandres-Meriz J., Majali-Martinez A., Hoch D., Morante M., Glasner A., van Poppel M.N.M., Desoye G., Herrera E.. Maternal C-Peptide and Insulin Sensitivity, but Not BMI, Associate with Fatty Acids in the First Trimester of Pregnancy. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms221910422 50. Boden G., Chen X., Ruiz J., White J.V., Rossetti L.. Mechanisms of Fatty Acid-Induced Inhibition of Glucose Uptake. J. Clin. Investig. (1994) 93 2438-2446. DOI: 10.1172/JCI117252 51. Sivan E., Homko C.J., Whittaker P.G., Reece E.A., Chen X., Boden G.. Free Fatty Acids and Insulin Resistance during Pregnancy. J. Clin. Endocrinol. Metab. (1998) 83 2338-2342. DOI: 10.1210/jc.83.7.2338 52. Pellonperä O., Mokkala K., Houttu N., Vahlberg T., Koivuniemi E., Tertti K., Rönnemaa T., Laitinen K.. Efficacy of Fish Oil and/or Probiotic Intervention on the Incidence of Gestational Diabetes Mellitus in an At-Risk Group of Overweight and Obese Women: A Randomized, Placebo-Controlled, Double-Blind Clinical Trial. Diabetes Care (2019) 42 1009-1017. DOI: 10.2337/dc18-2591 53. Desoye G., Herrera E.. Adipose Tissue Development and Lipid Metabolism in the Human Fetus: The 2020 Perspective Focusing on Maternal Diabetes and Obesity. Prog. Lipid Res. (2021) 81 101082. DOI: 10.1016/j.plipres.2020.101082 54. Wang Q., Holmes M.V., Smith G.D., Ala-Korpela M.. Genetic Support for a Causal Role of Insulin Resistance on Circulating Branched-Chain Amino Acids and Inflammation. Diabetes Care (2017) 40 1779-1786. DOI: 10.2337/dc17-1642 55. Uusitupa M., Lindi V., Louheranta A., Salopuro T., Lindström J., Tuomilehto J.. Long-Term Improvement in Insulin Sensitivity by Changing Lifestyles of People with Impaired Glucose Tolerance. Diabetes (2003) 52 2532-2538. DOI: 10.2337/diabetes.52.10.2532 56. Koivusalo S.B., Rönö K., Klemetti M.M., Roine R.P., Lindström J., Erkkola M., Kaaja R.J., Pöyhönen-Alho M., Tiitinen A., Huvinen E.. Gestational Diabetes Mellitus Can Be Prevented by Lifestyle Intervention: The Finnish Gestational Diabetes Prevention Study (RADIEL). Diabetes Care (2016) 39 24-30. DOI: 10.2337/dc15-0511 57. Dimeski G., Yow K.S., Brown N.N.. What Is the Most Suitable Blood Collection Tube for Glucose Estimation?. Ann. Clin. Biochem. (2015) 52 270-275. DOI: 10.1177/0004563214544708
--- title: Effects of Combined Intervention of rTMS and Neurotransmitter Drugs on the Brain Functional Networks in Patients with Cognitive Impairment authors: - Mengyun Li - Zhiming Qin - Haifeng Chen - Zhiyuan Yang - Lianlian Wang - Ruomeng Qin - Hui Zhao - Feng Bai journal: Brain Sciences year: 2023 pmcid: PMC10046663 doi: 10.3390/brainsci13030419 license: CC BY 4.0 --- # Effects of Combined Intervention of rTMS and Neurotransmitter Drugs on the Brain Functional Networks in Patients with Cognitive Impairment ## Abstract Alzheimer’s disease (AD) causes extensive neural network dysfunction. Memantine and donepezil are commonly used as monotherapy or in combination with non-drug interventions, such as repetitive transcranial magnetic stimulation (rTMS), for its treatment. However, no studies have reported any differences between the effects of combined neurotransmitter and rTMS interventions versus rTMS alone on the brain networks of patients with cognitive impairment. Therefore, it is crucial to explore the advantages of different intervention methods to guide clinical practice. We used resting-state functional magnetic resonance imaging (rs-fMRI) to investigate the impact of neurotransmitter superimposed rTMS and rTMS alone on the brain functional network of patients with cognitive impairment. We divided patients with cognitive impairment who had received rTMS into two groups based on whether they received neurotransmitters: the combined intervention group and the rTMS-alone intervention group. We conducted rs-fMRI scans and comprehensively assessed cognitive function in these patients. To examine the effects of the superimposed interventions, we utilized independent component analysis to evaluate the functional connectivity of brain networks in these patients. Compared to the rTMS-alone intervention group, co-intervention of neurotransmitter drugs and rTMS exhibited potential for cognitive enhancement via the reconstructed inter-network connectivity of the cerebellum and the enhanced intra-network connectivity of the frontal-parietal regions in these patients with cognitive impairment. We hypothesized that the combination of neurotransmitter drugs and rTMS intervention could have greater clinical benefits than rTMS intervention alone, leading to improved cognitive function in patients with cognitive impairment. ## 1. Introduction Normal aging and neurodegenerative diseases lead to changes in brain structure and function, and decreased strength of functional network connections which are closely related to cognitive decline. Alzheimer’s disease (AD) manifested as cognitive dysfunction such as memory, executive function, and visual space function which is a typical representative of degenerative diseases. There is already evidence that AD is a complex polygenic disorder that is the result of multiple factors acting abnormally [1,2,3]. Although the pathogenesis of AD is not fully understood, current theories have guided the exploration of clinical interventions to delay the progression of AD. At present, the drugs used clinically mainly include acetylcholinesterase inhibitors (donepezil, etc.) and NMDA receptor antagonists (memantine). Acetylcholinesterase inhibitors (CHEIs) prevent acetylcholine from being destroyed in patients by blocking acetylcholinesterase activity, compensating for acetylcholine in the brain, thereby improving cognitive function [4]. Memantine is an N-methyl-D-aspartate (NMDA) receptor antagonist that non-competitively blocks NMDA receptors, reduces glutamate-induced NMDA receptor overexcitation, prevents apoptosis, and is effective in improving memory in patients with AD [5]. Prior research has indicated that administration of donepezil to patients with mild cognitive impairment (MCI) for a duration of 3 to 6 months may result in enhanced brain activation during memory processing in the left inferior frontal gyrus, a region linked to attention and memory processes such as encoding and retrieving short- and long-term memory [6]. Additional research demonstrated that patients with MCI who received donepezil for 3 months displayed increased activation in the right medial temporal lobe, including the hippocampus/parahippocampal gyrus, during episodic memory coding tasks. Increased functional connectivity was significantly linked to improved fMRI task performance. Results suggest that donepezil may partially “normalize” brain activation patterns and connections during tasks, and these effects are linked to cognitive changes [7]. Furthermore, there is growing evidence that memantine treatment can have a beneficial effect in patients with moderate to severe Alzheimer’s disease (AD), as it can lead to increased resting default mode network (DMN) activity in the precuneus region over 6 months [8]. Therefore, the drugs may improve cognitive function by improving functional connectivity of brain networks. Non-drug interventions have been shown to have important implications for slowing the progression of Alzheimer’s disease (AD) and improving cognitive function. Repeated transcranial magnetic stimulation (rTMS) has been shown as one of the effective strategies to improve cognitive function in patients with cognitive impairment [9]. Previous studies have demonstrated that repetitive transcranial magnetic stimulation (rTMS) directly stimulates cortical neurons by generating magnetic fields that induce electrical currents and activate the synaptic activities of neuronal circuits in the central nervous system [10]. Moreover, an rTMS study located in the left angular gyrus demonstrated that one month of treatment with rTMS was effective in improving patient episodic memory [3]. It was reported that rTMS can also affect cognitive function by affecting functional connectivity of brain networks. For example, studies have shown that rTMS modulation directed upon the precuneus for 2 weeks could improve middle cognitive region (HIPc) connectivity with the left parahippocampal gyrus and posterior perceptual region (HIPp) connectivity with the left middle temporal gyrus, potentially improving episodic memory [11]. Moreover, a study conducted by Wang et al. [ 2014] demonstrated that repetitive transcranial magnetic stimulation (rTMS) administered to the posterior parietal cortex (PPC) five times daily resulted in a significant increase in functional connectivity within hippocampus-centric networks and an improvement in cognitive function [12,13]. As an effective non-invasive intervention that can be combined with drugs for the treatment of patients with clinical cognitive impairment, this study was designed to explore the differences between the effect of neurotransmitter-drug-superimposed rTMS treatment and rTMS treatment alone on the functional connectivity of patients’ brain networks. Independent component analysis (ICA) is an effective tool for separating statistical independent source signals in blood oxygen level dependent (BOLD) data. In the absence of a prior seed region, the separation of different functional magnetic resonance (fMRI) signal sources by maximizing the non-Gaussian nature of the source signal can identify intracranial function during rest. These networks are often referred to as “intrinsically connected networks” (ICNs) or “rest state networks” (RSNs) [14,15]. This study attempted to explore: (i) how neurotransmitter-superimposed rTMS interventions affect the functional connections of brain networks; and (ii) whether the changes in the functional connections of brain networks were associated with behavior performance in these patients. This study will provide a deeper understanding of the impact of combined use of interventions on brain functional networks in patients with cognitive impairments. ## 2.1. Participants This study was approved by the Ethics Committee of Nanjing Drum Tower Hospital and the written informed consent of all patients was obtained before entering the study. In this study, patients with cognitive impairment who had received rTMS treatment were screened in the Department of Neurology, Drum Tower Hospital, Nanjing University School of Medicine. Four of the patients were excluded due to excessive head movement (>3 degrees) during fMRI scans, and two patients were excluded due to incomplete behavioral data. As shown in Table 1, the study enrolled 28 patients with cognitive impairment (all Chinese Han and right-handed), including 11 males and 17 females. These participants included patients with MCI and AD, and the diagnostic criteria for patients with AD were based on standards developed by the National Neurological and Communication Disorders and Stroke Institute and the Association for AD-Related Diseases (NINCDS-ADRDA Criteria) [16], while combining cerebrospinal fluid pathology and neuroimaging student markers for judgment. Patients with MCI were diagnosed based on criteria established by the previous studies [17,18,19,20,21,22]. According to research conducted by May and colleagues (Petersen et al., 1999, published in the Archives of Neurology), the diagnostic criteria for Alzheimer’s disease include: (i) a memory complaint, preferably confirmed by an informant; (ii) objective evidence of memory impairment for the individual’s age; (iii) generally preserved cognitive function for the individual’s age; (iv) activities of daily living that are essentially intact; and (v) the individual is not considered to be demented. All participant exclusion criteria included: (i) a history of neurological or psychiatric disorders (brain tumors, epilepsy, Parkinson’s disease, severe anxiety and depression, thyroid dysfunction, or other neurological or psychiatric disorders that may contribute to memory loss); (ii) any MRI contraindications or poor image quality; (iii) all neuropsychological assessments could not be completed. ## 2.2. Repetitive Transcranial Magnetic Stimulation Intervention Our intervention targets are provided by another study in our research group, who used the left hippocampus as the seed, calculated by seed-based functional connectivity analysis, and ultimately located the intervention target in the left corner gyrus (MNI: −45, −67, 38). rTMS was delivered using a commercially available magnetic stimulator (CCY-IV model; YIRUIDE Inc., Wuhan, China) with a 70 mm figure-eight coil and an electromyography device. Each stimulation session consisted of forty circulations of 2 s delivered at 20 Hz spaced by 28 s of no stimulation, for a grand total of 1600 stimulations. The stimulation target is defined as a sphere of 6 mm radius centered at MNI coordinates (MNI: −45, −67, 38) and the treatments lasted about 20 min. For detailed information, please refer to the study of Yang 2022, published in J Alzheimers Dis [3]. Participants in this study were all treated with rTMS for a period of four weeks, followed by an fMRI scan and a complete neuropsychological assessment. At the same time, statistics of patients taking neurotransmitter drugs were captured. To the best of our knowledge, patients were administered a daily dose of 10 mg of donepezil or 10 mg of memantine, and the average duration of medication was more than eight weeks. The neurotransmitters refer to the AChEIs (e.g., donepezil) and NMDA receptor antagonist (e.g., memantine). ## 2.3. Neuropsychological Assessment To assess the behavioral effects of the treatment, we employed a standardized neuropsychological test protocol that includes global cognitive assessment and multiple cognitive domain examinations. We also completed the Clinical Dementia Rating scale to assess the extent of cognitive impairment in participants. We used the Activity of Daily Living Scale to assess the patients’ ability to perform activities of daily living. Global cognitive function was assessed by MMSE and Montreal Cognitive Assessment Beijing (MoCA-BJ). The raw test scores were converted to Z-scores, which were used to calculate the compound cognitive index. Episodic memory was calculated as the average of the Z-scores from the AVLT-DR scores and the Webster Memory Scale-Visual Reproduction-delayed Recall (VR-DR). AVLT-DR was determined based on the number of words retrieved 20 min after learning trials of 15 words, and VR-DR assessed ability to reproduce difficult to verbalize designs after a brief exposure. The information processing speed was calculated as the average Z-scores of the Trail Making Test-A (TMT-A) and the Stroop Color and Word Tests A and B (Stroop A and B). Language function included the Boston Naming Test and the Category Verbal Fluency test. The execution function was a composite score of the average Z-scores of the digital Span Test-backward, Trail Making Test-B (TMT-B), and Stroop Color and Word tests C (Stroop C). The visual spatial function was the average of a composite score that includes the Z-scores of the Clock Drawing Test and the Visual Replication-copy Test [3]. ## 2.4. fMRI Acquisition All participants were examined on a Philips 3.0T scanner (Philips Medical System). The inspection protocol included a high-resolution T1-weighted turbine gradient echo sequence (repeat time [TR] = 9.8 ms, flip angle [FA] = 8°, echo time [TE] = 4.6 ms, FOV = 250 × 250 mm2, number of slices = 192, acquisition Matrix = 256 × 256, thickness = 1.0 mm), the FLAIR Sequence (TR = 4.500 ms, TE = 333 ms, Time Interval [TI] = 1.600 ms, number of slices = 200, voxel size = 0.95 × 0.95 × 0.95 mm3, acquisition matrix = 270 × 260) and the resting-state function scans imaging sequence (TR = 2000 ms, TE = 30 ms, FA = 90, acquisition matrix = 64 × 64, number of slices = 35, thickness = 4.0 mm, FOV = 240 × 240 mm2) [3]. ## 2.5. Data Preprocessing We used statistical parametric mapping (SPM)-12 software (http://www.fil.ion.ucl.ac.uk/spm) and the Resting State fMRI Data Processing Assistant (http://rfmri.org/DPARSF) to preprocess resting BOLD data [23]. The following steps were adopted: (i) Converting the data from the medical digital imaging format to the NIfTI format. ( ii) The first 10 volumes were discarded to allow the signal to reach equilibrium and the participants to adapt to the environment, leaving 220 images for further processing. ( iii) The remaining 220 time points were corrected for the acquisition time delay between different slices, ensuring that all voxels in a volume were scanned instantaneously at the same moment. ( iv) The rigid-body head movement during scans was corrected. Excessive motion was defined as more than 3.0 mm of translation or greater than a 3.0° rotation in any direction and two patients were excluded; thus, 30 subjects were included for further analysis. ( v) The functional images were then spatially normalized to the Montreal Neurological Institute (MNI) space using a unified segmentation algorithm. Then, the data were resampled to an isotropic resolution of 3 mm using the parameters estimated during unified segmentation [24]. ( vi) Spatial smoothing: the normalized images were spatially smoothed using an isotropic Gaussian filter with a full width half maximum (FWHM) of 6 mm to reduce spatial noises [25,26,27]. ## 2.6. Identification of Resting-State Networks We used the GIFTv4.0 software (ICA in the fMRI toolbox, http://icatb.sourceforge.net, 26 February 2023) to analyze the preprocessed images, following three main steps: (i) data reduction, (ii) group ICA, and (iii) back reconstruction. First, principal components analysis (PCA) was applied to reduce the data dimensionality for each subject. The reduced data from all subjects were then concatenated and entered into a second data reduction step using PCA. Second, the simplified data were then estimated using the Informax algorithm to estimate the spatially independent components (34 in this study), and Informax was run 20 times in the ICASSO toolbox algorithms to increase the reliability of independent component decomposition. Finally, individual subject components were back reconstructed from the group components using the GICA approach, during which the aggregate components and the results from the data reduction step were used to compute the individual subject components. Each back-reconstructed component consists of a spatial z-map reflecting the component’s functional connectivity pattern across space and an associated time course reflecting the component’s activity across time [2,15,28]. The group-level components corresponding to the visual network (VN), the auditory network (AUN), the sensorimotor network (SMN), the left frontoparietal network (LFPN), the right frontoparietal network (RFPN), the cerebellum network (CN), the anterior default mode network (aDMN), the posterior DMN (pDMN), the dorsal attention network (DAN), the ventral attentional network (VAN), and the salience network (SN) were selected by visual inspection and confirmed using the template-matching procedure. The template was provided in GIFT software (the RSN template), and the map of each component was spatially correlated with a specific network template. The component with the largest spatial correlation coefficients with each of these templates was chosen and reconfirmed by visual inspection. Finally, we identified 12 networks for subsequent analysis (the detailed descriptions are presented in the Results section, and the operation process is shown in Figure 1) [29]. ## 2.7. Inter-Network Connectivity Analysis Temporal correlations between different resting-state networks (RSNs) were calculated to investigate the effects of rTMS and neurotransmitter drug interventions on brain functional networks. The time course of each RSN was extracted from the ICA procedure, and the time courses of each pair of the 12 RSNs were used to calculate the functional network connectivity (FNC) and normalized with the Fisher r-to-z transformation. A two-sample t-test was employed to compare group differences in the FNCs at an uncorrected significance level of $p \leq 0.05$ [2,26]. ## 2.8. Intra-Network Connectivity Analysis First, a single-sample t-test ($p \leq 0.05$, FWE-corrected) was used in the two groups to create a sample-specific component map and a network mask. A two-sample t-test was then conducted to compare the intra-network difference between the two groups for each component within the corresponding network mask. The statistical threshold was as follows: voxel level, $$p \leq 0.005$$; cluster level, $p \leq 0.05$ (FWE-corrected), and cluster size >10 voxels. Therefore, a t-value map with a significant between-group difference was obtained, and the results were displayed using xjview software [15,26]. ## 2.9. Statistical Analysis We used SPSS software (IBM SPSS Statistics 26) for data statistics, and the measurement data conforming to the normal distribution were expressed as the mean ± standard deviation, and the non-normal distribution data were expressed as the median in (lower quartile, upper quartile). The measurement data between the two groups were normalized using a stand-alone sample t-test and the non-normally distributed data were tested using Mann–Whitney U. The Spearman rank correlation analysis between brain functional characteristics and cognitive performance was further performed. The thresholds were set at $p \leq 0.05.$ ## 3.1. Demographic and Neuropsychological Data In this study, four subjects were excluded from the analysis due to excessive head movement ($$n = 4$$). In addition, two subjects were not included in the cognition analysis due to incomplete cognitive scale data ($$n = 2$$). Ultimately, 30 subjects were included in the imaging analysis, and 28 subjects from the rTMS group were included in the cognition analysis (Figure 1 and Table 1). As illustrated in Table 1, the study enrolled 28 patients with cognitive impairment (11 males and 17 females), and the demographic data indicated no significant between-group differences in age, sex, educational level, MoCA scores, information processing speed, language, or visuospatial processing function ($p \leq 0.05$). The rTMS-alone intervention group exhibited significantly lower scores in general cognition (MMSE, $$p \leq 0.002$$), episodic memory ($$p \leq 0.022$$), and executive function ($$p \leq 0.002$$) than the combined intervention group. ## 3.2. ICA and Determination of RSNs Out of 34 components, we screened 12 networks that were in line with the previous studies [25,28,30,31,32]). The left and right frontoparietal network (LFPN, RFPN) consists primarily of the dorsolateral prefrontal cortex (DLPFC) and posterior parietal cortex (PPC). The anterior default mode network (aDMN) mainly involves the medial prefrontal (mPFC) and anterior cingulate cortex (ACC). The posterior DMN (pDMN) primarily includes the posterior cingulate cortex (PCC)/precuneus (Pcu) and the bilateral lateral parietal cortex. The sensorimotor network (SMN) mainly includes the bilateral precentral gyrus (PreCG) and postcentral gyrus (PostCG). The salience network (SN) mainly includes the dorsal ACC (dACC), bilateral VLPFC, and anterior insula (AI). The dorsal attention network (DAN) includes the interparietal sulcus and the junction of the precentral superior frontal sulcus bilaterally. Core regions of the ventral networks (VANs) are the inferior parietal lobule (IPL) and the adjacent temporo-parietal junction (TPJ). The visual network (VN) consists of the bilateral occipital lobe (OL). The auditory network (AN) primarily includes the bilateral superior temporal gyrus (STG) [2,27,33,34,35,36,37,38] (as shown in Figure 2). ## 3.3. Combined Intervention Reconstruct Inter-Network Connectivity of Cerebellum In comparison to the rTMS-alone intervention group, the combined intervention group demonstrated functional network connectivity (FNC) enhancement, including increased connectivity of CN-SN and subN-VAN ($p \leq 0.05$, uncorrected). Additionally, significantly reduced inter-network connectivity was observed in the CN and aDMN ($p \leq 0.05$, uncorrected). Figure 3 and Table 2 depict between-group differences in inter-network connectivity. ## 3.4. Combined Intervention Improves the Strength of Intra-Network Connectivity within Frontal-Parietal Regions The combined intervention group exhibited significantly increased intra-network connectivity when compared to the rTMS-alone intervention group, primarily involving frontal-parietal regions (Figure 4 and Table 3, $p \leq 0.005$, uncorrected). Specifically, (i) functional connectivity of Frontal_Sup_Medial_L/Angular_L/Frontal_Mid_L/Precuneus_L of LFPN exhibited significant increases. ( ii) Significant increases were observed in Cingulum_Ant_R of RFPN. ( iii) Connectivity improvements in Postcentral_L/Postcentral_R/Parietal_Inf_L/Frontal_Mid_R/Precentral_L were associated with DAN. ## 3.5. Differential Inter-Network/Intra-Network Connectivity Patterns and Behavioral Significance Inter-Network connectivity: The Spearman rank correlation analysis revealed that the connectivity between CN and SN was correlated with multiple cognitive domains, including general cognition (i.e., MMSE) (r = −0.398, $$p \leq 0.036$$), episodic memory (r = −0.388, $$p \leq 0.041$$), and executive function (r = −0.532, $$p \leq 0.004$$) (Figure 5A). Intra-network connectivity: this study determined correlations between the 12 RSNs and conducted cognitive assessments, and found that increased functional connectivity within the LFPN involving the left angular gyrus ($r = 0.697$, $$p \leq 0.025$$) and left precuneus ($r = 0.782$, $$p \leq 0.008$$) was positively correlated with episodic memory. Decreased functional connectivity of the left precentral gyrus within the DAN was negatively associated with executive function (r = −0.685, $$p \leq 0.029$$). Additionally, increased functional connectivity of the anterior cingulate and paracingulate gyri within RFPN was positively correlated with MMSE scores (Figure 5B). ## 4. Discussion This study is the first to explore the effects of combining neurotransmitter drugs with rTMS intervention on brain network in patients with cognitive impairment. We compared the differences in the 12 RSNs between the combined intervention group and rTMS-alone intervention group, and found that the combined intervention improved the inter-network connectivity of the cerebellum and enhanced the strength of intra-network connectivity within frontal-parietal regions, which was closely related to cognitive improvement. ## 4.1. Neurotransmitter Drugs Combined with rTMS Intervention Can Reconstruct Functional Connectivity Associated with Cerebellum The cerebellum has been known to play an important role in motor control [38,39]. There is growing evidence that the cerebellum is involved not only in motor function, but also in sensory and cognitive processes [40]. The cerebellum is a complex structural and functional region that is functionally connected to multiple brain networks and is involved in processes such as cognitive, emotional, and sensorimotor processes [41]. For example, Ferrari et al. demonstrated that the cerebellum plays a role in short-term memory for the order of incoming visual stimuli [40]. In addition, Gatti’s research shows that the right cerebellum has a causal relationship with words related to integrative semantics [42]. Therefore, previous studies have suggested that the cerebellum was closely associated with cognitive function [30,43,44,45,46]. Studies have shown that in patients with MCI and AD, there is a significant disruption of cortico-cerebellar functional connectivity, particularly in regions of the default mode network (DMN) and frontoparietal network (FPN). The DMN is a brain network that includes the medial prefrontal cortex, posterior cingulate cortex, and bilateral parietal cortex, and it is involved in various cognitive processes such as self-referential thinking, mind-wandering, and memory. It is considered the most vulnerable brain network in AD, and a reduced connection of the cerebellar network to the DMN in AD patients may be a contributing factor to cognitive impairment [47]. The present study found that the combined intervention group had enhanced connectivity between CN and SN, while decreasing connectivity between CN and aDMN when compared to the rTMS-alone intervention group. The functional connectivity between CN and SN was also found to be correlated with MMSE scores and episodic memory. Our results were consistent with those previous studies. For example, a large number of studies have shown that rTMS intervention can affect cerebellar function, including motor and cognitive domains. In the motor field, rTMS can affect cerebellum visual guidance, motor surround inhibition, motor adaptation, and learning. In the cognitive domain, rTMS can influence the cognitive processes involved in the cerebellum, including verbal working memory, semantic association, and predictive language processing [48]. It is important to acknowledge that while rTMS has a direct effect on neurons in the stimulated area, it may also indirectly modulate neural responses in other nearby or distant regions, such as the motor cortex or prefrontal cortex [40,49,50]. Transcranial magnetic stimulation (TMS) is known to directly modulate inhibitory activity of Purkinje cells located in the cerebellar cortex, which in turn affects the activity of the cerebral cortex through the thalamus. As a result, rTMS not only targets the intended brain region but can also indirectly modulate neural responses in other proximal or distal regions, such as the motor cortex or prefrontal cortex. Therefore, the effects of rTMS on brain function and behavior may extend beyond the targeted area and could potentially result in functional changes in other brain regions [51,52]. The results suggest that rTMS intervention targeting the left angular gyrus in this study may affect cognitive function by affecting the functional connection strength between the cerebellar cortex and the cerebral cortex. In addition, much of the previous research on TMS affecting cerebellar function has focused on semantic memory [42]. The results of our study also proved this point. In this study, the FC between CN and SN was significantly enhanced in the combined intervention group, which was related to language function. Moreover, we also found that FC enhancement between CN and SN in patients in the combined intervention group was also related to MMSE scores and executive function. Summing up the above, we suspect that rTMS may affect cognitive function by inducing changes in the functional connectivity between the cerebellum and the cortex, regulating cerebellar excitability. Previous studies of drugs such as donepezil and memantine have not reported on the effects of neurotransmitters on functional connections between cerebellar networks and other networks, so we speculate that this may be a new finding regarding rTMS treatment, but the underlying mechanism needs more evidence to be proven. ## 4.2. Neurotransmitter Drugs Combined with rTMS Intervention Can Enhance Functional Connectivity within Frontal-Parietal Regions An experiment utilizing repetitive transcranial magnetic stimulation (rTMS) to target the left angular gyrus has demonstrated the crucial role of this brain region in scenario simulation and memory [53], with the left angular gyrus situated at the convergence of brain regions supporting various cognitive processes, such as language, attention, and semantic, numerical, and social cognition [54,55]. The angular gyrus, as a node of the DMN, exhibits connections with the frontoparietal control network, which plays a role in executive control processes during cognitive tasks [56]. Additionally, the angular gyrus exhibits connections with the precuneus and the mid-cingulate cortex, which are believed to play a role in mediating various aspects of memory function, including memory retrieval tasks [57]. Previous studies have shown that rTMS targeting the left angular gyrus may impact other core regions, such as the hippocampus and precuneus, which are associated with visual memory and posterior cingulate/precuneus activation [53]. In particular, the activation of visual memory is closely related to memory processing (hippocampus and parahippocampal cortex) and posterior cingulate/precuneus [58,59,60,61]. Similar studies have suggested that the application of rTMS can affect episodic memory performance by targeting specific lateral parietal regions that are connected to the hippocampus [13,62]. The fronto-parietal network included the bilateral DLPFC (middle frontal gyrus) and parietal (superior parietal gyrus) cortex, which play a crucial role in the top-down control of attention and in modulating cognition in AD. Prior studies have indicated that cognitive control is executed through the adaptable reorganization of the FPN in relation to various cognitive domains [63,64,65,66]. We found that the FC of the left precuneus and Frontal_Sup_Medial_L, which belong to LFPN, were improved significantly in the combined intervention group compared to in the rTMS-alone intervention group. The subsequent related analyses confirmed that the Z-values of the left precuneus and Frontal_Sup_Medial_L were positively correlated with the episodic memory. We suggest that this improvement was supported by changes in cortical activity of the precuneus and Frontal_Sup_Medial_L and their connectivity with the frontal-parietal network. Our findings are in accordance with previous studies. Giacomo et al. found a significant beneficial effect of rTMS intervention that targeted the precuneus in improving episodic memory [67]. Prior studies have demonstrated that the dorsal attention network (DAN) consists of both cortical and cerebellar nodes that are associated with attentional control [68]. Recent research has found that increased functional connectivity (FC) within the DAN and enhanced connections between the cerebellum and cortical network nodes were strongly correlated with increased activation during multiple attention tasks, indicating a significant relationship between connectivity and cognitive performance [50]. This finding is consistent with our previous discussion that rTMS can rebuild the functional connection between the cerebellum and cortex. Furthermore, our findings indicate that the functional connectivity (FC) of the Precentral_L node within the dorsal attention network (DAN) was greater in the combined intervention group than in the rTMS-only intervention group. Additional analyses revealed that the Z-values of the Precentral_L node were negatively associated with executive functions. In addition, the heightened connectivity within the DAN in the combined intervention group is related to poorer cognitive function, potentially indicating a compensatory mechanism. This mechanism proposes that hyperactive brain regions increase their activity to compensate for cognitive decline in other brain regions to preserve cognitive function [69]. The rTMS intervention in this study targeted the left hippocampus, and all participants received four weeks of treatment with rTMS aimed at the left angular gyrus. The results demonstrated enhanced functional connectivity (FC) of the precuneus, which was associated with episodic memory and MMSE scores, consistent with previous findings. Notably, neurotransmitters can have similar effects, as evidenced by a previous study which found that just five days of galantamine treatment in MCI patients resulted in increased activation in the left hippocampus, prefrontal, anterior cingulate, and occipital cortices during face encoding, as well as increased activity in the right precuneus and middle frontal gyrus during working memory tasks [70]. A previous study has reported that the ventral default mode network (DMN) comprises a collection of regions spanning from the precuneus and posterior cingulate cortex to the parahippocampal region via the retrosplenial cortex [71]. They also found that three months of donepezil treatment significantly enhanced functional connectivity within the network, including the para-hippocampal gyrus [72]. Moreover, previous studies have shown that donepezil can improve cognitive function in patients with Alzheimer’s disease (AD) through modulation of spontaneous brain activity. Specifically, improvements in cognitive function have been associated with increased spontaneous activity in the right gyrus rectus, precentral gyrus, and left superior temporal gyrus [73]. The results suggest that the functional connection of the precuneus and frontal lobe is enhanced, and positive correlation with episodic memory and cognition may be a synergistic effect between rTMS and neurotransmitters. Previous research has also shown that upregulating frontal systems activation can improve cognitive performance [6,74,75]. Thus, the enhancement of precentral gyrus functional connection in the combined intervention group may be due to the action of neurotransmitters. Overall, the results indicate that the combined intervention is more effective than rTMS alone in enhancing functional connectivity within FPN and DMN, which can improve cognitive function in patients with cognitive impairment. ## 4.3. Limitations and Prospects There were several limitations in our study. First, many imaging studies have demonstrated that neurotransmitters activate the brain, but it is still unclear whether the changes in brain activation caused by these drugs are associated with changes in cognitive outcomes, and whether these associations have significant cognitive benefits. In addition, due to sample size limitations, we chose a less stringent threshold for voxels and clumps, and the changes in brain functional network connectivity between groups were not strictly corrected. Second, our study was a short-term study; there was no long-term follow-up of patients and, due to conditional restrictions, we failed to include a group of patients who had not been treated with rTMS but were taking neurotransmitters as a control group. In future studies we will include control groups and expand sample sizes to draw stronger conclusions. In addition, we will conduct long-term follow-up of subjects to explore the long-term effects of drug and rTMS interventions. ## 5. Conclusions In summary, the present findings clearly demonstrated that co-intervention of neurotransmitter drugs and rTMS showed potential for cognitive improvement via the reconstructed inter-network connectivity of the cerebellum and the enhanced intra-network connectivity of frontal-parietal regions, although additional studies will be needed to shed light on more fine-grained questions regarding the effect of the combined intervention. ## References 1. Hebert L.E., Scherr P.A., Beckett L.A., Albert M.S., Pilgrim D.M., Chown M.J., Funkenstein H.H., Evans D.A.. Age-specific incidence of Alzheimer’s disease in a community population. JAMA (1995) 273 1354-1359. DOI: 10.1001/jama.1995.03520410048025 2. Zhao Z., Wu J., Fan M., Yin D., Tang C., Gong J., Xu G., Gao X., Yu Q., Yang H.. Altered intra- and inter-network functional coupling of resting-state networks associated with motor dysfunction in stroke. Hum. Brain Mapp. (2018) 39 3388-3397. DOI: 10.1002/hbm.24183 3. Yang Z., Sheng X., Qin R., Chen H., Shao P., Xu H., Yao W., Zhao H., Xu Y., Bai F.. Cognitive Improvement via Left Angular Gyrus-Navigated Repetitive Transcranial Magnetic Stimulation Inducing the Neuroplasticity of Thalamic System in Amnesic Mild Cognitive Impairment Patients. J. Alzheimers Dis. JAD (2022) 86 537-551. DOI: 10.3233/JAD-215390 4. Muir J.L.. Acetylcholine, aging, and Alzheimer’s disease. Pharmacol. Biochem. Behav. (1997) 56 687-696. DOI: 10.1016/S0091-3057(96)00431-5 5. Conway M.E.. Alzheimer’s disease: Targeting the glutamatergic system. Biogerontology (2020) 21 257-274. DOI: 10.1007/s10522-020-09860-4 6. Petrella J.R., Prince S.E., Krishnan S., Husn H., Kelley L., Doraiswamy P.M.. Effects of donepezil on cortical activation in mild cognitive impairment: A pilot double-blind placebo-controlled trial using functional MR imaging. AJNR Am. J. Neuroradiol. (2009) 30 411-416. DOI: 10.3174/ajnr.A1359 7. Risacher S.L., Wang Y., Wishart H.A., Rabin L.A., Flashman L.A., McDonald B.C., West J.D., Santulli R.B., Saykin A.J.. Cholinergic Enhancement of Brain Activation in Mild Cognitive Impairment during Episodic Memory Encoding. Front. Psychiatry (2013) 4 105. DOI: 10.3389/fpsyt.2013.00105 8. Lorenzi M., Beltramello A., Mercuri N.B., Canu E., Zoccatelli G., Pizzini F.B., Alessandrini F., Cotelli M., Rosini S., Costardi D.. Effect of memantine on resting state default mode network activity in Alzheimer’s disease. Drugs Aging (2011) 28 205-217. DOI: 10.2165/11586440-000000000-00000 9. Zhao J., Li Z., Cong Y., Zhang J., Tan M., Zhang H., Geng N., Li M., Yu W., Shan P.. Repetitive transcranial magnetic stimulation improves cognitive function of Alzheimer’s disease patients. Oncotarget (2017) 8 33864-33871. DOI: 10.18632/oncotarget.13060 10. Bauer P.R., Kalitzin S., Zijlmans M., Sander J.W., Visser G.H.. Cortical excitability as a potential clinical marker of epilepsy: A review of the clinical application of transcranial magnetic stimulation. Int. J. Neural Syst. (2014) 24 1430001. DOI: 10.1142/S0129065714300010 11. Chen J., Ma N., Hu G., Nousayhah A., Xue C., Qi W., Xu W., Chen S., Rao J., Liu W.. rTMS modulates precuneus-hippocampal subregion circuit in patients with subjective cognitive decline. Aging (2020) 13 1314-1331. DOI: 10.18632/aging.202313 12. Freedberg M., Reeves J.A., Toader A.C., Hermiller M.S., Voss J.L., Wassermann E.M.. Persistent Enhancement of Hippocampal Network Connectivity by Parietal rTMS Is Reproducible. eNeuro (2019) 6. DOI: 10.1523/ENEURO.0129-19.2019 13. Wang J.X., Rogers L.M., Gross E.Z., Ryals A.J., Dokucu M.E., Brandstatt K.L., Hermiller M.S., Voss J.L.. Targeted enhancement of cortical-hippocampal brain networks and associative memory. Science (2014) 345 1054-1057. DOI: 10.1126/science.1252900 14. Beckmann C.F., DeLuca M., Devlin J.T., Smith S.M.. Investigations into resting-state connectivity using independent component analysis. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. (2005) 360 1001-1013. DOI: 10.1098/rstb.2005.1634 15. Wei L., Hu X., Zhu Y., Yuan Y., Liu W., Chen H.. Aberrant Intra- and Internetwork Functional Connectivity in Depressed Parkinson’s Disease. Sci. Rep. (2017) 7 2568. DOI: 10.1038/s41598-017-02127-y 16. McKhann G.M., Knopman D.S., Chertkow H., Hyman B.T., Jack C.R., Kawas C.H., Klunk W.E., Koroshetz W.J., Manly J.J., Mayeux R.. The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. J. Alzheimers Assoc. (2011) 7 263-269. DOI: 10.1016/j.jalz.2011.03.005 17. Ding D., Zhao Q., Guo Q., Meng H., Wang B., Luo J., Mortimer J.A., Borenstein A.R., Hong Z.. Prevalence of mild cognitive impairment in an urban community in China: A cross-sectional analysis of the Shanghai Aging Study. Alzheimers Dement. J. Alzheimers Assoc. (2015) 11 300-309.e2. DOI: 10.1016/j.jalz.2013.11.002 18. Petersen R.C.. Mild cognitive impairment as a diagnostic entity. J. Intern. Med. (2004) 256 183-194. DOI: 10.1111/j.1365-2796.2004.01388.x 19. Petersen R.C., Smith G.E., Waring S.C., Ivnik R.J., Tangalos E.G., Kokmen E.. Mild cognitive impairment: Clinical characterization and outcome. Arch. Neurol. (1999) 56 303-308. DOI: 10.1001/archneur.56.3.303 20. Petersen R.C., Doody R., Kurz A., Mohs R.C., Morris J.C., Rabins P.V., Ritchie K., Rossor M., Thal L., Winblad B.. Current concepts in mild cognitive impairment. Arch. Neurol. (2001) 58 1985-1992. DOI: 10.1001/archneur.58.12.1985 21. Zhao Z., Lu J., Jia X., Chao W., Han Y., Jia J., Li K.. Selective changes of resting-state brain oscillations in aMCI: An fMRI study using ALFF. BioMed Res. Int. (2014) 2014 920902. DOI: 10.1155/2014/920902 22. Guo Q., Zhao Q., Chen M., Ding D., Hong Z.. A comparison study of mild cognitive impairment with 3 memory tests among Chinese individuals. Alzheimer Dis. Assoc. Disord. (2009) 23 253-259. DOI: 10.1097/WAD.0b013e3181999e92 23. Chao-Gan Y., Yu-Feng Z.. DPARSF: A MATLAB Toolbox for “Pipeline” Data Analysis of Resting-State fMRI. Front. Syst. Neurosci. (2010) 4 13. PMID: 20577591 24. Ashburner J., Friston K.J.. Unified segmentation. NeuroImage (2005) 26 839-851. DOI: 10.1016/j.neuroimage.2005.02.018 25. Song J., Qin W., Liu Y., Duan Y., Liu J., He X., Li K., Zhang X., Jiang T., Yu C.. Aberrant functional organization within and between resting-state networks in AD. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0063727 26. Li X., Wang F., Liu X., Cao D., Cai L., Jiang X., Yang X., Yang T., Asakawa T.. Changes in Brain Function Networks in Patients With Amnestic Mild Cognitive Impairment: A Resting-State fMRI Study. Front. Neurol. (2020) 11 554032. DOI: 10.3389/fneur.2020.554032 27. Zhao Z., Huang T., Tang C., Ni K., Pan X., Yan C., Fan X., Xu D., Luo Y.. Altered resting-state intra- and inter- network functional connectivity in patients with persistent somatoform pain disorder. PLoS ONE (2017) 12. DOI: 10.1371/journal.pone.0176494 28. Liao H., Cai S., Shen Q., Fan J., Wang T., Zi Y., Mao Z., Situ W., Liu J., Zou T.. Networks Are Associated With Depression in Patients With Parkinson’s Disease: A Resting-State Imaging Study. Front. Neurosci. (2020) 14 573538. DOI: 10.3389/fnins.2020.573538 29. Houck J.M., Çetin M.S., Mayer A.R., Bustillo J.R., Stephen J., Aine C., Cañive J., Perrone-Bizzozero N., Thoma R.J., Brookes M.J.. Magnetoencephalographic and functional MRI connectomics in schizophrenia via intra- and inter-network connectivity. NeuroImage (2017) 145 96-106. DOI: 10.1016/j.neuroimage.2016.10.011 30. Li F., Lu L., Shang S., Hu L., Chen H., Wang P., Zhang H., Chen Y.C., Yin X.. Disrupted functional network connectivity predicts cognitive impairment after acute mild traumatic brain injury. CNS Neurosci. Ther. (2020) 26 1083-1091. DOI: 10.1111/cns.13430 31. Wang S., Chen B., Yu Y., Yang H., Cui W., Fan G., Li J.. Altered resting-state functional network connectivity in profound sensorineural hearing loss infants within an early sensitive period: A group ICA study. Hum. Brain Mapp. (2021) 42 4314-4326. DOI: 10.1002/hbm.25548 32. Lei Y., Zhang D., Qi F., Gao J., Tang M., Ai K., Yan X., Lei X., Shao Z., Su Y.. Dysfunctional Interaction Between the Dorsal Attention Network and the Default Mode Network in Patients With Type 2 Diabetes Mellitus. Front. Hum. Neurosci. (2021) 15 796386. DOI: 10.3389/fnhum.2021.796386 33. Zuo X.N., Kelly C., Adelstein J.S., Klein D.F., Castellanos F.X., Milham M.P.. Reliable intrinsic connectivity networks: Test-retest evaluation using ICA and dual regression approach. NeuroImage (2010) 49 2163-2177. DOI: 10.1016/j.neuroimage.2009.10.080 34. Wang D., Qin W., Liu Y., Zhang Y., Jiang T., Yu C.. Altered resting-state network connectivity in congenital blind. Hum. Brain Mapp. (2014) 35 2573-2581. DOI: 10.1002/hbm.22350 35. Seeley W.W., Menon V., Schatzberg A.F., Keller J., Glover G.H., Kenna H., Reiss A.L., Greicius M.D.. Dissociable intrinsic connectivity networks for salience processing and executive control. J. Neurosci. Off. J. Soc. Neurosci. (2007) 27 2349-2356. DOI: 10.1523/JNEUROSCI.5587-06.2007 36. Wu C., Matias C., Foltynie T., Limousin P., Zrinzo L., Akram H.. Dynamic Network Connectivity Reveals Markers of Response to Deep Brain Stimulation in Parkinson’s Disease. Front. Hum. Neurosci. (2021) 15 729677. DOI: 10.3389/fnhum.2021.729677 37. Beckmann C.F., Smith S.M.. Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans. Med. Imaging (2004) 23 137-152. DOI: 10.1109/TMI.2003.822821 38. Habas C.. Functional Connectivity of the Cognitive Cerebellum. Front. Syst. Neurosci. (2021) 8 15. DOI: 10.3389/fnsys.2021.642225 39. Hardwick R.M., Therrien A.S., Lesage E.. Non-invasive stimulation of the motor cerebellum has potential cognitive confounds. Brain Stimul. (2021) 14 922-923. DOI: 10.1016/j.brs.2021.05.011 40. Ferrari C., Cattaneo Z., Oldrati V., Casiraghi L., Castelli F., D’Angelo E., Vecchi T.. TMS Over the Cerebellum Interferes with Short-term Memory of Visual Sequences. Sci. Rep. (2018) 8 6722. DOI: 10.1038/s41598-018-25151-y 41. Zheng W., Liu X., Song H., Li K., Wang Z.. Altered Functional Connectivity of Cognitive-Related Cerebellar Subregions in Alzheimer’s Disease. Front. Aging Neurosci. (2017) 9 143. DOI: 10.3389/fnagi.2017.00143 42. Gatti D., Van Vugt F., Vecchi T.. A causal role for the cerebellum in semantic integration: A transcranial magnetic stimulation study. Sci. Rep. (2020) 10 18139. DOI: 10.1038/s41598-020-75287-z 43. Bernard J.A., Seidler R.D.. Moving forward: Age effects on the cerebellum underlie cognitive and motor declines. Neurosci. Biobehav. Rev. (2014) 42 193-207. DOI: 10.1016/j.neubiorev.2014.02.011 44. Schmahmann J.D.. The cerebellum and cognition. Neurosci. Lett. (2019) 688 62-75. DOI: 10.1016/j.neulet.2018.07.005 45. Balsters J.H., Whelan C.D., Robertson I.H., Ramnani N.. Cerebellum and cognition: Evidence for the encoding of higher order rules. Cereb. Cortex (2013) 23 1433-1443. DOI: 10.1093/cercor/bhs127 46. Fan L., Hu J., Ma W., Wang D., Yao Q., Shi J.. Altered baseline activity and connectivity associated with cognitive impairment following acute cerebellar infarction: A resting-state fMRI study. Neurosci. Lett. (2019) 692 199-203. DOI: 10.1016/j.neulet.2018.11.007 47. Tang F., Zhu D., Ma W., Yao Q., Li Q., Shi J.. Differences Changes in Cerebellar Functional Connectivity Between Mild Cognitive Impairment and Alzheimer’s Disease: A Seed-Based Approach. Front. Neurol. (2021) 12 645171. DOI: 10.3389/fneur.2021.645171 48. Grimaldi G., Argyropoulos G.P., Boehringer A., Celnik P., Edwards M.J., Ferrucci R., Galea J.M., Groiss S.J., Hiraoka K., Kassavetis P.. Non-invasive cerebellar stimulation—A consensus paper. Cerebellum (2014) 13 121-138. DOI: 10.1007/s12311-013-0514-7 49. Farzan F., Pascual-Leone A., Schmahmann J.D., Halko M.. Enhancing the Temporal Complexity of Distributed Brain Networks with Patterned Cerebellar Stimulation. Sci. Rep. (2016) 6 23599. DOI: 10.1038/srep23599 50. Esterman M., Thai M., Okabe H., DeGutis J., Saad E., Laganiere S.E., Halko M.A.. Network-targeted cerebellar transcranial magnetic stimulation improves attentional control. NeuroImage (2017) 156 190-198. DOI: 10.1016/j.neuroimage.2017.05.011 51. Daskalakis Z.J., Paradiso G.O., Christensen B.K., Fitzgerald P.B., Gunraj C., Chen R.. Exploring the connectivity between the cerebellum and motor cortex in humans. J. Physiol. (2004) 557 689-700. DOI: 10.1113/jphysiol.2003.059808 52. Gatti D., Rinaldi L., Cristea I., Vecchi T.. Probing cerebellar involvement in cognition through a meta-analysis of TMS evidence. Sci. Rep. (2021) 11 14777. DOI: 10.1038/s41598-021-94051-5 53. Thakral P.P., Madore K.P., Schacter D.L.. A Role for the Left Angular Gyrus in Episodic Simulation and Memory. J. Neurosci. Off. J. Soc. Neurosci. (2017) 37 8142-8149. DOI: 10.1523/JNEUROSCI.1319-17.2017 54. Seghier M.L.. The angular gyrus: Multiple functions and multiple subdivisions. Neurosci. Rev. J. Bringing Neurobiol. Neurol. Psychiatry (2013) 19 43-61. DOI: 10.1177/1073858412440596 55. Binder J.R., Desai R.H.. The neurobiology of semantic memory. Trends Cogn. Sci. (2011) 15 527-536. DOI: 10.1016/j.tics.2011.10.001 56. Vincent J.L., Kahn I., Snyder A.Z., Raichle M.E., Buckner R.L.. Evidence for a frontoparietal control system revealed by intrinsic functional connectivity. J. Neurophysiol. (2008) 100 3328-3342. DOI: 10.1152/jn.90355.2008 57. Gilmore A.W., Nelson S.M., McDermott K.B.. A parietal memory network revealed by multiple MRI methods. Trends Cogn. Sci. (2015) 19 534-543. DOI: 10.1016/j.tics.2015.07.004 58. Daselaar S.M., Porat Y., Huijbers W., Pennartz C.M.. Modality-specific and modality-independent components of the human imagery system. NeuroImage (2010) 52 677-685. DOI: 10.1016/j.neuroimage.2010.04.239 59. Gilboa A., Winocur G., Grady C.L., Hevenor S.J., Moscovitch M.. Remembering our past: Functional neuroanatomy of recollection of recent and very remote personal events. Cereb. Cortex (2004) 14 1214-1225. DOI: 10.1093/cercor/bhh082 60. Addis D.R., Moscovitch M., Crawley A.P., McAndrews M.P.. Recollective qualities modulate hippocampal activation during autobiographical memory retrieval. Hippocampus (2004) 14 752-762. DOI: 10.1002/hipo.10215 61. Legostaeva L., Poydasheva A., Iazeva E., Sinitsyn D., Sergeev D., Bakulin I., Lagoda D., Kremneva E., Morozova S., Ryabinkina Y.. Stimulation of the Angular Gyrus Improves the Level of Consciousness. Brain Sci. (2019) 9. DOI: 10.3390/brainsci9050103 62. Nilakantan A.S., Bridge D.J., Gagnon E.P., VanHaerents S.A., Voss J.L.. Stimulation of the Posterior Cortical-Hippocampal Network Enhances Precision of Memory Recollection. Curr. Biol. CB (2017) 27 465-470. DOI: 10.1016/j.cub.2016.12.042 63. Cocchi L., Halford G.S., Zalesky A., Harding I.H., Ramm B.J., Cutmore T., Shum D.H., Mattingley J.B.. Complexity in relational processing predicts changes in functional brain network dynamics. Cereb. Cortex (2014) 24 2283-2296. DOI: 10.1093/cercor/bht075 64. Cole M.W., Reynolds J.R., Power J.D., Repovs G., Anticevic A., Braver T.S.. Multi-task connectivity reveals flexible hubs for adaptive task control. Nat. Neurosci. (2013) 16 1348-1355. DOI: 10.1038/nn.3470 65. Fornito A., Harrison B.J., Zalesky A., Simons J.S.. Competitive and cooperative dynamics of large-scale brain functional networks supporting recollection. Proc. Natl. Acad. Sci. USA (2012) 109 12788-12793. DOI: 10.1073/pnas.1204185109 66. Hearne L.J., Cocchi L., Zalesky A., Mattingley J.B.. Reconfiguration of Brain Network Architectures between Resting-State and Complexity-Dependent Cognitive Reasoning. J. Neurosci. Off. J. Soc. Neurosci. (2017) 37 8399-8411. DOI: 10.1523/JNEUROSCI.0485-17.2017 67. Koch G., Bonnì S., Pellicciari M.C., Casula E.P., Mancini M., Esposito R., Ponzo V., Picazio S., Di Lorenzo. F., Serra L.. Transcranial magnetic stimulation of the precuneus enhances memory and neural activity in prodromal Alzheimer’s disease. NeuroImage (2018) 169 302-311. DOI: 10.1016/j.neuroimage.2017.12.048 68. Brissenden J.A., Levin E.J., Osher D.E., Halko M.A., Somers D.C.. Functional Evidence for a Cerebellar Node of the Dorsal Attention Network. J. Neurosci. Off. J. Soc. Neurosci. (2016) 36 6083-6096. DOI: 10.1523/JNEUROSCI.0344-16.2016 69. Reuter-Lorenz P.A., Cappell K.A.. Neurocognitive Aging and the Compensation Hypothesis. Curr. Dir. Psychol. Sci. (2008) 17 177-182. DOI: 10.1111/j.1467-8721.2008.00570.x 70. Goekoop R., Rombouts S.A., Jonker C., Hibbel A., Knol D.L., Truyen L., Barkhof F., Scheltens P.. Challenging the cholinergic system in mild cognitive impairment: A pharmacological fMRI study. NeuroImage (2004) 23 1450-1459. DOI: 10.1016/j.neuroimage.2004.08.006 71. Solé-Padullés C., Bartrés-Faz D., Lladó A., Bosch B., Peña-Gómez C., Castellví M., Rami L., Bargalló N., Sánchez-Valle R., Molinuevo J.L.. Donepezil treatment stabilizes functional connectivity during resting state and brain activity during memory encoding in Alzheimer’s disease. J. Clin. Psychopharmacol. (2013) 33 199-205. DOI: 10.1097/JCP.0b013e3182825bfd 72. Goveas J.S., Xie C., Ward B.D., Wu Z., Li W., Franczak M., Jones J.L., Antuono P.G., Li S.J.. Recovery of hippocampal network connectivity correlates with cognitive improvement in mild Alzheimer’s disease patients treated with donepezil assessed by resting-state fMRI. J. Magn. Reson. Imaging JMRI (2011) 34 764-773. DOI: 10.1002/jmri.22662 73. Cheng J., Yang H., Zhang J.. Donepezil’s Effects on Brain Functions of Patients With Alzheimer Disease: A Regional Homogeneity Study Based on Resting-State Functional Magnetic Resonance Imaging. Clin. Neuropharmacol. (2019) 42 42-48. DOI: 10.1097/WNF.0000000000000324 74. Saykin A.J., Wishart H.A., Rabin L.A., Flashman L.A., McHugh T.L., Mamourian A.C., Santulli R.B.. Cholinergic enhancement of frontal lobe activity in mild cognitive impairment. Brain J. Neurol. (2004) 127 1574-1583. DOI: 10.1093/brain/awh177 75. Griffanti L., Wilcock G.K., Voets N., Bonifacio G., Mackay C.E., Jenkinson M., Zamboni G.. Donepezil Enhances Frontal Functional Connectivity in Alzheimer’s Disease: A Pilot Study. Dement. Geriatr. Cogn. Disord. Extra (2016) 6 518-528. DOI: 10.1159/000450546
--- title: LC-MS/MS-Based Proteomics Approach for the Identification of Candidate Serum Biomarkers in Patients with Narcolepsy Type 1 authors: - Akeem Sanni - Mona Goli - Jingfu Zhao - Junyao Wang - Chloe Barsa - Samer El Hayek - Farid Talih - Bartolo Lanuzza - Firas Kobeissy - Giuseppe Plazzi - Monica Moresco - Stefania Mondello - Raffaele Ferri - Yehia Mechref journal: Biomolecules year: 2023 pmcid: PMC10046664 doi: 10.3390/biom13030420 license: CC BY 4.0 --- # LC-MS/MS-Based Proteomics Approach for the Identification of Candidate Serum Biomarkers in Patients with Narcolepsy Type 1 ## Abstract Narcolepsy type 1 (NT1) is the most common type of narcolepsy known to be caused by the loss of specific neurons responsible for producing peptide neurotransmitters (orexins/hypocretins), resulting in a sleep-wake cycle disorder. It is characterized by its association with cataplexy and abnormalities in rapid eye movement. To date, no cure has been established for this life-threatening condition. Misdiagnosis of NT1 is also quite common, although it is not exceedingly rare. Therefore, successfully identifying candidate serum biomarkers for NT1 would be a head start for accurate diagnosis and development of therapeutics for this disorder. This study aims to identify such potential serum biomarkers. A depletion protocol was employed for 27 human serum samples (16 NT1 and 11 healthy controls), followed by applying LC-MS/MS bottom-up proteomics analysis, then LC-PRM-MS for validation. The comparison of the proteome profiles of the low-abundant proteins in the samples was then investigated based on age, sex, sample groups, and the presence of the Human Leukocyte Antigen (HLA) DQB10602 allele. The results were tracked to gene expression studies as well as system biology to identify key proteins and understand their relationship in the pathogenesis of NT1. Our results revealed 36 proteins significantly and differentially expressed. Among the impaired pathways and bioprocesses, the complement activation pathway is impaired by six of the differentially expressed proteins (DEPs). They are coded by the genes C2, CFB, C5, C1R, C1S, and MASP1, while 11 DEPs are involved in Acute Phase Response Signaling (APRS), which are coded by the genes FN1, AMBP, APOH, CFB, CP, ITIH2, C5, C2, F2, C1, and ITIH4. The combined AUCs of the downregulated and upregulated DEPs are 0.95 and 0.76, respectively. Overall, this study reveals potential serum-protein biomarkers of NT1 and explains the possible correlation between the biomarkers and pathophysiological effects, as well as important biochemical pathways involved in NT1. ## 1. Introduction Narcolepsy type 1 (NT1) is a sleep disorder associated with cataplexy, a neurological condition characterized by muscle paralysis and decreased sensitivity to pain [1]. The disease is characterized by a high level of loss or dysfunction of orexin/hypocretin neurons in the lateral region of the hypothalamus without affecting the melanin-concentrating hormone neurons [2]. Apart from sleep-wake regulation, it is important to state that the dysfunction of the orexinergic system affects other metabolic processes and complicates some metabolic diseases such as type-2 diabetes and obesity [3,4]. In the 1950s, excessive daytime sleepiness and sleep paralysis, hypersomnia, and cataplexy were considered the major symptoms of narcolepsy [5]. Over time, the correct and scientific diagnosis of NT1 became more complicated and not limited to these symptoms. Christian and Claudio, in 2004, compared the two types of narcolepsy and attempted to differentiate their common cataplexy-like symptoms from ‘true’ cataplexy in NT1 [6]. Since NT1 comes with cataplexy-like symptoms, it differs from narcolepsy without definite cataplexy, and it is also the most common type of narcolepsy [7]. The existence of co-occurring symptoms proves the scientific need to investigate and identify biomarkers for NT1; this would improve the accurate diagnosis and categorization of this disease. On the other hand, proteome profiling is one of the vital experimental studies for biomarker discovery in biological and biomedical samples [8]. Using this method, the proteome of a particular cell or biofluid is identified, and the proteins that are differentially expressed are identified and further analyzed using bioinformatics techniques. Therefore, serum-based proteomics gives the advantage of revealing disease progression. Developing techniques for biomarker identification, exploring expression levels of the comprehensively identified proteins, and associating those changes to different disease conditions using bioinformatics software help in the discovery of clinically relevant biomarkers [9,10]. Validation of the data is necessary to factor in the effect of FDR during in-silico digestion of most bioinformatic tools utilized in proteomics studies. Western blotting has often been applied, but to improve analysis throughout, mass spectrometry (MS)-based targeted analysis was developed, and parallel reaction monitoring (PRM) is now commonly used. MS is a very sensitive analytical tool that takes advantage of its high mass accuracy and resolution for detecting and quantifying analytes [11,12]. MS coupled with other instruments for separation, such as capillary electrophoresis and liquid chromatography, improves the ability of MS to serve a multi-purpose bioanalytical role [13,14]. This leads to an improvement in proteomics studies and the development of biomarkers in the scientific community [15]. Liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) is a versatile bioanalytical technique for high-throughput quantitative analysis of complex biological samples, especially in proteomics [16,17]. Previously, several studies have considered LC-MS/MS for measurements of orexin-A in narcolepsy and proteomics profiling of the hypothalamus in mouse models of narcolepsy [18,19]. However, to the best of our knowledge, identifying and quantifying potential biomarkers in human serum samples of NT1 has never been done. In this study, we apply LC-MS/MS techniques for label-free proteomics quantification (LFPQ) and targeted proteomics, followed by system biology, to identify potential biomarkers of human sera samples in NT1. In addition, this study evaluates the correlation between narcolepsy and immunity/autoimmunity, inflammation, and degeneration, as well as other biological, cellular, and molecular processes involved in this disease. Overall, the results of this study will provide information on candidate serum-protein biomarkers for narcolepsy type 1 that could be targeted for accurate serum diagnosis and development of therapeutics for narcolepsy type 1. This should directly and significantly indicate new ways to improve the diagnosis of NT1 in serum samples of patients, as well as assist in the development of potential therapeutics. ## 2.1. Chemicals and Reagents (HPLC)-grade water (H2O), acetonitrile (ACN), and MS-grade formic acid (FA) were obtained from Fisher Scientific (Fair Lawn, NJ, USA). Ammonium bicarbonate (ABC), dithiothreitol (DTT), and iodoacetamide (IAA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Trypsin/Lys-C mix MS-grade was purchased from Promega (Madison, WI, USA). ## 2.2. Study Participants Patients with a clinical diagnosis of NT1 were recruited from the Department of Neurology and Sleep Medicine, Oasi Research Institute (IRCCS), and the Department of Neurology, University of Bologna, Italy, between April and June 2018. For patients with NT1, a clinical and laboratory diagnosis of NT1, according to the 3rd edition of the International Classification of Sleep Disorders, was required. It was recommended that the following criteria be met [20]: (a) unequivocal cataplexy during laboratory testing; (b) persistent daytime sleepiness; (c) at least 2 SOREMPs and an 8-min mean sleep latency during the multiple sleep latency tests; (d) evidence of cerebrospinal fluid orexin A deficiency, when available. The presence of human leukocyte antigen (HLA) DQB10602 was assessed in all patients. Patients were excluded if they had any other neurologic or medical conditions (including hypertension and diabetes, in particular) and if they were taking any medications. All patients were enrolled soon after the diagnosis was made; for this reason, they were all drug-naïve at the time of the blood sampling, none were obese, and their body mass indexes were not excessive (5 overweight patients out of 16, with values ≥25 <30). None of these patients had Streptococcus infection, H1N1 flue or use of AS03-adjuvanted vaccine prior to the onset of narcoleptic symptoms. Healthy participants served as controls. The study was approved by the local ethics committee. All participants provided written informed consent before participating in the study, and their demographic and some clinical information is presented in Table 1. The serum samples were collected between the hours of 9 AM–12 PM. ## 2.3. Depletion of High-Abundance Proteins in Serum Samples As seen in Figure 1, the experimental processes followed in this study started with the depletion of the collected serum samples. A total of 27 serum samples, consisting of 11 healthy controls and 16 NT1 samples, were subjected to depletion using Agilent Human 14 multiple affinity removal column with dimension 4.6 × 100 mm. This column depletes the 14 high-abundant proteins, namely albumin, IgA, transferrin, haptoglobin, IgG, antitrypsin, transferrin, alpha 1-acid glycoprotein, IgM, apolipoprotein AI, apolipoprotein AII, transthyretin, complement C3, and alpha 2-macroglobulin. The depletion process helps expand the dynamic range of quantitative and qualitative analysis of the low-abundant proteins that might be differentially expressed as a result of NT1 disease. Therefore, a 30 µL aliquot of serum was depleted following the protocol provided by the manufacturer. After depletion, using 5KDa MWCO 4 mL spin concentrator Agilent filter (Santa Clara, CA, USA), the buffer in the samples was exchanged with 50 mM ammonium bicarbonate (ABC) (pH 8.0) which is more compatible with Mass spectrometry, and was used for tryptic digestion. ## 2.4. Tryptic Digestion The depleted sera were further prepared based on previously established protocols [21]. The protein concentration of depleted sera samples was determined before tryptic digestion using the micro-BCA protein assay following the protocol recommended by the vendor (Thermo Scientific/Pierce, Rockford, IL, USA). An aliquot of 15 µg of depleted serum proteins corresponding to 0.2 µL of original serum was transferred to an Eppendorf tube, followed by the addition of 50 mM ammonium bicarbonate to have the final volume of 50 µL. Thermal denaturation was performed at 80 °C for 30 min. The denatured proteins were reduced and alkylated with DTT and IAA, respectively [22]. DTT and IAA solutions were prepared in 50 mM ammonium bicarbonate. Initially, 1.25 µL of DTT was added to the samples and incubated at 60 °C for 45 min, then samples were alkylated with 5 µL of IAA solution and incubated at 37.5 °C for 45 min. Subsequently, alkylation was quenched with another 1.25 µL of DTT, and incubation was done at 37.5 °C for 30 min. To digest the reduced and alkylated proteins into peptides, 0.6 µg of trypsin/lysine C was added to the samples and then incubated at 37.5 °C for 18 h. Later, the digestion was terminated by the addition of 0.3 µL of neat FA. The samples were speed-vacuum dried, resuspended in a solution containing $98\%$ HPLC water, $2\%$ ACN, and $0.1\%$ FA, and then subjected to liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) analysis [22]. ## 2.5. Identification of Potential Biomarkers Using Untargeted LC-MS/MS Proteomics Analysis A Dionex 3000 Ultimate nano-LC system (Sunnyvale, CA, USA) connected to an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, MA, USA), which was equipped with a nano-ESI source, was used for the analysis. LC-MS/MS analysis was conducted on a tryptic digest serum sample corresponding to 1 µg of proteins. Two solvents were used as gradients; 1 is more aqueous, and the 2nd is more organic: mobile phase A was composed of $97.9\%$ HPLC water, $2\%$ can, and $0.1\%$ FA, while mobile phase B was composed of $99.9\%$ ACN and $0.1\%$ FA. Firstly, all 27 samples (Control and NT1 samples) were loaded on the trapping column (Thermo Scientific Acclaim PepMap C18, 75 µm × 20 mm, 3 µm, 100 Å), which helped enrich and desalt the samples (online purification). Then, the samples were further separated with a Thermo Scientific PepMap C18 column (75 µm × 150 mm, 2 µm, 100 Å) with a flow rate of 0.350 µL/min in a 120 min run. Mobile phase B was kept at a slope of $5\%$ for the first 10 min of the run, then increased to $20\%$ over 55 min, $30\%$ over 25 min, $50\%$ over 20 min, and $80\%$ over 1 min, then kept at $5\%$ until the end of the run. The separated peptides were sent into the LTQ Orbitrap Velos mass spectrometer for detection through electrospray ionization. The mass spectrometer was set at Data Dependent Acquisition (DDA) mode for the two MS scans; the first was an FTMS scan from 400–2000 m/z and Normalized collision energy (CE) of $35\%$ at a resolution of 60,000 while it worked in positive ion mode. From the MS spectrum, the top 10 most intense ions were selected for fragmentation in the second MS event (MS2), and Collision Induced Dissociation (CID) was set as the fragmentation technique with a mass resolution of 7500. The isolation width was 3 m/z, the dynamic exclusion duration was 90 s, the activation Q was set at 0.250, and the activation time was 10 min. ## 2.6. Untargeted LC-MS/MS Data Analysis Raw data obtained from the untargeted LC-MS/MS analysis of the 27 serum samples were processed by MaxQuant software, version 1.5.6.4 (Matrix Science Inc., Boston, MA) to attain the peptide intensities when searched against the Swiss Protein human database. The acetylation of the protein N-terminal and the oxidation of methionine were set as variable modifications, and the fixed modification was set for carbamidomethylating of cysteine. The proteomics experiment was conducted on label-free quantification (LFQ); thus, this was set on the MaxQuant, and the LFQ average number of neighbors was set as 6. Peptides were identified at an m/z tolerance of 6 ppm, while the minimum peptide length was set to 7, with a maximum of two missed cleavages. Finally, the false discovery rate (FDR) was set at 0.01 for identifying peptides and proteins. Further to all these settings, proteins with a minimum of two identified peptides were only considered. In order to visualize the results from the MaxQuant analysis, Perseus software, version 1.5.5.09 (Max Planck Institute of Biochemistry, Munich, Germany), was used, which generated a master file containing all proteins identified in the 27 samples, along with their LFQ intensities. The ion peak intensities represent the relative abundance of the proteins in each sample. The data was further analyzed by GraphPad Prism 9.3.1 (GraphPad Software Company, La Jolla, CA, USA). Since the data extracted did not satisfy the normality test criteria, a nonparametric test was applied. The levels of identified proteins in the NT1 samples were compared with Control samples using Mann-Whitney U-Test. All the identified proteins were compared based on sex differences in the sample, age, and the presence of human leukocyte antigen (HLA). The GenBank ID and gene names of each protein were exported to an excel file to form an experimental data set. Principal Component Analysis (PCA) and heatmap were performed to visualize the differences among different groups. Unsupervised PCA was conducted with OriginPro2022b software ($95\%$ confidence level), and a protein-specific heatmap was created with Genesis software version 1.8.1. In addition, to evaluate the predictive ability, selectivity, and sensitivity of each of the DEPs, SPSS® version 28 (IBM) software was used to create receiver operating characteristic curves as well as to obtain the area under curves (AUC) values [23]. ## 2.7. Ingenuity Pathway Analysis (IPA) and Gene Ontology for System Biology Tracking the functions of the DEPs in NT1 in different biological and functional biochemical pathways gives an idea of the possible effect of the DEPs in the pathophysiology of NT1. This could assist in the therapeutic design and drug targeting for the disease. Both pathway studio and IPA software were used for system biology analysis in this study. For the IPA, the gene name of the proteins, p-value, as well as the Log2 of the fold change were used for the analysis. Core analysis was performed in a flexible format, and the Gene symbol (human) was set as a platform for the experiment. GenBank or Uniprot/SwissProtein Accession was set as an identity platform for the proteins. According to previous literature, pathways related to NT1 disease were further investigated and scanned to better understand the roles of the significantly expressed proteins in those pathways. Elsevier’s Pathway Studio v 10.0 (https://www.elsevier.com/solutions/pathway-studio-biological-research (accessed on 23 August 2022)) was employed to establish the relationships between the DEPs and biological processes related to NT1. The validated proteins were designed in hollow color for easy identification and reference. The software generates a proteome-interactome network using a direct-interaction algorithm that maps cellular processes and interactions among genes of the DEPs. Of the 36 proteins, interactions related to the brain and circulation were generated with proteins considered statistically significant after Fisher’s statistical test. This was achieved by the Subnetwork Enrichment Analysis (SNEA) algorithm, which also helps extract statistically significant altered functional pathways. SNEA creates a central seed from all the related entities in the database and retrieves associated entities based on their relationship with the seed—expression targets, regulation targets, and protein modification targets. The gene ontology (GO) of the DEPs was conducted with “Panther17.0 Released” to understand pathway functions, molecular function, and the effect of those proteins in biological processes (p-value cutoff = 0.05). A Hierarchical clustering tree was obtained by online bioinformatic software “Shiny GO v0.75”, which was used to visualize the strength of the relationship between pathways affected by the DEPs [24]. After the quantitative validation of the DEPs, the proteins were processed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment and another GO, with only the top 10 pathways considered. “ Shiny GO v0.75” was also used to predict the biological processes affected by the quantitatively validated DEPs and the proteins’ cellular and molecular functions [24]. ## 2.8. Subcellular Localization After quantitatively validating the proteins, their subcellular distribution was determined by a commonly used prediction software, CELLO v.2.5 [25]. The result was also double-checked with WoLF PSORT, another commonly used subcellular localization software [26]. The software works with different types of sequence coding strategies: the dipeptide composition, partitioned amino acid composition, and the physicochemical properties of the primary structure of the proteins. In addition, WoLF PSORT considers the functional motifs of the proteins. K-nearest neighbor classifier values assigned to the DEPs were then used for prediction. ## 2.9. Targeted Proteomics (LC-PRM-MS) Strategy A targeted PRM approach was applied to validate some of the DEPs (those with the highest level of fold change and more related to NT1, as predicted by the IPA analysis). First, 1 µL of each of the 27 tryptic digested samples were pooled together, dried, resuspended in the mobile phase, and then run on the Dionex 3000 Ultimate nano-LC system interfaced with a Q-Exactive HF mass spectrometer (Thermo Scientific). Next, 1 µg/µL of the pooled sample was injected into the instrument for the identification of the precursors’ ions, their fragments, as well as retention times. The peptide information of the proteins of interest was identified, collected on Xcalibur, and then used to prepare a transition list for the targeted proteomics. The gradient used for the untargeted proteomics analysis was kept the same. As usual, after the chromatographic separation, peptides were detected in the Q-Exactive mass spectrometer Orbitrap with the following parameters: run-time 120 min; detection in positive mode; and an MS full-scan range of 200–2000 m/z followed by MS2 scanning at the range of 400–2000 m/z. A fragmentation pattern of HCD was attempted at normalized collision energy (NCE) 25 and 35 but established at 35 for this analysis. Precursors of the proteins of interest that met the requirements to be considered for a good PRM assay were selected and included in the transition list with a retention time window of ±6 min and a mass range of ±2Da relative to the target mass [27,28]. This was in accordance with an established protocol for targeted proteomics in Yehia Mechref’s Lab [29]. The expected RT of the precursors (obtained from untargeted proteomics) was manually re-checked from the raw data of the pooled sample using Xcalibur (Thermo Scientific) and compared with those in the untargeted proteomics study. Peptides absent in the pooled sample raw data were excluded from the transition list, and 42 peptides were finally targeted, representing 14 DEPs from untargeted proteomics results. For quantitative validation, PRM data was analyzed and quantified by Skyline software version 21.2.0.536, and the normalized data acquired were used for the statistical analysis. The Mann-Whitney U-Test was applied to compare the PRM data in NT1 and Control samples using GraphPad Prism 9.3.1 (GraphPad Software Company, La Jolla, CA, USA); precursors with a p-value < 0.05 were considered significant. ## 3.1. Unsupervised PCA for Comparative Proteomics Analysis of NT1 and Control Samples The combination of MaxQuant version 1.5.6.4 and Perseus 1.6.15.0 identified 195 low-abundant proteins in the 27 depleted serum samples. The unsupervised PCA generated with OriginPro2022b software at a confidence level of $95\%$ for all 195 identified proteins is shown in Figure 2A, and a reasonable level of clustering is observed. The differences in the NT1 and Control sample groups can be seen in the primary principal component 1 (PC1) and the secondary principal component 2 (PC2). Most importantly, this proves that the two groups (NT1 samples and control samples) can be considered as two different cohorts. The NT1 samples are expected to have a different expression level of proteins compared to the healthy control samples. This helps track the proteome as it changes from a healthy state to an NT1 condition in humans. ## 3.2. Heatmap of DEPs Of the 195 low-abundant proteins, 36 showed statistically significant expression changes (differentially expressed proteins) in a comparison between the NT1 sample group and the control group. The DEPs were used for protein-specific heatmap to visualize their expression levels. Figure 2B shows the protein-specific heatmap, revealing that the proteome of the NT1 serum samples has differentially expressed proteins that can be further investigated. 32 DEPs were observed to be downregulated in NT1, while 4 proteins were upregulated. The 32 proteins have relatively low levels of expression in NT1, while the protein product of the genes FN1, NID, C1RL, and PCYOX are observed to have a higher level of expression in NT1 (Table 2). This is consistent with the calculated fold change of 36 DEPs. The range of visualization is from green (−3.0) to red (3.0), depicting the level of proteome changes between NT1 and healthy samples (control). ## 3.3. Sex and Age-Based Comparison and Volcano Plot Aside from the 36 DEPs discovered based on the comparison between NT1 samples vs. control, out of 195 identified proteins across all samples, 11 DEPs were determined by sex-based comparison, and 15 DEPs were determined by age-based comparison. The Venn diagram investigating the overlapping proteins is shown in Figure 3A, and the protein content representing the element of each set is given in Table S1. From the gender dimorphism observed in the proteome profiles of NT1 in this study, out of the 11 DEPs, 10 proteins are upregulated in Female NT1 compared to male NT1, while one protein is downregulated in female NT1 compared to Male NT1. Additionally, the volcano plot shows all the identified proteins, with a threshold line generated at a p-value of 0.05 after the Mann–Whitney U-test. The data points above the line represent the statistically significant proteins or differentially expressed proteins (DEPs) (Figure 3B) and are painted magenta. The DEPs at the right-hand section of the plot are downregulated in NT1, while the DEPs at the left-hand section of the plot are upregulated. Information about the IDs, p-value, FC, and Area Under the Curve (AUC) of the DEPs are shown in Table 2. The down-regulated DEPs were combined; the range of their AUC is 0.73–0.91, with a p-value of 0.0001–0.04, whereas the AUC range for the upregulated proteins is 0.45–0.68 when combined (Figure 3C,D). The combined AUC for the downregulated DEPs in NT1 is 0.95, and for the upregulated DEPs in NT1 is 0.76, which indicates that the proteins are relatively selective and a good analyte to differentiate between NT1 and the healthy controls [30]. Because the AUC of Q9NZP8 falls below 0.50, it was not considered for the quantitative validation of DEPs in this study. ## 3.4. Gene Ontology for the Untargeted Proteomics Result The relationship between the top 30 pathways altered by the DEPs in NT1 is predicted by ShinyGO v0.75, which technically reveals the biological relevance of the 36 DEPs in NT1 (Figure S1A). The gene ontology of the DEPs considering molecular functions and biological functions are also shown in Figure S1B,C, respectively. Essentially, most of the DEPs are involved in the regulation of complement and coagulation cascades (immune response), as well as proteolysis; this is expected because the loss of orexin/hypocretin neuropeptides is well attributed to NT1 [31,32]. ## 3.5. PRM Validates DEPs After the untargeted proteomics study of the NT1, out of the 36 identified DEPs, we targeted 14 proteins that have the highest level of fold change. They are also mostly related to NT1 progression based on the IPA and GO results of the untargeted proteomics analysis (Figures S1 and S2). 11 DEPs were detected by the PRM assay, and eight DEPs were validated by PRM. The eight validated DEPs maintained the same trend of fold change when the untargeted proteomics and the targeted proteomics study of NT1 were carried out. Seven validated proteins were downregulated in NT1 (CP, CFB, C5, ITIH4, AMBP, C2, ITIH2), while one (FN1) was upregulated. As a reference point for comparison, the fold changes of the validated DEPs during targeted and untargeted proteomics studies are provided in Table S1. The table shows the list of the validated DEPs, including their charge state, m/z, RT, FC, and p-values. The pathway studio generated a global interactome of all differentially expressed proteins. The interactome result shows seven altered processes (Figure 4A), including pathways and other biological processes, as well as the network of DEPs that may be associated with NT1. These processes include the complement activation/alternative pathway, complement activation/classical pathway, acute-phase reaction, immune response, hemostasis, innate immune response, and phagocytosis. The interactome also shows eight altered cellular processes and the DEPs involved (Figure 4B), which include neutrophil adhesion, blood vessel injury, T-cell response, proteolysis, cell damage, ischemia, edema, and blood vessel permeability. ## 3.6. GO, KEGG and Subcellular Localization of the Three Quantitatively Validated DEPs Among the eight validated DEPs in the targeted proteomics study, three were quantitatively validated and coded by the genes ITIH4, FN1, and C5. The charge states, m/z, RT, FC, and p-value are given in Table S2, and Figure S3 shows the box plots and ROC curves of the three mentioned proteins. Figure S4A shows the functional and fold enrichment of the three quantitively validated DEPs. The KEGG analysis and GO analysis results for molecular function (Figure S4B), cellular component (Figure S4C), biological function (Figure S4D), and pie-chart of their subcellular distribution (Figure S4E) are also shown. Using GO, we analyzed the top 10 classifications most significantly enriched in cellular and molecular function and biological processes. The functional/fold enrichment analysis predicted that the three quantitatively validated proteins are primarily involved in complement and coagulation cascades, which supports the previously reported information [33]. In addition, the functional/fold enrichment analysis predicted other conditions such as infection of the epithelial cells, AGE-RAGE signaling pathway in diabetic complications, pertussis, and ECM-receptor interaction. The KEGG and gene ontology enrichment analysis results predicted that the three quantitatively validated proteins are also involved in the acute phase response, regulation of proteolysis and peptidase activity, hydrolase activities (Figure S4D), as well as regulation of important biological processes (Table S3). The loss of orexin/hypocretin neuropeptides in NT1 could be attributed to the dysregulation of proteolysis and peptidases. Essentially, the top 10 cellular components affected by the quantitively validated DEPs include the membrane attack complex and the formation of a fibrinogen complex (Figure S4C). This is also revealed in the IPA result, where a cascade of events in the complement activation pathway is shown. C5 is the first complement protein that constitutes the membrane attack complex that is formed in the pathway (Figure S2A). Finally, the subcellular localization result predicted that the quantitatively validated DEPs are primarily located in the extracellular and nuclear membrane, but plasma membrane, mitochondria, and cytoplasm were also predicted (Figure S4E). ## 4. Discussion In this study, we applied the technique of proteome profiling in identifying serum biomarkers specifically for the most common type of narcolepsy (NT1). To our knowledge, this is necessary for improving diagnostic tools for NT1 as it gives the advantage of diagnosing NT1 simply from a serum sample. In addition, we hope this study increases the understanding of the pathogenesis of NT1, which could help in designing therapeutic and prophylactic measures for the disease. Such work is essential for bridging the gaps in the current understanding of narcolepsy in the scientific community [1]. Previous studies have predicted that narcolepsy is associated with autoimmunity after investigating the ‘criteria of autoimmune disease’ in narcolepsy cases [7,34,35,36]. The relationship between human leukocyte antigen (HLA) and NT1 was significantly discussed in a study conducted by Giannoccaro et al. [ 36]. Thus, we investigated the presence of HLA in the patients’ samples selected for this study. Out of 16 narcolepsy patients, 15 tested positive for HLA, and one did not. We further identified 21 differentially expressed proteins by comparing HLA presence in our samples. In most cases, the expression of low-abundant proteins in serum and the alterations in their expressions are associated with a change in the organism’s physiological state. However, due to the wide dynamic range of serum proteins, the presence of low-abundant proteins is often masked by the high-abundant serum proteins, thereby affecting the detection of the low-abundant proteins that are currently considered potential biomarkers of diseases [37,38,39]. Hence, in this study, we focused on the low-abundant proteins by depleting the samples off the high-abundant proteins prior to the proteome profiling of the samples. When we compared the NT1 and healthy control groups, eventually, we found some similarities and many differences in the proteomes of the two groups. When the two groups were compared, 36 unique proteins were significantly expressed, and 32 of those were downregulated in NT1. Among the DEPs, the protein products of the genes C1RL, NID1, FN1, and PCYOX were upregulated in NT1. C1RL was previously found to be involved in immunological activities and reported as an independent prognostic biomarker in glioblastoma (GBM); interestingly, C1RL was also found to be upregulated (in GBM) in the study [40,41]. FN1 is a gene commonly reported in gene expression studies related to cancer and neurodegenerative diseases [19,42,43]. The synergetic effect of FN1 and genes of some DEPs in NT1 might have resulted in the dysregulation of neutrophils and leukocytes and might contribute to the complex relationship between orexin and cancer or neurodegeneration that has recently been indicated as a possible case of inverse comorbidity [44]. The function of PCYOX, which was previously unrecognized, was recently found and reported to be involved in controlling neutrophil Rhebs levels, thereby affecting neutrophil bactericidal potential [45]. Increased obesity and type-2 diabetes due to insulin resistance in the orexin deficiency state of narcolepsy patients have been reported in prior studies [4,46,47]. Excitingly, our KEGG functional enrichment analysis in this study predicted the effect of the quantitatively validated DEPs in the AGE-RAGE signaling pathway in diabetic complications and bacterial invasion of epithelial cells (Figure S4A), which further confirms the relationship between diabetes and narcolepsy. Furthermore, the Mignot group from Stanford University previously reported that infection could be the highest environmental risk for narcolepsy [48]. FN1, C5, and ITIH4 are all quantitatively validated. The summary of the functions of DEPs in this study can be understood from the results of the gene ontology analysis (Figure S1). In a study that investigated the alterations of classical component pathway factors in narcolepsy, C1Q was reported to be upregulated; it was also found to be upregulated in our study but did not meet the criteria to be statistically significant [33]. Some other complement proteins identified in these studies are found to be downregulated. Also, ITIH3 and PTPRN2 are proteins found in this study and have also been reported in proteomics profiling in narcolepsy rat models, as well as some other gene expression studies [19]. The acute phase response signaling pathway explains the biochemistry behind the cascade of events leading to the inflammatory response that confers protection of the body against infections. This signaling pathway is also triggered by tissue injury and acute-phase cytokines, and it has been previously reported that the acute phase response leads to changes in the body system, including sleepiness (excessive), which is a major symptom of narcolepsy; anorexia; and protein production, especially the complement proteins [49]. Interestingly, six of the DEPs identified by this study are involved in the complement activation pathway, coded by gene C2, CFB, C5, C1R, C1S, and MASP1; and eleven are involved in Acute Phase Response Signaling (APRS): FN1, AMBP, APOH, CFB, CP, ITIH2, C5, C2, F2, C1, and ITIH4. Notably, the integrin signaling pathway, p53 pathways, and axon guidance mediated by ‘semaphorins’ revealed by GO analysis are understandable as this may support the relationship between signaling pathways and inflammatory regulation as well as immune response mediation [50]. The p53 pathway has previously been reported to play a pivotal role in neurodegenerative diseases [51]. Figure S5 shows the interconnection of processes affected by the three quantitatively validated proteins. In conclusion, autoimmunity does not necessarily have to be confirmed by upregulation of all complement proteins but rather complement function dysfunction, which could be upregulation, downregulation, or dysfunction of complement proteins, as all of these can synergistically lead to autoimmunity; this aligns with a recent review study that explored the relationship between the complement system and autoimmunity [52]. Hence, to the best of our knowledge, our study confirms the relationship between autoimmunity and NT1 and explains some insights regarding the involved bioprocesses. However, it is important to stress the concept that this proteomic study most probably reflects the combined effect of genetic and epigenetic factors, which definitely interact with the pathophysiological mechanisms of NT1 [53]. For this reason, it is important to underline the high AUC found with the combined assembly of underregulated proteins, which seems to indicate that a complex proteomic pattern can be more helpful than single proteins in the identification of NT1 patients and in the characterization of their dysfunctional pathways. ## References 1. Kornum B.R., Knudsen S., Ollila H.M., Pizza F., Jennum P.J., Dauvilliers Y., Overeem S.. Narcolepsy. Nat. Rev. Dis. Prim. (2017) 3 16100. DOI: 10.1038/nrdp.2016.100 2. Bassetti C.L.A., Adamantidis A., Burdakov D., Han F., Gay S., Kallweit U., Khatami R., Koning F., Kornum B.R., Lammers G.J.. Narcolepsy—Clinical spectrum, aetiopathophysiology, diagnosis and treatment. Nat. Rev. Neurol. (2019) 15 519-539. DOI: 10.1038/s41582-019-0226-9 3. Tsujino N., Sakurai T.. Orexin/hypocretin: A neuropeptide at the interface of sleep, energy homeostasis, and reward system. Pharmacol. Rev. (2009) 61 162-176. DOI: 10.1124/pr.109.001321 4. Mohammadi S., Dolatshahi M., Zare-Shahabadi A., Rahmani F.. Untangling narcolepsy and diabetes: Pathomechanisms with eyes on therapeutic options. Brain Res. (2019) 1718 212-222. DOI: 10.1016/j.brainres.2019.04.013 5. Yoss R.E., Daly D.D.. Criteria for the diagnosis of the narcoleptic syndrome. Proc. Staff Meet. Mayo Clin. (1957) 32 320-328. PMID: 13441766 6. Sturzenegger C., Bassetti C.L.. The clinical spectrum of narcolepsy with cataplexy: A reappraisal. J. Sleep Res. (2004) 13 395-406. DOI: 10.1111/j.1365-2869.2004.00422.x 7. Han F., Lin L., Schormair B., Pizza F., Plazzi G., Ollila H.M., Nevsimalova S., Jennum P., Knudsen S., Winkelmann J.. *HLA DQB106:02 negative narcolepsy with hypocretin/orexin deficiency*. Sleep* (2014) 37 1601-1608. DOI: 10.5665/sleep.4066 8. Sun X., Sun H., Han X., Chen P.C., Jiao Y., Wu Z., Zhang X., Wang Z., Niu M., Yu K.. Deep Single-Cell-Type Proteome Profiling of Mouse Brain by Nonsurgical AAV-Mediated Proximity Labeling. Anal. Chem. (2022) 94 5325-5334. DOI: 10.1021/acs.analchem.1c05212 9. Kulasingam V., Diamandis E.P.. Strategies for discovering novel cancer biomarkers through utilization of emerging technologies. Nat. Clin. Pract. Oncol. (2008) 5 588-599. DOI: 10.1038/ncponc1187 10. Srinivas P.R., Verma M., Zhao Y., Srivastava S.. Proteomics for cancer biomarker discovery. Clin. Chem. (2002) 48 1160-1169. PMID: 12142368 11. Zhong L., Zhu L., Cai Z.W.. Mass Spectrometry-based Proteomics and Glycoproteomics in COVID-19 Biomarkers Identification: A Mini-review. J. Anal. Test (2021) 5 298-313. DOI: 10.1007/s41664-021-00197-6 12. Yu A., Zhao J., Peng W., Banazadeh A., Williamson S.D., Goli M., Huang Y., Mechref Y.. Advances in mass spectrometry-based glycoproteomics. Electrophoresis (2018) 39 3104-3122. DOI: 10.1002/elps.201800272 13. Sun L., Zhu G., Yan X., Dovichi N.J.. High sensitivity capillary zone electrophoresis-electrospray ionization-tandem mass spectrometry for the rapid analysis of complex proteomes. Curr. Opin. Chem. Biol. (2013) 17 795-800. DOI: 10.1016/j.cbpa.2013.07.018 14. Pejchinovski M., Hrnjez D., Ramirez-Torres A., Bitsika V., Mermelekas G., Vlahou A., Zürbig P., Mischak H., Metzger J., Koeck T.. Capillary zone electrophoresis on-line coupled to mass spectrometry: A perspective application for clinical proteomics. Proteom. Clin. Appl. (2015) 9 453-468. DOI: 10.1002/prca.201400113 15. Harvey W.T., Carabelli A.M., Jackson B., Gupta R.K., Thomson E.C., Harrison E.M., Ludden C., Reeve R., Rambaut A., Peacock S.J.. SARS-CoV-2 variants, spike mutations and immune escape. Nat. Rev. Microbiol. (2021) 19 409-424. DOI: 10.1038/s41579-021-00573-0 16. Jemal M.. High-throughput quantitative bioanalysis by LC/MS/MS. Biomed Chromatogr. (2000) 14 422-429. DOI: 10.1002/1099-0801(200010)14:6<422::AID-BMC25>3.0.CO;2-I 17. Gallien S., Duriez E., Demeure K., Domon B.. Selectivity of LC-MS/MS analysis: Implication for proteomics experiments. J. Proteom. (2013) 81 148-158. DOI: 10.1016/j.jprot.2012.11.005 18. Lindström M., Schinkelshoek M., Tienari P.J., Kukkonen J.P., Renkonen R., Fronczek R., Lammers G.J., Itkonen O.. Orexin-A measurement in narcolepsy: A stability study and a comparison of LC-MS/MS and immunoassays. Clin. Biochem. (2021) 90 34-39. DOI: 10.1016/j.clinbiochem.2021.01.009 19. Azzam S., Schlatzer D., Nethery D., Saleh D., Li X., Akladious A., Chance M.R., Strohl K.P.. Proteomic profiling of the hypothalamus in two mouse models of narcolepsy. Proteomics (2017) 17 1600478. DOI: 10.1002/pmic.201600478 20. Cartwright R.D.. Alcohol and NREM parasomnias: Evidence versus opinions in the international classification of sleep disorders, 3rd edition. J. Clin. Sleep Med. (2014) 10 1039-1040. DOI: 10.5664/jcsm.4050 21. Mondello S., Kobeissy F., Mechref Y., Zhao J., Talih F.R., Cosentino F., Antelmi E., Moresco M., Plazzi G., Ferri R.. Novel biomarker signatures for idiopathic REM sleep behavior disorder: A proteomic and system biology approach. Neurology (2018) 91 e1710-e1715. DOI: 10.1212/WNL.0000000000006439 22. Mondello S., Kobeissy F.H., Mechref Y., Zhao J., El Hayek S., Zibara K., Moresco M., Plazzi G., Cosentino F.I.I., Ferri R.. Searching for Novel Candidate Biomarkers of RLS in Blood by Proteomic Analysis. Nat. Sci. Sleep (2021) 13 873-883. DOI: 10.2147/NSS.S311801 23. Cao X.L., Song J.Y., Sun Z.G.. Quantitative label-free proteomic analysis of human follicle fluid to identify novel candidate protein biomarker for endometriosis-associated infertility. J. Proteom. (2022) 266 104680. DOI: 10.1016/j.jprot.2022.104680 24. Ge S.X., Jung D., Yao R.. ShinyGO: A graphical gene-set enrichment tool for animals and plants. Bioinformatics (2020) 36 2628-2629. DOI: 10.1093/bioinformatics/btz931 25. Yu C.S., Chen Y.C., Lu C.H., Hwang J.K.. Prediction of protein subcellular localization. Proteins (2006) 64 643-651. DOI: 10.1002/prot.21018 26. Horton P., Park K.J., Obayashi T., Fujita N., Harada H., Adams-Collier C.J., Nakai K.. WoLF PSORT: Protein localization predictor. Nucleic. Acids Res. (2007) 35 W585-W587. DOI: 10.1093/nar/gkm259 27. Rauniyar N.. Parallel Reaction Monitoring: A Targeted Experiment Performed Using High Resolution and High Mass Accuracy Mass Spectrometry. Int. J. Mol. Sci. (2015) 16 28566-28581. DOI: 10.3390/ijms161226120 28. Gutierrez Reyes C.D., Huang Y., Atashi M., Zhang J., Zhu J., Liu S., Parikh N.D., Singal A.G., Dai J., Lubman D.M.. PRM-MS Quantitative Analysis of Isomeric N-Glycopeptides Derived from Human Serum Haptoglobin of Patients with Cirrhosis and Hepatocellular Carcinoma. Metabolites (2021) 11. DOI: 10.3390/metabo11080563 29. Tsai T.H., Song E., Zhu R., Di Poto C., Wang M., Luo Y., Varghese R.S., Tadesse M.G., Ziada D.H., Desai C.S.. LC-MS/MS-based serum proteomics for identification of candidate biomarkers for hepatocellular carcinoma. Proteomics (2015) 15 2369-2381. DOI: 10.1002/pmic.201400364 30. Jiménez-Valverde A.. Insights into the area under the receiver operating characteristic curve (AUC) as a discrimination measure in species distribution modelling. Glob. Ecol. Biogeogr. (2012) 21 498-507. DOI: 10.1111/j.1466-8238.2011.00683.x 31. Straat M.E., Schinkelshoek M.S., Fronczek R., Lammers G.J., Rensen P.C.N., Boon M.R.. Role of Brown Adipose Tissue in Adiposity Associated With Narcolepsy Type 1. Front Endocrinol. (2020) 11 145. DOI: 10.3389/fendo.2020.00145 32. Barateau L., Liblau R., Peyron C., Dauvilliers Y.. Narcolepsy Type 1 as an Autoimmune Disorder: Evidence, and Implications for Pharmacological Treatment. CNS Drugs (2017) 31 821-834. DOI: 10.1007/s40263-017-0464-6 33. Yüceer H., Gezen Ak D., Şenel G.B., Dursun E., Yılmaz V., Karadeniz D., Tüzün E., Küçükali C.. Classical complement pathway factor alterations in narcolepsy. Acta Neuropsychiatr. (2022) 34 212-219. DOI: 10.1017/neu.2021.42 34. Rose N.R., Bona C.. Defining criteria for autoimmune diseases (Witebsky’s postulates revisited). Immunol. Today (1993) 14 426-430. DOI: 10.1016/0167-5699(93)90244-F 35. Latorre D., Kallweit U., Armentani E., Foglierini M., Mele F., Cassotta A., Jovic S., Jarrossay D., Mathis J., Zellini F.. T cells in patients with narcolepsy target self-antigens of hypocretin neurons. Nature (2018) 562 63-68. DOI: 10.1038/s41586-018-0540-1 36. Giannoccaro M.P., Liguori R., Plazzi G., Pizza F.. Reviewing the Clinical Implications of Treating Narcolepsy as an Autoimmune Disorder. Nat. Sci. Sleep (2021) 13 557-577. DOI: 10.2147/NSS.S275931 37. de Jesus J.R., de Campos B.K., Galazzi R.M., Martinez J.L., Arruda M.A.. Bipolar disorder: Recent advances and future trends in bioanalytical developments for biomarker discovery. Anal. Bioanal. Chem. (2015) 407 661-667. DOI: 10.1007/s00216-014-8341-3 38. Polaskova V., Kapur A., Khan A., Molloy M.P., Baker M.S.. High-abundance protein depletion: Comparison of methods for human plasma biomarker discovery. Electrophoresis (2010) 31 471-482. DOI: 10.1002/elps.200900286 39. Araújo J.E., Lodeiro C., Capelo J.L., Rodríguez-González B., dos Santos A.A., Santos H.M., Fernández-Lodeiro J.. Novel nanocomposites based on a strawberry-like gold-coated magnetite (Fe@Au) for protein separation in multiple myeloma serum samples. Nano Res. (2015) 8 1189-1198. DOI: 10.1007/s12274-014-0599-4 40. Wang J., Tong L., Lin G., Wang H., Zhang L., Yang X.. Immunological and clinicopathological characteristics of C1RL in 2120 glioma patients. BMC Cancer (2020) 20. DOI: 10.1186/s12885-020-07436-6 41. Fan S., Wu N., Chang S., Chen L., Sun X.. The immune regulation of BCL3 in glioblastoma with mutated IDH1. Aging Albany N. Y. (2022) 14 3856-3873. DOI: 10.18632/aging.204048 42. Song Y., Feng T., Cao W., Yu H., Zhang Z.. Identification of Key Genes in Nasopharyngeal Carcinoma Based on Bioinformatics Analysis. Comput. Intell. Neurosci. (2022) 2022 9022700. DOI: 10.1155/2022/9022700 43. Chen B., Zhou X., Yang L., Zhou H., Meng M., Wu H., Liu Z., Zhang L., Li C.. Glioma stem cell signature predicts the prognosis and the response to tumor treating fields treatment. CNS Neurosci. (2022) 28 2148-2162. DOI: 10.1111/cns.13956 44. Mogavero M.P., Silvani A., DelRosso L.M., Salemi M., Ferri R.. Focus on the Complex Interconnection between Cancer, Narcolepsy and Other Neurodegenerative Diseases: A Possible Case of Orexin-Dependent Inverse Comorbidity. Cancers (2021) 13. DOI: 10.3390/cancers13112612 45. Gadjeva M., Kugadas A., Petenkova A., Geddes-McAlister J., Mansour M.K., Sykes D.. Neutrophil maturation and their response to infectious pathogens are regulated by microbiota. J. Immunol. (2019) 202 127-22. DOI: 10.4049/jimmunol.202.Supp.127.22 46. Tsujino N., Sakurai T.. Role of orexin in modulating arousal, feeding, and motivation. Front. Behav. Neurosci. (2013) 7 28. DOI: 10.3389/fnbeh.2013.00028 47. Tsuneki H., Murata S., Anzawa Y., Soeda Y., Tokai E., Wada T., Kimura I., Yanagisawa M., Sakurai T., Sasaoka T.. Age-related insulin resistance in hypothalamus and peripheral tissues of orexin knockout mice. Diabetologia (2008) 51 657-667. DOI: 10.1007/s00125-008-0929-8 48. Bonvalet M., Ollila H.M., Ambati A., Mignot E.. Autoimmunity in narcolepsy. Curr. Opin. Pulm. Med. (2017) 23 522-529. DOI: 10.1097/MCP.0000000000000426 49. Berczi I., Stephano A.Q., Berczi I.. Chapter 8—Vasopressin, the Acute Phase Response and Healing. Insights to Neuroimmune Biology (2016) 185-199. DOI: 10.1016/B978-0-12-801770-8.00008-2 50. Zhang W., Tettamanti G., Bassal T., Heryanto C., Eleftherianos I., Mohamed A.. Regulators and signalling in insect antimicrobial innate immunity: Functional molecules and cellular pathways. Cell Signal (2021) 83 110003. DOI: 10.1016/j.cellsig.2021.110003 51. Chang J.R., Ghafouri M., Mukerjee R., Bagashev A., Chabrashvili T., Sawaya B.E.. Role of p53 in neurodegenerative diseases. Neuro-Degener. Dis. (2012) 9 68-80. DOI: 10.1159/000329999 52. Conigliaro P., Triggianese P., Ballanti E., Perricone C., Perricone R., Chimenti M.S.. Complement, infection, and autoimmunity. Curr. Opin. Rheumatol. (2019) 31 532-541. DOI: 10.1097/BOR.0000000000000633 53. Mogavero M.P., DelRosso L.M., Bruni O., Salemi M., Salsone M., Novellino F., Zucconi M., Ferini Strambi L., Ferri R.. Genetics and Epigenetics of Rare Hypersomnia. A Possible Keystone for their Diagnosis and Treatment within the Framework of Precision Medicine. Trends Genet. TIG (2023)
--- title: Pharmaceutical Payments to Japanese Board‐Certified Head and Neck Surgeons Between 2016 and 2019 authors: - Anju Murayama - Haruki Shigeta - Sae Kamamoto - Erika Yamashita - Hiroaki Saito - Toyoaki Sawano - Divya Bhandari - Sunil Shrestha - Eiji Kusumi - Tetsuya Tanimoto - Akihiko Ozaki journal: OTO Open year: 2023 pmcid: PMC10046701 doi: 10.1002/oto2.31 license: CC BY 4.0 --- # Pharmaceutical Payments to Japanese Board‐Certified Head and Neck Surgeons Between 2016 and 2019 ## Body The concern for the influence of physicians' financial conflicts of interest (COIs) with pharmaceutical companies on healthcare jeopardized trust in healthcare and evoked motivation demanding greater transparency in the relationships worldwide. 1, 2 In Japan, all pharmaceutical companies belonging to the Japan Pharmaceutical Manufacturers Association (JPMA), the largest pharmaceutical trade organization in Japan, have been demanded to disclose their payments to physicians on their company webpage since 2013. 3 This payment disclosure enabled publications of the financial relationships between physicians and pharmaceutical companies with detailed amounts of payments in several specialties. 4, 5, 6, 7, 8, 9, 10 Among several specialties, we speculated head and neck surgeons (HNSs) have increasing financial ties with pharmaceutical companies. One of the main diseases in the specialty is head and neck cancers including oral cavity cancer, pharyngeal cancer, and thyroid cancer. While surgery and radiotherapy are the key treatment for head and neck cancers, chemotherapy has advanced dramatically over the past 2 decades. 11 In 2017, nivolumab (OPDIVO®) was approved for head and neck cancers in Japan. Another immune checkpoint inhibitor, pembrolizumab (KEYTRUDA®) was approved for head and neck cancers in 2019. Considering the introduction of several novel drugs for head and neck cancers and the increased role of board‐certified HNSs, the HNSs were speculated to have substantial financial relationships with pharmaceutical companies. HNSs' financial relationships with pharmaceutical companies in the United States (US) were well‐described in studies conducted on otolaryngologist‐head and neck surgeons since the launch of Open Payments Database. 12, 13, 14, 15, 16 The Open Payments Database is the legal‐binding payment database including all financial transfers from pharmaceutical and medical devices companies to physicians. However, there was no study assessing financial relationships with pharmaceutical companies among the HNSs in Japan. This study aimed to elucidate the magnitude and trend of personal payments from pharmaceutical companies to the board‐certified HNSs through the recent years in Japan. ## Abstract ### Objective To evaluate the magnitude, prevalence, and trend of the financial relationship between Japanese head and neck surgeons and pharmaceutical companies between 2016 and 2019. ### Study Design Cross‐sectional analysis. ### Setting Japan. ### Methods This study evaluated personal payments concerning lecturing, consulting, and writing paid by 92 major pharmaceutical companies to all Japanese head and neck surgeons board‐certified by the Japan Society for Head and Neck Surgery between 2016 and 2019. The payments were descriptively analyzed and payment trend were assessed using population‐averaged generalized estimating equations. Further, the payments to board executive board members with specialist certification were also evaluated separately. ### Results Of all 443 board‐certified head and neck surgeons in Japan, 365 ($82.4\%$) received an average of $6443 (standard deviation: $12,875), while median payments were $2002 (interquartile ranges [IQR] $792‐$4802). Executive board specialists with a voting right received much higher personal payments (median $26,013, IQR $12,747‐$35,750) than the non‐executive specialists (median $1926, IQR $765‒$4134, $p \leq .001$) and the executive board specialists without a voting right (median $4411, IQR $963‐$5623, $$p \leq .015$$). The payments per specialist and prevalence of specialists with payments annually increased by $11.4\%$ ($95\%$ CI: $5.8\%$‐$17.2\%$; $p \leq .001$) and $7.3\%$ ($95\%$ CI: $3.8\%$‐$11.0\%$; $p \leq .001$), respectively. ### Conclusion There were increasingly widespread and growing financial relationships with pharmaceutical companies among head and neck surgeons in Japan, alongside of introduction of novel drugs. The leading head and neck surgeons received much higher personal payments from pharmaceutical companies, and no sufficient regulation was implemented by the society in Japan. ## Study Design This study is a cross‐sectional analysis evaluating the financial relationships between all board‐certified HNS specialists and pharmaceutical companies in Japan. All HNSs who were certified by the Japan Society of Head and Neck Surgery (JSHNS) were included in this study. The JSHNS, the sole and largest professional medical society for head and neck surgery in Japan, trains and certifies HNSs who have abundant skills and knowledge in head and neck surgery and can provide multidisciplinary treatment of head and neck cancers under the name of “Head and Neck Cancer Specialist.” ## Data Collection HNSs' names and affiliation were extracted from the official JSHNS webpage (https://www.jshns.org/modules/list_specialist/index.php) on February 10, 2022. At the time of our data extraction, the name list of board‐certified HNSs was last updated on January 20, 2022. The name list of executive board members of the JSHNS between 2020 and 2021 were also collected from the JSHNS webpage (https://www.jshns.org/modules/about/index.php?content_id=6) and the affiliation and position of the board‐certified HNSs were manually collected online to evaluate the strength of financial relationships between pharmaceutical companies and HNSs with a leading role. 17, 18, 19, 20 To evaluate the association between payment amounts and the number of drugs with new approval and additional indications for head and neck cancers and thyroid cancers, all drugs approved between 2011 and 2021 were extracted from the approval drug list issued by the Pharmaceuticals and Medical Devices Agency. The Pharmaceutical and Medical Devices Agency is the sole official agency reviewing and approving drugs in Japan, similar to the US Food and Drug Administration. 21 The payments concerning lecturing, writing, and consulting paid to the HNSs were extracted from all pharmaceutical companies belonging to the JPMA between 2016 and 2019. The JPMA transparency guidance voluntarily demands all member companies to disclose their payments to healthcare professionals and organizations. This payment disclosure is self‐regulated by the pharmaceutical industry association and there is no penalty for deviation from the guidance, which is one of the major differences between the US Open Payments program and JPMA transparency guidance. As of February 2022, the payment data of 2019 were the latest analyzable data in Japan. The payments from a total of 92 pharmaceutical companies were included in this study. The pharmaceutical companies disclosed payments for lecturing, writing, and consulting on the basis of individual physicians, but smaller and more prevalent payment categories such as food and beverages, travel and accommodation fees, and reimbursement for trial enrollment were not individually disclosed by the companies, as we noted previously. 5, 7, 8, 22 The extracted raw payment data were included as Supplemental Material 1. ## Analysis First, we conducted descriptive analyses for payment data. Average and median values were reported based on only HNSs receiving payment in each year, as in other studies. 8, 15, 16, 23, 24 Second, to evaluate payment concentration among the HNSs, the Gini index and the shares of the payment values per specialist were calculated, as performed previously 5, 8, 9, 25, 26 Third, to evaluate the trend between affiliations and positions, we used the robust adjustment. We also observed the affiliations and positions of HNSs who received more than $1000 continuously for 4 years. Fourth, we descriptively calculated payments between the HNSs with nonboard membership, the executive board HNSs without a voting right, and executive board HNSs with a voting right. The difference of payments between the 3 groups were evaluated by the Kruskal‐Wallis H test, and then the differences between each 2 group were assessed by Mann‐Whitney U test with the Bonferroni correction, as the payments were not normally distributed. Fifth, to evaluate the payment trends between 2016 and 2019, the population‐averaged generalized estimating equation (GEE) was performed, using the panel data of the personal payments in each specialist. As the payment distribution was highly skewed (Supplemental Material 2) negative binomial GEE model for the payment values per HNS, and linear GEE log‐linked model with binomial distribution for the prevalence of HNSs with payments were selected. Because several pharmaceutical companies disaffiliated from the JPMA and newly joined the JPMA, there were several companies without payment data over the 4 years. Thus, the trends of payments were calculated based on payments from all data‐collected companies and companies with payment data for the 4 years, as in our previous studies. 6, 8, 9, 27 Finally, we calculated Spearman's correlation between number of new approvals or additional indications for head and neck cancers and [1] 4‐years total payments and [2] number of HNSs with payments on the pharmaceutical company level. As the total payments and number of HNSs with payments were not normally distributed, Spearman's correlation was used. The payment values were converted from Japanese yen (¥) to US dollars ($) using 2019 average monthly exchange rates of ¥109.0 per $1. ## Ethical Approval The Ethics Committee of the Medical Governance Research Institute approved this study. As this study was a cross‐sectional analysis of publicly available information, informed consent was waived by the Ethics Committee. ## Overview and Per‐Specialist Payments A total of 443 HNSs were identified on the JSHNS webpage as of February 10, 2022. Of the 443 eligible board‐certified HNSs, 365 ($82.4\%$) received at least 1 payment from 55 ($60.0\%$) pharmaceutical companies between 2016 and 2019. Total payment amounts and number of instances were $2,351,621 and 3348 instances over the 4 years. The median was $2002 (interquartile range [IQR] $792‐$4802) in payments; 4.0 (IQR 2.0‐9.0) in payment instances; and 3.0 (IQR 2.0‐5.0) in number of pharmaceutical companies per specialist (Table 1). Table 1 \| Variables \| Unnamed: 1 \| \| --- \| --- \| \| Total \| \| \| Payment values, $ \| 2351621 \| \| Instances, n \| 3348 \| \| Companies, n \| 52 \| \| Average per specialist (SD) \| \| \| Payment values, $ \| 6443 (12,875) \| \| Instances, n \| 9.2 (13.9) \| \| Companies, n \| 4.1 (3.8) \| \| Median per specialist (IQR) \| \| \| Payment values, $ \| 2002 (792‐4802) \| \| Instances, n \| 4.0 (2.0‐9.0) \| \| Companies, n \| 3.0 (2.0‐5.0) \| \| Range \| \| \| Payment values, $ \| 95‐102,113 \| \| Instances, n \| 1.0‐105 \| \| Companies, n \| 1.0‐23.0 \| \| Physicians with specific payments, n (%) \| \| \| Any payments \| 365 (82.4) \| \| Payments > $500 \| 320 (72.2) \| \| Payments > $1000 \| 254 (57.3) \| \| Payments > $5000 \| 88 (19.9) \| \| Payments > $10,000 \| 55 (12.4) \| \| Payments > $50,000 \| 10 (2.3) \| \| Payments > $100,000 \| 1 (0.23) \| \| Gini index \| 0.764 \| \| Category of payments \| \| \| Lecturing \| \| \| Payment value, $ (%) \| 2,129,986 (90.6) \| \| Instances, n (%) \| 3050 (91.1) \| \| Consulting \| \| \| Payment value, $ (%) \| 106,786 (4.5) \| \| Instances, n (%) \| 112 (3.3) \| \| Writing \| \| \| Payment value, $ (%) \| 99,830 (5.1) \| \| Instances, n (%) \| 1086 (5.0) \| \| Other \| \| \| Payment value, $ (%) \| 15,020 (0.6) \| \| Instances, n (%) \| 15 (0.4) \| For the payment distribution, $72.2\%$, $57.3\%$, $19.9\%$, $12.4\%$, and $2.3\%$ of HNSs received more than $500, $1000, $5000, $10,000, and $50,000, respectively (Supplemental Material 3). The Gini index for the 4‐year cumulative payments per HNS was 0.764. Top $1\%$, $5\%$, $10\%$, and $25\%$ of HNSs occupied $14.1\%$ ($95\%$ confidence interval (CI) $10.8\%$‐$17.4\%$), $46.5\%$ ($95\%$ CI $41.0\%$‐$52.1\%$), $66.5\%$ ($95\%$ CI $62.0\%$‐$71.0\%$), and $84.6\%$ ($95\%$ CI $81.8\%$‐$87.5\%$) of total payments, respectively (Supplemental Material 3). The highest payment was $102,113. The most common payment category was lecturing and $80.1\%$ [355] of HNSs received 1 or more lecturing payments over the 4 years. ## Personal Payments and the Physicians' Affiliations and Positions Among 443 board‐certified HNSs, 221 ($49.9\%$) worked at universities or university hospital (Supplemental material 4). Compared to university staff who are not professors, university professors significantly received higher per‐physician personal payments (relative monetary value: 9.1times [$95\%$ CI 6.5‐12.8], $p \leq .001$), while the proportion of HNSs receiving payments did not reach statistical significance ($87.3\%$ vs $94.6\%$, relative proportion: 1.1 [$95\%$ CI 0.996‐1.2], $$p \leq .06$$). 52 HNSs ($11.7\%$) continuously received more than $1000 for the 4 years, and among them 31 ($59.6\%$) were university professors (Supplemental material 5). ## Payments to the JSHNS Executive Board Members We further investigated all 31 executive board members in 2021. Of 31 members, 18 ($58.1\%$) were the board‐certified HNSs. All 18 executive board members with specialist certification accepted personal payments from pharmaceutical companies between 2016 and 2019 (Table 2). Moreover, there was a statistically significant higher payment among the executive board HNSs with a voting right (median $26,013 [IQR $12,747‐$35,750]) than the HNSs with nonboard membership (median $1926 [IQR $765‐$4134], $p \leq .001$) and the executive board HNSs without a voting right (median $4411 [IQR $963‐$5623], $$p \leq .015$$). Table 2 \| Unnamed: 0 \| Unnamed: 1 \| Unnamed: 2 \| Payments, $ \| Payments, $.1 \| Payments, $.2 \| Payments, $.3 \| Payments, $.4 \| \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| \| Executive board members \| Position in the Society executive boarda \| Rankingb \| 2016 \| 2017 \| 2018 \| 2019 \| Four‐years combined \| \| A \| Executive board director \| 2 \| 6233 \| 15887 \| 25565 \| 20165 \| 67849 \| \| B \| Executive board chairperson \| 3 \| 13037 \| 26395 \| 15715 \| 11512 \| 66658 \| \| C \| Executive board secretary \| 13 \| 2759 \| 5211 \| 9707 \| 22594 \| 40271 \| \| D \| Executive board director \| 14 \| 8272 \| 13278 \| 11541 \| 7124 \| 40215 \| \| E \| Executive board director \| 20 \| 7077 \| 6361 \| 11437 \| 6410 \| 31285 \| \| F \| Executive board director \| 23 \| 8070 \| 6812 \| 12404 \| 2673 \| 29958 \| \| G \| Executive board director \| 26 \| 4965 \| 5544 \| 9560 \| 7895 \| 27964 \| \| H \| Executive board director \| 30 \| 1737 \| 6866 \| 4391 \| 11068 \| 24063 \| \| I \| Executive board director \| 35 \| 4829 \| 5501 \| 4291 \| 4496 \| 19116 \| \| J \| Executive board director \| 42 \| 1430 \| 4189 \| 3707 \| 5770 \| 15097 \| \| K \| Executive board director \| 51 \| 4189 \| 867 \| 2657 \| 2686 \| 10398 \| \| L \| Executive board director \| 54 \| 1941 \| 0 \| 2657 \| 5708 \| 10306 \| \| M \| Executive board director \| 70 \| 550 \| 1174 \| 2043 \| 2465 \| 6233 \| \| N \| Executive board secretary \| 76 \| 826 \| 1941 \| 1173 \| 1683 \| 5623 \| \| O \| Executive board secretary \| 91 \| 0 \| 2554 \| 1072 \| 1224 \| 4851 \| \| P \| Executive board secretary \| 108 \| 511 \| 1306 \| 817 \| 1338 \| 3972 \| \| Q \| Executive board secretary \| 256 \| 0 \| 275 \| 275 \| 413 \| 963 \| \| R \| Executive board secretary \| 310 \| 0 \| 0 \| 511 \| 0 \| 511 \| ## Payment Trend Between 2016 and 2019 The median annual payments per specialist increased from $817 (IQR $511‐$2248) in 2016 to $1027 (IQR $520‐$2284) in 2019, with an average annual change of $12.4\%$ ($95\%$ CI $6.8\%$‐$18.4\%$, $p \leq .001$) (Table 3). The number of HNSs receiving payments also annually increased from 189 ($42.7\%$) in 2016 to 260 ($58.7\%$) in 2018, while decreased to 245 ($55.3\%$) in 2019. Increasing trend of number of HNSs receiving payments by $8.0\%$ ($95\%$ CI $4.4\%$‐$11.7\%$; $p \leq .001$) per year was observed. Limiting payments from 50 ($90.9\%$) companies with 4‐years data, the payments per HNS and fraction of HNSs with payments also annually increased by $11.4\%$ ($95\%$ CI $5.8\%$‐$17.2\%$; $p \leq .001$) and $7.3\%$ ($95\%$ CI $3.8\%$‐$11.0\%$; $p \leq .001$), respectively. $20.3\%$ to $29.1\%$ of HNSs received more than $1000 per year and a total of $45.4\%$ [201] of HNSs received more than $1000 per year at least 1 year. Table 3 \| Variables \| 2016 \| 2017 \| 2018 \| 2019 \| Average yearly change (95% CI), % \| p value \| Combined total \| \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| \| All pharmaceutical companies \| \| \| \| \| \| \| \| \| Total payments, $ \| 422572 \| 656954 \| 630954 \| 641141 \| ‒ \| ‒ \| 2351621 \| \| Average payments (SD), $ \| 2236 (3172) \| 2760 (4728) \| 2427 (3968) \| 2617 (4311) \| 12.4 (6.8‐18.4) \| <0.001 \| 6443 (12,875) \| \| Median payments (IQR), $ \| 817 (511‐2248) \| 1022 (511‐2350) \| 970 (511‐2146) \| 1027 (520‐2284) \| 12.4 (6.8‐18.4) \| <0.001 \| 2002 (792‐4802) \| \| Payment range, $ \| 95‐17,160 \| 138‐28,503 \| 95‐25,565 \| 102‐36,111 \| ‒ \| \| 95‐102,113 \| \| Physicians with specific payments, n (%) \| \| \| \| \| \| \| \| \| Any payments \| 189 (42.7) \| 238 (53.7) \| 260 (58.9) \| 245 (55.3) \| 8.0 (4.4‐11.7) \| <0.001 \| 365 (82.4) \| \| Payments > $500 \| 148 (33.4) \| 187 (42.2) \| 203 (45.8) \| 209 (47.2) \| 11.0 (6.6‐15.5) \| <0.001 \| 320 (72.2) \| \| Payments > $1000 \| 90 (20.3) \| 127 (28.7) \| 129 (29.1) \| 129 (29.1) \| 10.3 (4.8‐16.2) \| <0.001 \| 254 (57.3) \| \| Payments > $5000 \| 26 (5.9) \| 38 (8.6) \| 33 (7.4) \| 37 (8.3) \| 8.7 (−0.9‐19.3) \| 0.78 \| 88 (19.9) \| \| Payments > $10,000 \| 8 (1.8) \| 17 (3.8) \| 17 (3.8) \| 15 (3.4) \| 15.9 (0.0‐34.2) \| 0.049 \| 55 (12.4) \| \| Payments > $50,000 \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| No observation \| ‒ \| 10 (2.3) \| \| Payments > $100,000 \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| No observation \| ‒ \| 1 (0.23) \| \| Gini index \| 0.836 \| 0.813 \| 0.789 \| 0.794 \| ‒ \| ‒ \| 0.764 \| \| Pharmaceutical companies with 4‐years payment data \| \| \| \| \| \| \| \| \| Total payments, $ \| 419932 \| 655933 \| 625448 \| 619380 \| ‒ \| ‒ \| 2320693 \| \| Average payments (SD), $ \| 2222 (3149) \| 2756 (4705) \| 2415 (3946) \| 2581 (4209) \| 11.4 (5.8‒17.2) \| <0.001 \| 6358 (12,702) \| \| Median payments (IQR), $ \| 817 (511‐2248) \| 1,022 (511‐2350) \| 970 (511‐2146) \| 1,025 (520‐2261) \| 11.4 (5.8‒17.2) \| <0.001 \| 1941 (765‐4802) \| \| Payment range, $ \| 95‐17,160 \| 138‐28,503 \| 95‐25,565 \| 95‐32,024 \| ‒ \| \| 95‐95,983 \| \| Physicians with specific payments, n (%) \| \| \| \| \| \| \| \| \| Any payments \| 189 (42.7) \| 238 (53.7) \| 259 (58.5) \| 240 (54.2) \| 7.3 (3.8‐11.0) \| <0.001 \| 365 (82.4) \| \| Payments > $500 \| 147 (33.2) \| 187 (42.2) \| 202 (45.6) \| 204 (46.0) \| 10.3 (5.3‐14.8) \| <0.001 \| 316 (71.3) \| \| Payments > $1,000 \| 89 (20.1) \| 127 (28.7) \| 128 (28.9) \| 125 (28.2) \| 9.5 (4.1‐15.2) \| <0.001 \| 250 (56.4) \| \| Payments > $5,000 \| 26 (5.9) \| 38 (8.6) \| 33 (7.4) \| 36 (8.1) \| 7.8 (−1.7‐18.1) \| 0.11 \| 87 (19.6) \| \| Payments > $10,000 \| 8 (1.8) \| 17 (3.8) \| 17 (3.8) \| 15 (3.4) \| 15.9 (0.0‒34.2) \| 0.49 \| 55 (12.4) \| \| Payments > $50,000 \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| No observation \| ‒ \| 10 (2.3) \| \| Payments > $100,000 \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| 0 (0.0) \| No observation \| ‒ \| 0 (0.0) \| \| Gini index \| 0.836 \| 0.812 \| 0.789 \| 0.799 \| ‒ \| ‒ \| 0.860 \| In a subgroup analysis on the HNSs with executive board memberships, this increasing trend of payment values and prevalence of were observed among the HNSs without executive board memberships in both of payment values (average annual change: $9.8\%$ [$95\%$ CI $4.2\%$‐$15.7\%$]; $p \leq .001$) and prevalence of HNSs with payments (average annual change: $7.5\%$ [$95\%$ CI $3.7\%$‐$11.5\%$]; $p \leq .001$). However, there were no increasing trend of payment values and prevalence of HNSs with payments among the executive board HNSs with a voting right. ## Payment by Pharmaceutical Companies Among the 92 pharmaceutical companies from which we collected data, 55 companies paid 1 or more payments to the board‐certified HNSs between 2016 and 2019. Payments from top 5 companies represented $60.1\%$ ($1,412,381) of total payments, respectively. The largest payments were made by Ono Pharmaceutical ($344,844; $14.7\%$), followed by Merck Biopharma ($334,019, $14.2\%$), and Taiho Pharmaceutical ($272,528, $11.6\%$). Taiho Pharmaceutical distributed personal payments to the largest number of 165 HNSs, totaling $37.2\%$ of all HNSs, followed by Merck Biopharma (160, $36.1\%$), Eisai (149, $33.6\%$). Payment trends and payment categories were described in Figures 1 and 2, respectively. There were 12 new or additional indications for head and neck cancers and thyroid cancers between 2011 and 2021 (Supplemental Material 6). Nine ($75.0\%$) drugs were for chemotherapy and 2 ($16.7\%$) drugs were for immunochemotherapy. Five ($41.7\%$) drugs were approved for head and neck carcinoma, 5 ($41.7\%$) drugs were approved for thyroid carcinoma, and remaining 2 ($16.7\%$) drugs were approved for other cancers including parathyroid carcinoma and pituitary tumor. Rakuten Medical and Stella Pharma were not member companies of JPMA, so the 2 companies did not disclose payments to healthcare professionals in Japan. Figure 1: Payment trends by company. Figure 2: Payment categories by company. There were moderate positive correlations between number of new or additional indications for head and neck cancers between 2011 and 2021 and the 4‐years total payments to the board‐certified HNSs (r[50] = 0.53; $p \leq .001$), as well as number of HNSs with payments (r[50] = 0.54, $p \leq .001$) in the Spearman's correlation. ## Discussion This study found $82.4\%$ of the Japanese board‐certified HNSs received a total of $2,351,621 in the personal payments for the reimbursement of lecturing, consulting, and writing between 2016 and 2019. However, only a very small proportion of these HNSs, such as the executive board members of the Japan Society for Head and Neck Surgery, received substantial personal payments from pharmaceutical companies. The substantial payments were often by the pharmaceutical companies which produced novel drugs for head and neck cancers. Our findings suggest several important similarities and differences from previous studies in the Japan and other developed countries. First, surprisingly the Japanese HNSs, representing $4.8\%$ of all Japanese otolaryngologists‐head and neck surgeons, 28 received an average of $2236 to $2760 annual personal payments, and the median annual payments were $817 to $1027. One study reported that the average and median personal payments from pharmaceutical and medical device companies to the US otolaryngologists were $1096 and $169 in 2014, respectively. 14 Comparing their findings, the Japanese HNSs received more than 5 times in median annual personal payments than those the US otolaryngologists received. Simply comparing the prevalence of physicians with payments, the percentage of the Japanese HNSs with payments ($82.4\%$) was similar to that in the US otolaryngologists ($84\%$‐$86\%$) shown by the previous studies. 14, 15, 16 However, our payment data only included payments for lecturing, consulting, and writing. Other common types of payments among otolaryngologists, 14 such as meals, travel, and accommodations were not included. Despite these limited categories of payments, which would significantly underestimate the prevalence of HNSs receiving payments, as high as $82.4\%$ of the Japanese HNSs were financially tied with pharmaceutical companies. Furthermore, we found that the Japanese HNSs received as substantial personal payments as Japanese physicians in other specialties. We previously reported that the median personal payments from pharmaceutical companies between 2016 and 2019 were $596 (IQR $0‐$2436) in the pediatric hematologists/oncologists, 27 $2210 ($715‐$8178) in the pulmonologists, 6 and $2471 ($851‐$9677) in the hematologists. 9 Furthermore these financial ties with pharmaceutical companies had become increasingly stronger and more prevalent. The payment analysis on the company level has provided plausible reason for the increasing trends. We found positive associations between number of novel approved drugs and magnitude of personal payments to the HNSs, as well as the number of HNSs distributed payments. Payments from top 5 companies occupied for $60\%$ of total payments, and among them, 4 companies had novel drugs for head and neck cancers. Among the 5 companies, Ono Pharmaceutical and Bristol Myers Squibb increasingly made personal payments to the HNSs mainly for lecturing since 2017. Ono Pharmaceutical paid 4.1 times higher payments in 2017, when Nivolumab was approved for head and neck cancer in Japan, compared to those in 2016 ($101,989 vs $25,033). Also, Bristol Myers Squibb made 13.7 times and surprisingly 27.2 times higher payments in 2017 ($51,787) and 2018 ($102,939) than those in 2016 ($3780), respectively. Despite these increasing trend of payments as well as novel drugs for head and neck cancer, the JSHNS has not made any regulation for financial COI among the board‐certified HNSs. Accumulating evidence indicates financial COI influences physicians' clinical practice 29 and this affects the otolaryngologists' prescriptions. 30, 31 Given these situations, regulations of financial COI for the board‐certified HNSs are essential, such as limiting the maximum monetary value of personal payments from pharmaceutical companies. For example, the Danish Medicines Agency restricts all Danish medical physicians not to received more than DKK 200,000 (equal to about $30,000 and about one‐sixth of annual Danish physician salary) without permission from the Danish government. 32 We also found that the leading HNSs in Japanese head and neck surgery such as the JSHNS executive board members received more substantial payments and they develop and implement regulations and statements for the HNSs endorsed by the JSHNS. Financial COI among the JSHNS executive board members were declared by each executive member, according to the JSHNS policy, but the COI information was not publicly disclosed due to the privacy of members. Financial independency from any entities and integrity to the patients are the most essential bases of all professional medical societies. 33, 34 However, most of Japanese professional medical societies were more substantially tied with pharmaceutical companies compared to other developed countries. Saito et al found that $86.9\%$ of executive board members of major professional medical societies received a median of $7486 personal payments in a single year. 20 The prevalence of executive members with payments and median personal payments were much higher than those in Australia ($24.4\%$ and $9861 between October 2015 and April 2018), 19 the United States ($71.6\%$ and $6026 between 2017 and 2019), 18 and France ($83\%$ and about $4200 per year). 17 This substantial concentration of payments to Japanese executive members could be due to less transparency in healthcare and immature discourse on financial COIs compared to other developed countries. 10 We believe full transparency in the financial relationships with pharmaceutical companies should be implemented among the leading physicians with public and authoritative position such as society executive board members and board‐certified HNSs, as payment disclosure increases public trust 35 and simultaneously patients desire physicians to disclose their financial relationships and to be free from financial ties with industries. 36, 37 Uniform payment database should be developed and more transparent and rigorous COI policy should be implemented among leading physicians, constantly updating in accordance with public demands 10, 37 and global standards. 38, 39 This study included several limitations. First, there would be underestimated payments to the HNSs from nonmember companies of JPMA. However, as the member companies accounted for $80.8\%$ of total pharmaceutical sales in Japan in 2018, 40 the underestimation of payments could be minimized by including data from all member companies. Second, currently, pharmaceutical companies do not disclose other categories of payments such as meals, travel, and stock ownerships, according to the JPMA guidance. 41 This could have led to underestimations of the extent and prevalence of overall financial relationships between HNSs and pharmaceutical industry. Third, this study included all HNSs as of February 2022, as the JSHNS did not disclose its list of HNSs for previous years. Therefore, there would have been some HNSs who did not hold the specialist certification. Fourth, the payment magnitude and trend may not be applicable to other countries. In conclusion, the majority of the Japanese board‐certified HNSs were financially tied with pharmaceutical companies manufacturing novel drugs between 2016 and 2019. These financial ties became increasingly prevalent and strong in overall HNSs. Additionally, the HNSs in leading roles received much higher personal payments from pharmaceutical companies than those in other developed countries, and no sufficient regulation was implemented by the professional medical society in Japan. ## Author Contributions Anju Murayama, data collection, study concept and design, resources, statistical analysis, drafting of the manuscript, reviewing of the manuscript, and study supervision; Haruki Shigeta, data collection, study concept and design, and drafting of the manuscript; Sae Kamamoto, data collection, study concept and design, and drafting of the manuscript; Erika Yamashita, data collection, reviewing of the manuscript; Hiroaki Saito, study concept and design, statistical analysis, drafting of the manuscript, and reviewing of the manuscript; Toyoaki Sawano, study concept and design, and reviewing of the manuscript; Divya Bhandari, study concept and design, and critically reviewing of the manuscript; Sunil Shrestha, study concept and design, and critically reviewing of the manuscript; Eiji Kusumi, study concept and design, and critically reviewing of the manuscript; Tetsuya Tanimoto, study concept and design, drafting of the manuscript, and study supervision; Akihiko Ozaki, study concept and design, data analysis, drafting of the manuscript, and study supervision; all authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. ## Competing interests Dr Saito received personal fees from TAIHO Pharmaceutical Co., Ltd. outside the scope of the submitted work. Dr Kusumi received personal fees from Otsuka Pharmaceutical outside the scope of the submitted work. Drs Ozaki and Tanimoto received personal fees from Medical Network Systems, a dispensing pharmacy, outside the scope of the submitted work. Dr Tanimoto also received personal fees from Bionics Co., Ltd., a medical device company, outside the scope of the submitted work. Regarding nonfinancial conflicts of interest among the study authors, all are engaged in ongoing research examining financial and nonfinancial conflicts of interest among healthcare professionals and pharmaceutical companies in Japan. Individually, Anju Murayama, Hiroaki Saito, Toyoaki Sawano, Tetsuya Tanimoto, and Akihiko Ozaki have contributed to several published studies addressing conflicts of interest and quality of evidence among clinical practice guideline authors in Japan and the United States. The other authors have no example conflicts of interest to disclose. ## Sponsorships This study was funded in part by the Medical Governance Research Institute. This non‐profit enterprise receives donations from a dispensing pharmacy, namely Ain Pharmacies, Inc., other organizations, and private individuals. ## Funding sources This study received support from the Tansa (formerly known as the Waseda Chronicle), an independent non‐profit news organization dedicated to investigative journalism. None of the entities providing financial support for this study contributed to the design, execution, data analyses, or interpretation of study findings and the drafting of this manuscript. ## References 1. Pham‐Kanter G. Act II of the sunshine act. PLoS Med (2014) 11. PMID: 25369363 2. Fabbri A, Santos A, Mezinska S, Mulinari S, Mintzes B. Sunshine policies and murky shadows in Europe: disclosure of pharmaceutical industry payments to health professionals in nine European Countries. Int J Health Policy Manag (2018) 7 504-509. PMID: 29935127 3. Ozaki A, Saito H, Senoo Y. Overview and transparency of nonresearch payments to healthcare organizations and healthcare professionals from pharmaceutical companies in Japan: analysis of payment data in 2016. Health Policy (2020) 124 727-735. PMID: 32439213 4. Yamamoto K, Murayama A, Ozaki A, Saito H, Sawano T, Tanimoto T. Financial conflicts of interest between pharmaceutical companies and the authors of urology clinical practice guidelines in Japan. Int Urogynecol J (2021) 32 443-451. PMID: 33151353 5. Murayama A, Ozaki A, Saito H. Pharmaceutical company payments to dermatology clinical practice guideline authors in Japan. PLoS One (2020) 15. PMID: 33048952 6. Murayama A, Hoshi M, Saito H. Nature and trends in personal payments made to the respiratory physicians by pharmaceutical companies in Japan betwenn 2016 and 2019. Respiration (2022) 101 1088. PMID: 36353778 7. Murayama A, Kida F, Ozaki A, Saito H, Sawano T, Tanimoto T. Financial and intellectual conflicts of interest among Japanese clinical practice guidelines authors for allergic rhinitis. Otolaryngol Head Neck Surg (2021) 166 8. Murayama A, Kamamoto S, Saito H. Pharmaceutical payments to Japanese board‐certified infectious disease specialists: a four‐year retrospective analysis of payments from 92 pharmaceutical companies between 2016 and 2019. Int J Environ Res Public Health (2022) 19 7417. PMID: 35742661 9. Kusumi E, Murayama A, Kamamoto S. Pharmaceutical payments to Japanese certified hematologists: a retrospective analysis of personal payments from pharmaceutical companies between 2016 and 2019. Blood Cancer J (2022) 12 54. PMID: 35393417 10. Murayama A, Yamada K, Yoshida M. Evaluation of conflicts of interest among participants of the Japanese nephrology clinical practice guideline. Clin J Am Soc Nephrol (2022) 17 819-826. PMID: 35623883 11. Polverini PJ, Lingen MW. A history of innovations in the diagnosis and treatment of oral and head and neck cancer. J Dent Res (2019) 98 489-497. PMID: 31008698 12. Rathi VK, Samuel AM, Mehra S. Industry ties in otolaryngology: initial insights from the physician payment sunshine act. Otolaryngol Head Neck Surg (2015) 152 993-999. PMID: 25779469 13. Brauer PR, Morse E, Mehra S. Industry payments for otolaryngology research: a four‐year analysis of the open payments database. Laryngoscope (2020) 130 314-320. PMID: 31059584 14. Pathak N, Fujiwara RJT, Mehra S. Assessment of nonresearch industry payments to otolaryngologists in 2014 and 2015. Otolaryngol Head Neck Surg (2018) 158 1028-1034. PMID: 29437524 15. Morse E, Fujiwara RJT, Mehra S. Increasing industry involvement in otolaryngology: insights from 3 years of the open payments database. Otolaryngol Head Neck Surg (2018) 159 501-507. PMID: 29807484 16. Morse E, Berson E, Mehra S. Industry involvement in otolaryngology: updates from the 2017 open payments database. Otolaryngol Head Neck Surg (2019) 161 265-270. PMID: 30909808 17. Clinckemaillie M, Scanff A, Naudet F, Barbaroux A. Sunshine on KOLs: assessment of the nature, extent and evolution of financial ties between the leaders of professional medical associations and the pharmaceutical industry in France from 2014 to 2019: a retrospective study. BMJ Open (2022) 12 18. Moynihan R, Albarqouni L, Nangla C, Dunn AG, Lexchin J, Bero L. Financial ties between leaders of influential US professional medical associations and industry: cross sectional study. BMJ (2020) 369 m1505. PMID: 32461201 19. Karanges EA, Ting N, Parker L, Fabbri A, Bero L. Pharmaceutical industry payments to leaders of professional medical associations in Australia: focus on cardiovascular disease and diabetes. Aust J General Pract (2020) 49 151-154 20. Saito H, Ozaki A, Kobayashi Y, Sawano T, Tanimoto T. Pharmaceutical company payments to executive board members of professional medical associations in Japan. JAMA Internal Med (2019) 179 578-580. PMID: 30715087 21. 21 Pharmaceuticals and Medical Devices Agency . List of approved products; 2022. Accessed February 24, 2022. https://www.pmda.go.jp/english/review-services/reviews/approved-information/drugs/0002.html. (2022) 22. Ozaki A, Saito H, Onoue Y. Pharmaceutical payments to certified oncology specialists in Japan in 2016: a retrospective observational cross‐sectional analysis. BMJ Open (2019) 9 23. Tringale KR, Marshall D, Mackey TK, Connor M, Murphy JD, Hattangadi‐Gluth JA. JAMA (2017) 317 1774-1784. PMID: 28464140 24. Feng H, Wu P, Leger M. Exploring the industry‐dermatologist financial relationship: insight from the open payment data. JAMA Dermatol (2016) 152 1307-1313. PMID: 27706478 25. Ozieranski P, Csanadi M, Rickard E, Tchilingirian J, Mulinari S. Analysis of pharmaceutical industry payments to UK Health Care Organizations in 2015. JAMA Network Open (2019) 2. PMID: 31225896 26. Annapureddy A, Murugiah K, Minges KE, Chui PW, Desai N, Curtis JP. Industry payments to cardiologists. Circ Cardiovasc Qual Outcomes (2018) 11 27. Kamamoto S, Murayama A, Kusumi E. Evaluation of financial relationships between Japanese certified pediatric hematologist/oncologists and pharmaceutical companies: a cross‐sectional analysis of personal payments from pharmaceutical companies between 2016 and 2019. Pediatr Blood Cancer (2022) 69. PMID: 35949170 28. 28 Ministry of Health LaW . Survey of Physicians, Dentists and Pharmacists 2018. In: Ministry of Health, Labour and Welfare; 2018.. (2018) 29. Mitchell AP, Trivedi NU, Gennarelli RL. Are financial payments from the pharmaceutical industry associated with physician prescribing?: a systematic review. Ann Internal Med (2021) 174 353-361. PMID: 33226858 30. Morse E, Hanna J, Mehra S. The association between industry payments and brand‐name prescriptions in otolaryngologists. Otolaryngol Head Neck Surg (2019) 161 605-612. PMID: 31547772 31. Morse E, Fujiwara RJT, Mehra S. The association of industry payments to physicians with prescription of brand‐name intranasal corticosteroids. Otolaryngol Head Neck Surg (2018) 159 442-448. PMID: 29865931 32. 32 Danish Medicines Agency . Doctors' notification of and application for permission to establish relations with companies; 2020. February 25, 2022. https://laegemiddelstyrelsen.dk/en/licensing/relationships/doctors/. (2020) 33. Moynihan R, Bero L, Hill S. Pathways to independence: towards producing and using trustworthy evidence. BMJ (2019) 367 l6576. PMID: 31796508 34. 34 Institute of Medicine . Conflict of Interest in Medical Research, Education, and Practice. The National Academies Press; 2009.. . Conflict of Interest in Medical Research, Education, and Practice (2009) 35. Chakroun R, Milhabet I. Conflits d'intérêts des leaders d'opinion médicaux: effets sur la confiance du grand public et des médecins généralistes. Revue d'Épidémiologie et de Santé Publique (2011) 59 233-242 36. Fadlallah R, Nas H, Naamani D. Knowledge, beliefs and attitudes of patients and the general public towards the interactions of physicians with the pharmaceutical and the device industry: a systematic review. PLoS One (2016) 11. PMID: 27556929 37. Murayama A, Senoo Y, Harada K. Awareness and perceptions among members of a Japanese cancer patient advocacy group concerning the financial relationships between the pharmaceutical industry and physicians. Int J Environ Res Public Health (2022) 19 3478. PMID: 35329160 38. Lenzer J, Hoffman JR, Furberg CD, Ioannidis JPA.. Ensuring the integrity of clinical practice guidelines: a tool for protecting patients. BMJ (2013) 347 f5535. PMID: 24046286 39. Rothman DJ, McDonald WJ, Berkowitz CD. Professional medical associations and their relationships with industry: a proposal for controlling conflict of interest. JAMA (2009) 301 1367-1372. PMID: 19336712 40. 40 Japan Pharmaceutical Manufacturers Association . Data Book 2021; 2021. March 2, 2022. https://www.jpma.or.jp/news_room/issue/databook/2021_en/lofurc0000004we3-att/DB2021_en_full.pdf. (2021) 41. 41 Japan Pharmaceutical Manufacturers Association . Regarding the transparency guideline for the relation between corporate activities and medical institutions; 2018. Accessed March 4, 2022. https://www.jpma.or.jp/english/code/transparency_guideline/eki4g60000003klk-att/transparency_gl_intro_2018.pdf. (2018)
--- title: Office‐Based Multilevel Radiofrequency Ablation for Mild‐to‐Moderate Obstructive Sleep Apnea authors: - Howard Herman - Jordan Stern - David M. Alessi - Keith A. Swartz - Marion Boyd Gillespie journal: OTO Open year: 2023 pmcid: PMC10046721 doi: 10.1002/oto2.19 license: CC BY 4.0 --- # Office‐Based Multilevel Radiofrequency Ablation for Mild‐to‐Moderate Obstructive Sleep Apnea ## Body Obstructive sleep apnea syndrome (OSAS) causes repetitive episodes of upper airway obstruction during sleep, usually in association with a reduction in blood oxygen saturation. 1 The prevalence of moderate‐to‐severe OSAS in the middle‐aged population is estimated to be up to $23\%$ in women and $49\%$ in men. 2, 3 On the basis of these numbers, the global prevalence of sleep‐disordered breathing is estimated to be close to 1 billion people. 2 Untreated mild‐to‐moderate obstructive sleep apnea (OSA) is associated with increased healthcare costs, motor vehicle accidents, and loss of work productivity. 4, 5 First‐line treatment for many OSAS patients is nasal continuous positive airway pressure (CPAP). When used adequately, CPAP improves sleepiness, performance, quality of life, and cardiovascular risk, 6, 7, 8 however $10\%$ to $35\%$ of patients fail to maintain CPAP use over time. 9, 10, 11, 12, 13 Radiofrequency ablation (RFA) has demonstrated promise in reducing snoring and sleepiness symptoms. 14 The procedure can be performed in the ambulatory setting under topical and local anesthesia. RFA has been applied as a second‐line treatment or adjunctive therapy with other sleep procedures for mild‐to‐moderate OSA if CPAP therapy is not adhered to or tolerated. 15 The study's aim is to assess the treatment effect and safety of RFA in a cohort of non‐obese patients with mild‐to‐moderate OSAS. The study was specifically designed to address the effectiveness of multilevel (soft palate and tongue) treatment when applied over 3 treatment sessions. ## Abstract ### Objective Investigate multilevel radiofrequency ablation (RFA) as an alternative therapy for patients with mild‐to‐moderate obstructive sleep apnea (OSA). ### Study Design Prospective, open‐label, single‐arm, nonrandomized clinical trial. ### Setting Multicenter academic and private clinics. ### Methods Patients with mild‐to‐moderate OSA (apnea‐hypopnea index [AHI] 10‐30; body mass index ≤ 32) were treated with 3 sessions of office‐based RFA to the soft palate and tongue base. The primary outcome was a change in the AHI and oxygen desaturation index (ODI $4\%$). Secondary outcomes included subjective sleepiness level; snoring level; and sleep‐related quality of life. ### Results Fifty‐six patients were enrolled, with 43 ($77\%$) completing the study protocol. Following 3 sessions of office‐based RFA to the palate and base of the tongue, the mean AHI decreased from 19.7 to 9.9 ($$p \leq .001$$), while the mean ODI ($4\%$) decreased from 12.8 to 8.4 ($$p \leq .005$$). Mean Epworth Sleepiness Scale scores declined from 11.2 (±5.4) to 6.0 (±3.5) ($$p \leq .001$$), while Functional Outcomes of Sleep Questionnaire scores improved from a mean of 14.9 at baseline to 17.4 ($$p \leq .001$$). The mean visual analog scale snoring scale was reduced from 5.3 (±1.4) at baseline to 3.4 (±1.6) at 6 months posttherapy ($$p \leq .001$$). ### Conclusion Office‐based, multilevel RFA of the soft palate and base of the tongue is a safe and effective treatment option with minimal morbidity for properly selected patients with mild‐to‐moderate OSA who are intolerant or refuse continuous positive airway pressure therapy. ## Study Design and Objectives This study is a multicenter, prospective, nonrandomized, Food and Drug Administration (FDA)‐approved study (NCT02349893) performed from 2017 to 2020. Institutional Review Board (IRB) approval was granted at each site by various IRBs including the University of Tennessee Health Science Center (M.B.G.); Solutions IRB, LLC (J.S.; H.H.; D.M.A.); and WCG IRB (K.A.S.). The device used in the study is the Celon ProSleep Plus (Olympus), a single‐prong bipolar radiofrequency applicator currently available and FDA‐approved within the United States for submucosal coagulation of the soft palate for the treatment of habitual snoring. The study's aim was to evaluate the safety of multilevel (soft palate and tongue base) RFA therapy for patients with mild‐to‐moderate OSA and to demonstrate the clinical effect 6 months after treatment. The study was funded by Olympus Winter & Ibe GmbH which covered study costs for participants and study sites. The study design was approved by the FDA to support an application for extended use in the base of the tongue region for patients with mild‐to‐moderate OSA. The full study protocol with detailed descriptions of results is available in Supplemental Appendix I, available online. ## Participants Study participants were a cohort of adult patients (22 years and above) with mild‐to‐moderate OSA (apnea‐hypopnea index [AHI] 10‐30) and a body mass index (BMI) ≤ 32 kg/m2; intolerance or inadequate adherence to CPAP; self‐report of daytime somnolence; evidence of narrowing of the airway at the level of the soft palate and tongue base on supine fiberoptic examination; no prior surgical treatment for OSAS other than nasal surgery or tonsillectomy; and a regular nightly sleep partner. Exclusion criteria included comorbid sleep disorders; tonsillar hypertrophy (Brodsky 3‐4+); nasal or supraglottic obstruction on examination; ASA Classes III‐V; and drug or alcohol abuse or current participation in another research study. An AHI range of 10 to 30 is selected to conform with the Sher criteria which define surgical success as a $50\%$ reduction in AHI and an overall AHI <20. There was concern that patients with baseline AHI between 5 and 10 could be considered unsuccessful even if their overall AHI was normalized <5 following treatment. Screening studies included home WatchPAT 200S‐3 for diagnosis of OSAS or full in‐laboratory polysomnography (PSG; performed within 12 months of study enrollment). Although the WatchPAT device may underestimate the severity of OSA, screening with the device was considered acceptable since it would be followed by a full‐night baseline PSG. Following informed consent, all subjects underwent a subsequent baseline in‐laboratory PSG unless a full in‐laboratory PSG within 6 months was available. PSGs were scored using American Academy of Sleep Medicine criteria for apnea (>$90\%$ reduction in peak thermal sensor from baseline for ≥10 seconds) and hypopnea (≥$50\%$ reduction in baseline nasal pressure signal for ≥10 seconds with either ≥$3\%$ desaturation event or associated arousal). Soft palate RFA treatment: A single‐prong RFA applicator (CelonProSleep Plus; Olympus) was used to create 7 lesions of 54 joule (J) each (Celon Power Setting 12 W) in a prescribed pattern (Figure 1). Figure 1: Diagram of approximate treatment sites on the soft palate. Tongue base RFA treatment: Using the same single‐prong RFA applicator, each patient then underwent 6 lesions of RFA treatment (CelonProSleep plus single‐prong applicator 80‐84 J each) to the base of the tongue using the prescribed pattern (Figure 2). Figure 2: Diagram of the approximate treatment sites on the base of the tongue. ## Intervention Participants underwent 3 radiofrequency treatments (4‐6 weeks apart) in an outpatient setting. Topical Cetacaine (benzocaine $14.0\%$; butamben $2.0\%$; and tetracaine hydroclhloride $2.0\%$) or HurriCaine ($20\%$ benzocaine) spray was applied to mucosal surfaces, followed by injection of 5 to 8 cc of $1\%$ lidocaine with 1:100,000 epinephrine to both the body of the soft palate and the dorsal tongue at the level of the circumvallate papilla. ## Postprocedure Care Patients were monitored for 3 hours following treatment. In addition, participants were provided with prescriptions for an antibiotic (amoxicillin or clindamycin), oral steroid for 7 days (methylprednisolone taper pack), and oral pain medication (hydrocodone/acetominophen $\frac{5}{325}$ mg) to be taken as needed. Patient follow‐up occurred on Days 1, 3, and 10 posttreatment. At the follow‐up visits, patients completed the pain, speech, and swallowing visual analog scale (VAS) scale and a clinical examination was performed. The use of pain medication (hydrocodone‐acetaminophen $\frac{5}{325}$ mg; maximum 2 tablets every 6 hours or 8 tablets per day; Narco®) in the first 7 days following the procedure was recorded. Adverse events (AEs) were reviewed and coded by severity and attributed to either the surgical procedure or the device. ## Outcome Measures The study's endpoint was to demonstrate a clinically significant reduction of OSA symptoms by showing an adequate reduction in AHI determined by PSG results 6 months after treatment. Treatment response was defined as a ≥$50\%$ reduction in the baseline AHI and an overall AHI <20. Secondary endpoints included change in the Epworth Sleepiness Scale (ESS); VAS of speech and swallowing; the Functional Outcomes of Sleep Questionnaire (FOSQ) (score range is 5‐20 where a higher score indicates higher activity level); and a drowsiness in the past week VAS (score range, 0‐100; 0‐9 represents minimal drowsiness; 10‐39 represents mild drowsiness; 40‐69 represents moderate drowsiness; and 70‐100 represents significant drowsiness). The Bed Partner Questionnaire was completed by the regular bed partner (score range is 0‐10, while 0‐3 means no snoring problem; 4‐6 mild snoring; 7‐9 moderate snoring; and 10 represents severe snoring disturbance). ## Statistical Analysis An a priori power analysis estimated a minimum of 34 subjects were needed to complete the study to demonstrate study endpoints with an anticipated dropout rate of 10 with an enrollment goal of at least 8 subjects per study site. Data were analyzed using the statistical software R: A Language and Environment for Statistical Computing, R Core Team, R Foundation for Statistical Computing, 2021 Version 4.05 2021. For continuous parameters, descriptive statistics including mean, standard deviation, median, and range are reported. For ordinal parameters, counts and percentages are reported in addition to the mean, standard deviation, median, and range. The statistical analysis of the PSG data at baseline and follow‐up visits was analyzed using paired t test and linear‐by‐linear χ 2 tests for the AHI levels. Comparison of questionnaires' scores (baseline vs follow‐up visits) was tested using paired t test. Testing of the repeated measurements is carried out with a mixed model with random intercept using nlme package. All statistical tests were performed at a significance level of 0.05, with no corrections for multiple testing. ## Results Fifty‐six patients were recruited for the study with 43 patients completing the protocol. Thirteen dropouts included 12 patients who were lost to follow‐up and 1 patient with an unreadable final PSG due to device failure. This patient refused to undergo a repeat PSG. The baseline characteristics of study participants are shown in Table 1. In general, subjects were middle‐aged overweight men demonstrating OSAS‐related quality of life deficits and excessive daytime somnolence. Of the 43 patients who completed the protocol, 7 ($16\%$) had a prior tonsillectomy and 7 ($16\%$) had undergone a previous septoplasty. Table 1 \| Baseline characteristic \| Normative value \| Subject value \| \| --- \| --- \| --- \| \| Gender (% male) \| \| 30/43 (70) \| \| Mean age, y (SD) \| \| 50.7 (11.2) \| \| Mean body mass index, kg/m2 (SD) \| <25 \| 27.2 (3.7) \| \| Mean apnea‐hypopnea index (events/hour sleep time) (SD) \| <5 \| 19.7 (7.1) \| \| Mean oxygen desaturation index, 4% (SD) \| < 5 \| 12.8 (7.7) \| \| Mean lowest O2 saturation (%) (SD) \| >90% \| 84.2 (5.3) \| \| Mean Epworth Sleepiness Scale score (SD) \| <10 \| 11.2 (5.4) \| \| Mean snoring VAS score (SD) \| 0 \| 5.3 (1.3) \| \| Mean bed partner snoring VAS score (SD) \| 0‐3 \| 6.9 (2.2) \| \| Mean functional outcomes of sleep Questionnaire (FOSQ) (SD) \| >17.8 \| 14.9 (4.4) \| \| Mean drowsiness in the past week VAS score (SD) \| <10 \| 54.3 (26.3) \| ## Primary Outcome Measures The primary outcome measure of AHI change from baseline to 6 months postintervention is summarized in Table 2. Overall, $\frac{22}{43}$ ($51\%$) subjects were considered complete responders with a ≥$50\%$ reduction in baseline AHI and an overall AHI <20 at study completion. Whereas 25 patients ($58\%$) had AHI scores below 20 at baseline, $\frac{39}{43}$ ($91\%$) had scores below 20 following treatment. A statistically significant reduction in AHI ($$p \leq .001$$) was observed at 6 months follow‐up. Table 2 \| Outcome measure \| Baseline \| 6‐mo posttreatment \| p value \| \| --- \| --- \| --- \| --- \| \| Responder (AHI reduction ≥50; overall AHI <20) (%) \| \| 22/43 (51) \| \| \| Normal (AHI <5) (%) \| 0 (0) \| 16 (37) \| \| \| Mild OSA (AHI 5‐15) (%) \| 16 (37) \| 16 (37) \| \| \| Moderate OSA (AHI 15‐30) (%) \| 27 (63) \| 11 (26) \| \| \| Mean AHI (±SD)a \| 19.70 (7.10) \| 9.86 (8.28) \| 0.001 \| \| Median AHI (range)b \| 17.80 (10.40‐34.90) \| 7.5 (0.00‐35.90) \| 0.001 \| Subgroup analysis was performed on $\frac{27}{43}$ ($63\%$) of subjects with moderate OSA (AHI >15‐30) and $\frac{16}{43}$ ($37\%$) with mild OSA (AHI 10‐15) on the screening WatchPAT examination. The baseline AHI results of the PSG in the moderate group ranged from 15.9 to 34.9 and included 4 patients with AHI >30 (range, 30‐34.9) who were found to have severe OSA on baseline PSG. A total of $\frac{15}{27}$ ($56\%$) of the moderate group demonstrated a $50\%$ reduction of AHI with an overall AHI <20 at study completion with a mean AHI reduction of −13.1 ($$p \leq .0000023$$) from baseline mean of 23.7 (±5.9) to final mean of 10.7 (±9.6). The AHI results of the mild group (baseline AHI; range, 11‐14.9) demonstrated $\frac{8}{16}$ ($50\%$) with a $50\%$ reduction in final AHI with a mean AHI reduction of −4.41 ($$p \leq .009$$) from a baseline mean of 12.9 (±1.4) to 8.5 (±5.5). Oxygen desaturation index (ODI; $4\%$ desaturation) results are shown in Table 3. Eleven ($26\%$) patients had incomplete ODI scores due to the inability to obtain the baseline ODI from the historical PSG sleep study performed within 6 months of enrollment. Three of these 11 patients did not have ODI scored at the 6‐month follow‐up PSG. Overall, $\frac{23}{32}$ ($72\%$) demonstrated ODI reduction following treatment with a mean ODI reduction of $33\%$ ($$p \leq .006$$). An ODI reduction of ≥$50\%$ was noted in $\frac{16}{32}$ ($50\%$) of subjects with complete data. Table 3 \| Outcome measure \| Baseline \| 6‐mo posttreatment \| p value \| \| --- \| --- \| --- \| --- \| \| ODI 4 reduction ≥25% \| \| 23/32 (72%) \| \| \| Mean ODI (±SD)a \| 12.79 (7.74) \| 8.36 (8.74) \| 0.006 \| \| Median ODI (range)b \| 11.65 (0.00‐31.20) \| 6.32 (0.00‐30.40) \| 0.008 \| The 21 ($49\%$) treatment nonresponders were offered standard of care management of their OSA following the completion of the study including CPAP, oral appliance therapy, and/or additional surgical procedures. This care was outside of the study and was not included as part of the study data. ## Secondary Outcome Measures All self‐reported questionnaire responses at the 6‐month follow‐up visit demonstrated statistically significant improvement compared to baseline (Table 4). Based on the bed partner report, intrusive snoring (very intense snoring or bed partner leaving the room) was reduced from $63\%$ at baseline to $7\%$ at 6 months follow‐up. The percentage of participants who reported normal ESS scores (<10) increased from $67\%$ at baseline to $88\%$ at 6 months posttreatment with a reduction of 5.6 points on average. At baseline, only $14\%$ reported a normal FOSQ score (>17.9) but increased to $58\%$ at 6 months posttreatment with an average increase of 2.5 points. In addition, subjects endorsed a $43\%$ mean reduction in drowsiness over the prior week. Table 4 \| Item \| Baseline mean (SD) \| 6‐mo posttreatment mean (SD) \| p value (paired T test) \| \| --- \| --- \| --- \| --- \| \| Snoring VAS \| 5.33 (1.34) \| 3.41 (1.66) \| 0.001 \| \| Bed partner snoring VAS \| 7 (2.16) \| 3.12 (2.38) \| 0.001 \| \| Epworth Sleepiness Scale \| 11.19 (5.40) \| 5.95 (3.51) \| 0.001 \| \| Functional Outcome of Sleep Questionnaire \| 14.91 (4.43) \| 17.41 (2.17) \| 0.001 \| \| Drowsiness in the past week VAS \| 54.34 (26.33) \| 31.08 (24.72) \| 0.001 \| ## AEs No serious AEs were observed during the study. Eleven AEs were reported in 5 ($12\%$) patients. One patient ($2\%$) had mild dysphagia that resolved after 3 days. Five patients ($9\%$) had tissue edema with or without mucosal ulceration treated with saline gargles or oral antibiotics. Two of the patients received an additional oral steroid dose pack. Pain level was documented using the VAS scale (0‐10, where “0” indicates no pain and “10” indicates unbearable pain). The average pain level immediately after RFA treatment was 1.7 (±1.1) and decreased to a mean of 0.29 (0.72) by Day 7. In addition to the low levels of pain, the vast majority of study patients reported complete recovery of the soft palate and the tongue base at 6 weeks after the RFA treatment ($95\%$ and $97\%$, respectively) and $100\%$ at 6 months. ## Discussion OSAS is a prevalent disorder estimated to be present in $10\%$ to $17\%$ of adult men and $3\%$ to $9\%$ of adult women. 16 This translates into approximately 13 million individuals over the age of 30 within the United States. 17 The burden of the disorder increases with age with a prevalence of $50\%$ in age groups older than 65 years. 17 Untreated OSAS is associated with an increased risk of cardiovascular disease, cerebrovascular disease, hypertension, perioperative complications, and premature all‐cause mortality. 18 Untreated OSAS causes a profound reduction in quality of life due to symptoms of snoring, poor sleep quality, and daytime sleepiness. CPAP is the recognized first‐line therapy for OSAS, however, up to $40\%$ to $50\%$ of patients fail to adhere to the recommended use of 4 or more hours of therapy per night. With regard to mild‐to‐moderate OSA, it is estimated that up to 1 in 5 normal‐weight adults (BMI 25‐28) in the United States has mild OSA, and 1 in 15 has at least moderate OSA. 19 Therefore, there is a recognized need for alternative therapies to meet this public health challenge. Non‐CPAP treatment options for moderate OSA (AHI <30) include mandibular advancement devices (MAD) and/or various surgical interventions. MADs are effective for mild‐to‐moderate OSA but may not be acceptable to all patients due to cost; the need for nightly use; and side effects such as drooling, temporomandibular joint discomfort, tooth pain, and xerostomia. 20 Surgical interventions are effective in select patients but have variable outcomes, involving expense, anesthesia risk, and the potential for bleeding, infection, pain, and poor wound healing. The ideal treatment is affordable, device‐free, and office‐based with minimal side effects that effectively reduce snoring and daytime sleepiness. RFA has been used as a potentially less morbid approach to stiffen and provide structural support to collapsible upper airway segments. Radiofrequency energy causes tissue ions to become agitated due to changes in electrical flow inherent in alternating current at relatively low temperatures (60‐95°C). 21 The lesion created by RFA creates protein coagulation and results in congestion, edema, and an acute inflammatory response within the first 24 hours. Over a period of 72 hours, the treated area creates focal necrosis which is transformed into fibrotic tissue over the course of 10 days. 21 A variety of FDA‐cleared RFA devices have demonstrated promise as a treatment alternative for OSA in multiple published studies. 22, 23, 24, 25 In these studies, repeated RFA of the soft palate and base of the tongue region resulted in reductions in AHI and daytime sleepiness without significant complications. RFA has several advantages over traditional surgical approaches including its ability to address multiple levels of the airway (nose, palate, and tongue); its ability to perform in the office under local anesthesia; lower cost; and minimal pain and morbidity. The present trial was designed to maximize the above advantages of RFA for a cohort of patients most likely to demonstrate benefit, namely nonobese patients with symptomatic mild‐to‐moderate OSAS. Based on the prior literature, it is clear that RFA is most effective as a treatment is used to treat multilevel sites of collapse (soft palate and tongue) in a repeated fashion designed to allow sufficient volumetric tissue reduction and fibrosis to occur. The results of this study demonstrate that repeated application of RFA energy to multiple sites of airway collapse in appropriately selected patients results in a significant reduction in AHI with improvement in snoring and daytime sleepiness with a low level of patient morbidity. Limitations of the study are mainly due to the nonrandomized design without placebo control. Although subjective patient questionnaires are subject to the placebo effect, the primary study outcomes of AHI and ODI ($4\%$) were based on objective testing criteria. Thirteen ($23\%$) patients were lost to follow‐up and, therefore, it is difficult to know if their outcomes would be consistent with the group that completed the entire course of therapy. It is reasonable to assume that some patients may have difficulty tolerating a course of care that requires several invasive treatment sessions over a 3‐ to 4‐month period. The study endpoint at 6 months does not allow projection of long‐term results which will require further follow‐up in order to determine the length of treatment effect. In addition, the study cohort was limited to patients with mild‐to‐moderate OSA (AHI 10‐30) with BMI ≤32 kg/m2 and therefore similar results cannot be assumed in patients with more severe OSAS who may benefit from other currently available treatment options. ## Conclusions This study evaluated the safety and effectiveness of RFA ablation treatment in the base of the tongue and soft palate for improving mild‐to‐moderate OSAS. The study found that RFA provides significant improvements in PSG measures of OSA and clinically meaningful improvements in patient self‐reported outcomes. The therapy had few side effects and was well‐tolerated with a low level of pain and morbidity. ## Author Contributions Howard Herman, site PI, subject enrollment, data acquisition, data analysis, manuscript preparation, and review; Jordan Stern, site PI, subject enrollment, data acquisition, data analysis, manuscript preparation and review; David M. Alessi, site PI, subject enrollment, data acquisition, data analysis, manuscript preparation and review; Keith A. Swartz, site PI, subject enrollment, data acquisition, data analysis, manuscript preparation and review; Marion Boyd Gillespie, site PI, subject enrollment, data acquisition, data analysis, manuscript preparation and review; research presentation. ## Competing interests None. ## Sponsorships Olympus Winter & Ibe GmbH. ## Funding source This article was supported by Olympus Winter & Ibe GmbH, Hamburg, Germany (Institutional Review Board fees; compensation for study coordinator; surgeon; statistician; subject incentive; disposable devices; and equipment loan). ## References 1. 1 Diagnostic Classifications Steering Committee . The International Classification System of Sleep Disorders Diagnostic and Coding Manual. American Sleep Disorders Association; 1990.. The International Classification System of Sleep Disorders Diagnostic and Coding Manual (1990) 2. Heinzer R, Vat S, Marques‐Vidal P. Prevalence of sleep‐disordered breathing in the general population: the HypnoLaus study. Lancet Respir Med (2015) 3 310-318. PMID: 25682233 3. Goodday RH. Nasal respiration, nasal airway resistance, and obstructive sleep apnea syndrome. Oral Maxillofac Surg Clin North Am (1997) 9 167-177 4. Peppard PE, Young T, Palta M, Skatrud J. Prospective study of the association between sleep‐disordered breathing and hypertension. N Engl J Med (2000) 342 1378-1384. PMID: 10805822 5. Wolf J, Lewicka J, Narkiewicz K. Obstructive sleep apnea: an update on mechanisms and cardiovascular consequences. Nutr Metab Cardiovasc Dis (2007) 17 233-240. PMID: 17314035 6. Peker Y, Hedner J, Norum J, Kraiczi H, Carlson J. Increased incidence of cardiovascular disease in middle‐aged men with obstructive sleep apnea: a 7‐year follow‐up. Am J Respir Crit Care Med (2002) 166 159-165. PMID: 12119227 7. Engleman HM, Martin SE, Kingshott RN, Mackay TW, Deary IJ, Douglas NJ. Randomised placebo controlled trial of daytime function after continuous positive airway pressure (CPAP) therapy for the sleep apnoea/hypopnoea syndrome. Thorax (1998) 53 341-345. PMID: 9708223 8. Montserrat JM, Ferrer M, Hernandez L. Effectiveness of CPAP treatment in daytime function in sleep apnea syndrome. A randomized controlled study with an optimized placebo. Am J Respir Crit Care Med (2001) 164 608-613. PMID: 11520724 9. Malone S, Liu PP, Holloway R. Obstructive sleep apnea in patients with dilated cardiomyopathy: effects of continuous positive airway pressure. Lancet (1991) 338 1480-1484. PMID: 1683918 10. Waldhorn RE, Herrick TW, Nguyen MC, O'Donnell AE, Sodero J, Potolicchio SJ. Long‐term compliance with nasal continuous positive airway pressure therapy of obstructive sleep apnea. Chest (1990) 97 33-38. PMID: 2403899 11. Meurice JC, Dore P, Paquereau J. Predictive factors of long term compliance with nasal continuous positive airway pressure treatment in sleep apnea syndrome. Chest (1994) 105 429-433. PMID: 8306741 12. Rauscher H, Formanek D, Popp W, Zwick H. Nasal CPAP and weight loss in hypertensive patients with obstructive sleep apnoea. Thorax (1993) 48 529-533. PMID: 8322241 13. Grunstein RR. Sleep‐related breathing disorders, 5: nasal continuous positive airway pressure treatment for obstructive sleep apnoea. Thorax (1995) 50 1106-1113. PMID: 7491563 14. Powell NB, Riley RW, Troell RJ, Li K, Blumen MB, Guilleminault C. Radiofrequency volumetric tissue reduction of the palate in subjects with sleep‐disordered breathing. Chest (1998) 113 1163-1174. PMID: 9596289 15. Aurora RN, Casey KR, Kristo D. Practice parameters for the surgical modifications of the upper airway for obstructive sleep apnea in adults. Sleep (2010) 33 1408-1413. PMID: 21061864 16. Peppard PE, Young T, Barnet JH, Palta M, Hagen EW, Hla KM. Increased prevalence of sleep‐disordered breathing in adults. Am J Epidemiol (2013) 177 1006-1014. PMID: 23589584 17. Young T, Palta M, Dempsey J, Peppard PE, Nieto FJ, Hla KM. Burden of sleep apnea: rationale, design, and major findings of the Wisconsin Sleep Cohort study. WMJ (2009) 108 246-249. PMID: 19743755 18. Rich J, Raviv A, Raviv N, Brietzke SE. An epidemiologic study of snoring and all‐cause mortality. Otolaryngol Head Neck Surg (2011) 145 341-346. PMID: 21493281 19. Young T, Peppard PE, Gottlieb DJ. Epidemiology of obstructive sleep apnea: a population health perspective. Am J Respir Crit Care Med (2002) 165 1217-1239. PMID: 11991871 20. Fritsch VA, Gerek M, Gillespie MB, Har‐El G. Head and Neck Surgery (2013) 450-457 21. Powell NB, Riley RW, Troell RJ, Blumen MB, Guilleminault C. Radiofrequency volumetric reduction of the tongue. Chest (1997) 111 1348-1355. PMID: 9149593 22. Woodson BT, Steward DL, Weaver EM, Javaheri S. A randomized trial of temperature‐controlled radiofrequency, continuous positive airway pressure, and placebo for obstructive sleep apnea syndrome. Otolaryngol Head Neck Surg (2003) 128 848-861. PMID: 12825037 23. Steward DL, Weaver EM, Woodson BT. A comparison of radiofrequency treatment schemes for obstructive sleep apnea syndrome. Otolaryngol Head Neck Surg (2004) 130 579-586. PMID: 15138424 24. Riley RW, Powell NB, Li KK, Weaver EM, Guilleminault C. An adjunctive method of radiofrequency volumetric tissue reduction of the tongue for OSAS. Otolaryngol Head Neck Surg (2003) 129 37-42. PMID: 12869914 25. Steward DL, Weaver EM, Woodson BT. Multilevel temperature‐controlled radiofrequency for obstructive sleep apnea: extended follow‐up. Otolaryngol Head Neck Surg (2005) 132 630-635. PMID: 15806059
--- title: 'The Influence of Obesity on Melanoma and Sentinel Lymph Node Diagnosis: A Retrospective Monocentric Study in 1001 Patients' authors: - Filipa Almeida Oliveira - Julie Klose - Hans-Joachim Schulze - Marta Ribeiro Teixeira - Alexander Dermietzel - Sascha Wellenbrock - Grit-Sophie Herter-Sprie - Tobias Hirsch - Maximilian Kueckelhaus journal: Cancers year: 2023 pmcid: PMC10046741 doi: 10.3390/cancers15061806 license: CC BY 4.0 --- # The Influence of Obesity on Melanoma and Sentinel Lymph Node Diagnosis: A Retrospective Monocentric Study in 1001 Patients ## Abstract ### Simple Summary The impact of obesity on melanoma has rarely been researched. Incidence of obesity is rapidly growing and melanoma is one of the most prevalent types of cancers worldwide. Several studies have shown that overweight and obese populations not only have a higher risk of developing melanoma but also tend to present with thicker melanomas at the time of diagnosis. Given that melanoma thickness is one of the main predictors of the melanoma prognosis, a worse prognosis in a patient with obesity would be expectable. However, this has not yet been demonstrated in the literature. Our study is the first to show that obese patients are twice more likely to present with lymph-node metastases. Lymph node metastases is the second most important prognosis predictor of melanoma. Our findings, therefore, raise important questions regarding the screening and treatment of obese patients with melanoma. ### Abstract [1] Background: While obesity is a known independent risk factor in the development of melanoma, there is no consensus on its influence on melanoma prognosis. [ 2] Methods: In a monocentric retrospective study, data was collected from patients who underwent sentinel lymph node (SLN) biopsy for stage IB-IIC melanoma between 2013 and 2018. Patients were divided into groups according to their body mass index (BMI). The association between BMI and melanoma features, as well as the risk factors for metastases in SLN were examined. [ 3] Results: Of the 1001 patients, 336 had normal weight (BMI < 25), 402 were overweight (BMI >= 25 and <30), 173 obese (BMI >= 30 and <35) and 90 extremely obese (BMI >= 35). Overweightness and obesity were associated with higher tumor thicknesses at time of diagnosis. Ulceration was not influenced by the patient’s weight. Metastases in sentinel lymph node was almost twice more likely in extremely obese patients than in normal weight patients. Independent risk factors for metastases in SLN in our study were tumor thickness, ulceration, and BMI > 35. [ 4] Conclusions: This is the first study to show higher metastases rates in high-BMI patients with melanoma, raising important questions regarding the screening and treatment of this specific patient population. ## 1. Introduction Melanoma is one of the most common malign tumors worldwide and its incidence is increasing [1,2]. The risk of melanoma development depends on an interaction between environmental factors, most importantly UV radiation, and predisposing host factors. The latter consist of genetic predisposition, phenotype, family history, and number of melanocytic nevi [2]. Obese populations have been associated with a higher risk of developing several cancers, including melanoma. This was shown by Oh et al., who prospectively analyzed the risk of cancer development in a cohort of over 700,000 healthy men in Korea over a 10-year period [3] and by Samanic et al. in a cohort of 4,500,700 male American veterans, over a period of 27 years [4]. Similar results linking obesity to melanoma development were found in a prospective Scandinavian study with 362,552 Swedish men [5]. Dennis et al. also found a clear association between melanoma development and obesity in a prospective cohort of farmers and their spouses, who were followed up for 10 years [6]. A correlation between melanoma and obesity was also found in two case-control studies, in which obesity prevalence in melanoma patients was higher than in the control groups [7,8]. Obesity not only seems to predispose cancer development but also to negatively influence the prognosis of several cancer entities such as colorectal, liver, gallbladder, pancreatic, breast, and ovarian cancers, among others [9]. Whether obesity also negatively influences melanoma prognosis has been studied by several authors [10]. Melanoma prognosis at time of diagnosis is defined by three main tumor features: Breslow tumor thicknesses, ulceration, and metastases in sentinel lymph nodes (SLN) [11]. The association between Breslow tumor thickness and obesity is not unanimous. Skowron et al. studied 427 melanoma patients and concluded that BMI >= 30 was an independent risk factor for the development of thick melanoma [12]. Similar results were found by Gandini et al. in an multicentric Italian study with 2738 patients. In this study, this association could already be observed at BMI >= 25 [13]. Other authors described a gender-based relation between obesity and melanoma. Giorgi et al. found an association between melanoma with tumor thickness >= 1 mm and BMI >= 25 only in females, especially postmenopausal [14], whereas Stenehjem et al., with the largest cohort of 2570 patients, only discovered an association between tumor thickness and higher BMI in males [15]. Ulceration has been investigated for its relation to obesity, but no significant correlation was found [12,16,17]. Metastases in sentinel lymph nodes and their relationship with obesity has only been mentioned by Shreckengost et al. as not existent [18]. To our knowledge, no previous studies have addressed this possible causality. The aim of our study was to understand the influence of obesity on melanoma features and melanoma main predictors. The study was performed in a high-volume center for skin malignancy care. ## 2. Materials and Methods In our institution, stage IB-IIC melanoma patients received a sentinel lymph node biopsy in the Department of Plastic Surgery or in the Department of Oral-Maxillofacial Surgery, based on the respective tumor location. ## 2.1. Data Collection Retrospectively, patients were identified as those with melanoma Stage IB-IIC in the trunk and limbs and who were submitted to sentinel lymph node biopsy between 2013 and 2018 at our institution. Patient data (age, gender, and body mass index—BMI) and tumor characteristics (tumor location, Breslow tumor thickness, ulceration, metastases in sentinel lymph node, extracapsular spread, S100 value) were extracted from the hospital’s internal information system. Inclusion criteria were age over 18 years old, known body mass index and known Breslow tumor thickness. ## 2.2. Data Analysis Patients were divided in four groups according to their BMI: normal weight (BMI >= 18.5 and <25), overweight (BMI >= 25 and <30), obese (BMI >= 30 and <35) and extremely obese (BMI >= 35). Differences between BMI groups with respect to age, gender, and melanoma characteristics were analyzed. To analyze the overall differences between the BMI groups, Fisher’s test and Chi-square test were used for categorical characteristics (Table 1), and Kruskal–Wallis test was used for metric characteristics (Table 2). If the p value of these global tests was less than the selected significance level ($p \leq 0.5$), then the BMI groups were compared in pairs with the Mann–Whitney U test and the significance level was adjusted with the Bonferroni–Holm procedure to find significant differences between BMI groups (Table 3 and Table 4). Additionally, each patient and tumor characteristic was analyzed for the chance of metastases in the sentinel lymph node biopsy using odds ratio with $95\%$ confidence interval (OR ($95\%$ CI)) (Table 5 and Table 6). A sub-analysis of the risk of metastases in the sentinel lymph node biopsy was performed on patients with TD < 1 mm. Furthermore, two multivariate analyses (logistic regressions) were calculated, one where all characteristics were considered (full model) and one where only characteristics with significant influence on sentinel lymph node positivity ($p \leq 0.05$) were taken into account (stepwise selection) (Table 7). ## 3.1. Demographic Characteristics Patients whose tumor thickness was unknown or not accurately determined ($$n = 10$$) were excluded. A total of 1001 patients met the inclusion criteria for our study. The median age was 58.0 ± SD 15.3 years and the median Breslow tumor thickness was 2.9 ± SD 4.1 mm. Trunk was the most frequent tumor location (415–$41.5\%$), followed by the lower extremity (363–$36\%$) and the upper extremity (223–$22\%$). Twenty-two percent of the patients had ulcerated melanomas. Of 1001 patients, 483 ($48.3\%$) were male and 518 ($51.7\%$) were female. A total of 336 ($34\%$) patients had normal weight (BMI >= 18.5 and <25), 402 ($40\%$) were overweight (BMI >= 25 and <30), 173 ($17\%$) were obese (BMI >= 30 and <35) and 90 ($9\%$) extremely obese (BMI >= 35). Underweight patients (BMI < 18.5) were excluded from our study. A total of $37.9\%$ of patients had metastases in the sentinel lymph node biopsy and $10.0\%$ of the patients presented with extracapsular spread. ## 3.2. Analysis of BMI Groups BMI groups statistically differed regarding Breslow tumor thickness, metastases in sentinel lymph node biopsy, age, and gender. Breslow tumor thickness in the normal-weight patients (median 1.6 ± SD 5.9 mm) was lower than in all the other BMI groups ($p \leq 0.002$, p-Value Kruskal–Wallis test). Among overweight, obese, and extremely obese groups, differences in Breslow tumor thickness were not statistically significant (respectively, 2.1 ± SD 2.5 mm, 2.0 ± SD 3.2 mm and 2.3 ± SD 3.2 mm) ($p \leq 0.1$) (Table 1 and Table 3 and Figure 1a). Normal-weight patients were younger (median age 54 ± SD 16.9 years) than overweight patients (median 60.0 ± SD 14.6 years) and obese patients (median 60.0 ± SD 13.2 years) ($p \leq 0.001$). Extremely obese patients (median 58 ± SD 12.5) did not statistically differ from any other group regarding age (Table 1 and Table 3). The number of women was statically higher in the normal weight ($66.1\%$) and in the extremely obese groups ($62\%$) than in the overweight and obese groups (respectively, $40.0\%$ and $45.7\%$, $p \leq 0.001$ and $$p \leq 0.013$$) (Table 2 and Table 4). Extracapsular spread, tumor location and ulceration did not statistically differ between the BMI groups (Table 2). BMI groups differed regarding the percentage of patients with metastases in the sentinel lymph node biopsy, which increased with the BMI. A statistically significant difference between the groups could only be observed between the extremely obese BMI group ($52.2\%$) and the normal weight group ($31.8\%$) ($$p \leq 0.001$$) (Table 2 and Figure 1b). ## 3.3. Metastases in the Sentinel Lymph Node Biopsy Our analysis showed that the higher the BMI (OR 1.04, CI 1.01–1.06), age (OR 1.01, CI 1.00–1.02), Breslow tumor thickness (OR 1.11, CI 1.07–1.14), and S100 value (OR 1.45, CI 0.39–5.45), the higher the likelihood of the patient exhibiting metastases in the sentinel lymph node biopsy (respectively, $$p \leq 0.005$$, $$p \leq 0.057$$, $p \leq 0.001$, $$p \leq 0.0105$$, Table 5). Furthermore, extremely obese patients (BMI >= 35) were almost twice more likely (OR 1.91, CI 1.23–2.95) to have a metastatic sentinel lymph node than normal-weight patients ($$p \leq 0.004$$, Table 6). In the sub-analysis of the patients with Breslow tumor thickness under 1 mm ($$n = 122$$), extremely obese patients (BMI >= 35) were four times more likely (OR 4.27, CI 0.78–23.31) to have a metastatic sentinel lymph node than normal-weight patients ($$p \leq 0.100$$, Table 8). In the initial multivariate analysis, the chance of having a positive sentinel lymph node in our population was higher being a male (OR 0.79, CI 0.58–1.08, $$p \leq 0.142$$), being overweight or extremely obese (OR 1.20, CI 0.85–1.70, $$p \leq 0.299$$ and OR 2.10, CI 1.30–3.65, $$p \leq 0.009$$). Melanomas on trunk or lower extremity were more likely to have metastases in the sentinel lymph node biopsy (OR 1.52, CI 1.04–2.23, $$p \leq 0.032$$ and OR 1.36, CI 0.91–2.04, $$p \leq 0.131$$). Ulceration and Breslow tumor thickness were also associated with metastases in the sentinel lymph node biopsy (OR 1.76, CI 1.16–2.67, $$p \leq 0.008$$ and OR 1.18, CI 1.00–1.40, $$p \leq 0.050$$). In the stepwise selection, where only characteristics with a significant influence on sentinel lymph node positivity ($p \leq 0.05$) were taken into consideration, only an association between metastases in sentinel lymph node biopsy and BMI >= 35 (OR 1.99, CI 1.15–3.42, $$p \leq 0.013$$), Breslow tumor thickness (OR 1.79, CI 1.18–2.70, $$p \leq 0.006$$), and Ulceration (OR 1.18, CI 1.01–1.38, $$p \leq 0.040$$) could be shown. For the other characteristics, such as male gender, tumor location and overweight, the odds ratio was no longer significantly different from 1. ## 4. Discussion This study demonstrates a significantly higher tumor Breslow thickness in overweight and obese melanoma patients. Extreme obesity was identified as an independent risk factor for the presence of lymph-node metastases. In our study, only $34\%$ of the patients who received a sentinel lymph node biopsy between 2013 and 2018 had a normal body weight (BMI < 25). Most patients ($66\%$) were overweight, obese, or extremely obese (BMI >= 25). The prevalence of obesity and overweightness in the studied population was higher than in the general German population ($54\%$) [19]. These findings corroborate the results of other authors who show an increased risk of melanoma development in overweight and obese populations [3,4,6,7,8]. Furthermore, in our study, at time of diagnosis, overweight, obese, or extremely obese patients presented with higher tumor thicknesses than normal-weight patients. These results are in-line with findings in previous research that overweightness and obesity are associated with thicker melanomas [10,12,13]. Several theories have attempted to explain the relation between obesity and melanoma development and progression. The metabolic role of obesity in tumor growth has been well-documented for several tumor entities. Obesity and its chronic calory excess lead to abnormal levels of glycemia, insulin, cytokines, adipokines and steroid hormones. This, in turn, leads to a pro-inflammatory state and promotes tumor progression and angiogenesis [20]. The same mechanisms may also induce melanoma growth in overweight and obese patients [21,22]. In addition to the metabolic activity from adipose tissue, diet has been proposed to also influence tumor development. Preclinical studies have shown that a high-fat diet may induce melanoma progression [23] and, on the other hand, that caloric restriction may slow down melanoma growth [21]. In a clinical study from Norway with 50,752 participants, a diet rich in omega-3 fatty acids and in polyunsaturated fat was associated with an increased melanoma risk in women [24]. Some authors have searched for a genetic mutation conferring simultaneous susceptibility to melanoma and obesity [25]. Cauci et al. focused on polymorphisms from a vitamin-D receptor [26]. Li et al. studied single nucleotide polymorphisms in the FTO, MAP2K5, NEGR1, FLJ35779, ETV5, CADM2, and NUDT3 genes [27]. These hypotheses remain to be confirmed. The literature also refers to the relation between tumor progression and late tumor detection due to hidden tumor location. This theory was refuted by Skowron et al., who showed that melanomas in overweight or obese patients are not more frequent in non-visible body areas than in normal-weight patients [12]. Late tumor detection and melanoma progression due to avoidance of doctor appointments possibly because of lower self-esteem in overweight and obese populations was studied by Risica et al. The authors showed that the method of melanoma detection, through self-examination or doctor appointments, does not seem to differ with BMI [28]. In our results, although Breslow tumor thickness was higher in all overweight and obese patients (with a BMI >= 25), Breslow tumor thickness did not linearly increase with BMI, meaning it did not differ within overweight, obese, and extremely obese patients. Therefore, our study shows that there is a higher risk of melanoma progression in patients with a BMI >= 25, but this risk does not continue to increase from BMI >= 25 and is the same for overweight, obese, and extremely obese patients. Ulceration and S100 value do not seem to be influenced by overweight or obesity, since they did not differ within BMI groups in our study. These findings are in-line with other groups [12,15]. A higher rate of non-detection of the sentinel lymph node despite lymphoscintigraphy has been described for breast cancer in obese patients [29,30]. Similar studies regarding sentinel lymph node biopsy in melanoma could not be found. In our study, the non-detection of SLN was not documented. There is little research on the role of obesity in the development of melanoma metastases. To our knowledge, only Shreckengost et al. analyzed the impact of obesity on sentinel lymph node metastases [18]. Contrary to their results, our study showed a significant influence of BMI >= 35 on sentinel lymph node metastases, one of the main predictors of melanoma prognosis. However, our research differs from Shreckengost et al. ’s study with regards to the study design. Patients with melanoma in stage IA were included in Shreckengost et al. ’s research, while they were not included in ours. Moreover, in $43.1\%$ of patients, SLN data was missing in Shreckengost et al. ’s study. Thus, overall metastases in the sentinel lymph node biopsy in the aforementioned study was only found in $15.1\%$ of the patients in contrast to our population, where $37.9\%$ of the patients exhibited metastases in the sentinel lymph node biopsy. Furthermore, patient groups in Shreckengost et al. ’s study and in this study differed as we distinguished patients with a BMI >= 35. Moreover, in our study, extremely obese patients (BMI >= 35) were almost twice more likely (OR 1.99) to have metastases in the sentinel lymph node than normal-weight patients. Extracapsular spread, however, did not differ within both groups. In the initial multivariate analysis, male gender, tumor location on body trunk or lower extremity as well as advanced age seemed to be predictors of metastases in the sentinel lymph node. After the stepwise selection, only BMI >= 35, Breslow tumor thickness, and ulceration were independent predictors of metastases in sentinel lymph node in our population. Since the last two tumor characteristics are the most important and validated features of melanoma prognosis and, therefore, of melanoma survival [11], our findings regarding BMI >= 35 and metastases in the sentinel lymph node may be valid. BMI is a standardized and inexpensive method to assess obesity, but it does not consider muscle mass or differentiate visceral from subcutaneous fat [31]. Other assessment methods of obesity such as hip circumference, waist-to-height, and waist-to-hip ratio also fail to measure visceral fat [32]. Among others, air displacement plethysmography [33], dual-energy X-ray and magnetic resonance imaging [34] are very accurate assessments of visceral fat but are undoubtedly more expensive and difficult to reproduce in such a big population study as ours. Despite its limitations, the authors consider that patients with a BMI >= 35 unquestionably suffer from morbid obesity. The mechanisms leading to the higher frequency of metastases presence in patients with BMI >= 35 are not yet identified. The main differentiating feature of the extremely obese patients’ group is that of having the highest amount of adipose tissue in comparison to all other BMI groups. Adipose tissue, besides the aforementioned endocrine and metabolic activities, is known to cause a systemic immunological disfunction and a reduced response to cancer in obese patients [35,36]. Some authors have observed, contrary to expectation, that obese patients with metastatic melanoma seem to have a better therapeutic response than normal-weight patients, especially to immunotherapy [37,38,39]. This is referred to as the “obesity paradox” [35,37]. Wang et al. showed that the immunological function of T cells was altered in obese patients (BMI >= 30) and the expression of PD-1 was higher, making a better efficacy of autoimmune therapy possible [35]. Our findings raise at least two questions regarding the screening and treatment of overweight and obese patients: Firstly, if overweight and obese populations are more likely to develop melanomas and they present with higher tumor thicknesses, should skin-cancer screening be adjusted to patient’s weight? Skin-cancer screening in Germany is recommended from the age of 35 years and is paid for by the public medical insurance every 2 years [40]. An earlier and/or more frequent screening could lead to an earlier discovery of melanoma with lower tumor thickness and, consequently, to a better prognosis of these patients. Secondly, if extremely obese patients (BMI >= 35) are more likely to exhibit metastases in the sentinel lymph node (although tumor thickness does not differ from overweight or obese patients), should BMI >= 35 also be a criterion to offer sentinel-node surgery in patients with lower tumor thicknesses (<1.0 mm)? Sentinel lymph node surgery is offered to all patients with tumor Breslow thickness over 1.0 mm. An earlier sentinel lymph node surgery is also indicated in patients with tumor thicknesses >0.75 mm and <1.0 mm, who meet certain criteria such as ulceration, increased mitosis rate and age under 40 years [41]. An earlier SLN surgery in patients with BMI >= 35 could lead to an earlier detection of metastases and, therefore, improve the treatment and prognosis of these patients. Following this line of thought, we analyzed whether the same results could be found in patients with tumor thicknesses <1.0 mm by conducting the same statistical analysis of our study but restricted to these patients (Table 8 and Table 9). In this population group, a similar trend regarding metastases in the sentinel lymph node was identified. However, the differences did not reach statistical significance, possibly due to the small number of patients. A multicenter study with a higher number of patients may help explain whether extremely obese patients would benefit from these approaches, which may eventually lead to a necessity of adjustment of the melanoma treatment guidelines. Limitations of our study are the absence of head and neck melanomas, its retrospective nature, and its single-center approach. ## 5. Conclusions In conclusion, our study not only validated that BMI influences melanoma development but also affects two of the main melanoma predictors: Breslow tumor thickness and metastases in the sentinel lymph node. BMI >= 25 was associated with thicker melanomas at diagnosis and BMI >= 35 was associated with an almost twice higher likelihood of exhibiting metastases in the sentinel lymph node than normal patients. Therefore, patients with a BMI >= 25 and, in particular, patients with BMI >= 35 had a worse prognosis at the time of diagnosis. To our knowledge, this is the first study to show this association. Our study raises the question of whether the prognosis of obese patients can be optimized through earlier tumor detection or through sentinel-node surgery at lower tumor thicknesses. ## References 1. 3 Results 3.0 Overview of Incident Cancer Cases and Cancer Deaths Non-Hodgkin Lymphoma 3.6% Ovaries 3.1% Stomach 2.4% Kidney 2.4% Leukaemia 2.3% Bladder 2.0 % Oral Cavity and Pharynx 1.9% Cervix 1.9% Thyroid Gland 1.8% Vulva 1.4% Central Nervous System 1.3% Liver 1.2% Multiple Myeloma 1.2% Soft Tissue without Mesothelioma 0. (2017.0) 2. Rastrelli M., Tropea S., Rossi C.R., Alaibac M.. Melanoma: Epidemiology, risk factors, pathogenesis, diagnosis and classification. In Vivo (2014.0) 28 1005-1011. PMID: 25398793 3. Oh S.W., Yoon Y.S., Shin S.-A.. Effects of Excess Weight on Cancer Incidences Depending on Cancer Sites and Histologic Findings Among Men: Korea National Health Insurance Corporation Study. J. Clin. Oncol. (2005.0) 23 4742-4754. DOI: 10.1200/JCO.2005.11.726 4. Samanic C., Gridley G., Chow W.-H., Lubin J., Hoover R.N., Fraumeni J.F.. Obesity and cancer risk among white and black United States veterans. Cancer Causes Control (2004.0) 15 35-44. DOI: 10.1023/B:CACO.0000016573.79453.ba 5. Samanic C., Chow W.-H., Gridley G., Jarvholm B., Fraumeni J.F.. Relation of body mass index to cancer risk in 362,552 Swedish men. Cancer Causes Control. (2006.0) 17 901-909. DOI: 10.1007/s10552-006-0023-9 6. Dennis L.K., Lowe J.B., Lynch C.F., Alavanja M.C.. Cutaneous Melanoma and Obesity in the Agricultural Health Study. Ann. Epidemiol. (2008.0) 18 214-221. DOI: 10.1016/j.annepidem.2007.09.003 7. Gallus S., Naldi L., Martin L., Martinelli M.. Anthropometric measures and risk of cutaneous malignant melanoma: A case–control study from Italy. Melanoma Res. (2006.0) 16 83-87. DOI: 10.1097/01.cmr.0000194429.77643.76 8. Kirkpatrick C.S., White E., Lee J.A.. Case-Control Study of Malignant Melanoma in Washington State. II. Diet, alcohol, and obesity. Am. J. Epidemiol. (1994.0) 139 869-880. DOI: 10.1093/oxfordjournals.aje.a117093 9. Calle E.E., Rodriguez C., Walker-Thurmond K., Thun M.J.. Overweight, Obesity, and Mortality from Cancer in a Prospectively Studied Cohort of U.S. Adults. N. Engl. J. Med. (2003.0) 348 1625-1638. DOI: 10.1056/NEJMoa021423 10. Cassano N., Caccavale S., Vena G.A., Argenziano G.. Body Mass Index and Melanoma Prognosis. Dermatol. Pr. Concept. (2021.0) e2021106. DOI: 10.5826/dpc.1104a106 11. Balch C.M., Gershenwald J.E., Soong S.-J., Thompson J.F., Atkins M.B., Byrd D.R., Buzaid A.C., Cochran A.J., Coit D.G., Ding S.. Final Version of 2009 AJCC Melanoma Staging and Classification. J. Clin. Oncol. (2009.0) 27 6199-6206. DOI: 10.1200/JCO.2009.23.4799 12. Skowron F., Berard F., Balme B., Maucort-Boulch D.. Role of obesity on the thickness of primary cutaneous melanoma. J. Eur. Acad. Dermatol. Venereol. (2015.0) 29 262-269. DOI: 10.1111/jdv.12515 13. Gandini S., Montella M., Ayala F., Benedetto L., Rossi C.R., Vecchiato A., Corradin M.T., De Giorgi V., Queirolo P., Zannetti G.. Sun exposure and melanoma prognostic factors. Oncol. Lett. (2016.0) 11 2706-2714. DOI: 10.3892/ol.2016.4292 14. de Giorgi V., Gori A., Papi F., Grazzini M., Rossari S., Verdelli A., Crocetti E., Savarese I., Giorgi L., Massi D.. Excess body weight and increased Breslow thickness in melanoma patients. Eur. J. Cancer Prev. (2013.0) 22 480-485. DOI: 10.1097/CEJ.0b013e32835f3b5d 15. Stenehjem J., Veierød M., Nilsen L., Ghiasvand R., Johnsen B., Grimsrud T., Babigumira R., Støer N., Rees J., Robsahm T.. Anthropometric factors and Breslow thickness: Prospective data on 2570 cases of cutaneous melanoma in the population-based Janus Cohort. Br. J. Dermatol. (2018.0) 179 632-641. DOI: 10.1111/bjd.16825 16. von Schuckmann L.A., Smith D., Hughes M.C.B., Malt M., van der Pols J.C., Khosrotehrani K., Smithers B.M., Green A.C.. Associations of Statins and Diabetes with Diagnosis of Ulcerated Cutaneous Melanoma. J. Investig. Dermatol. (2017.0) 137 2599-2605. DOI: 10.1016/j.jid.2017.07.836 17. Newton-Bishop J.A., Davies J.R., Latheef F., Randerson-Moor J., Chan M., Gascoyne J., Waseem S., Haynes S., O’Donovan C., Bishop D.T.. 25-Hydroxyvitamin D. Int. J. Cancer (2015.0) 136 2890-2899. DOI: 10.1002/ijc.29334 18. Shreckengost C.S.H., Tariq M., Farley C.R., Zhang C., Delman K.A., Kudchadkar R.R., Lowe M.C.. The Impact of Obesity on Surgically Treated Locoregional Melanoma. Ann. Surg. Oncol. (2021.0) 28 6140-6151. DOI: 10.1245/s10434-021-09773-7 19. Übergewicht und Adipositas bei Erwachsenen. J. Health Monit. (2017.0) 2 21-28. DOI: 10.17886/RKI-GBE-2017-025 20. Hopkins B.D., Goncalves M.D., Cantley L.C.. Obesity and Cancer Mechanisms: Cancer Metabolism. J. Clin. Oncol. (2016.0) 34 4277-4283. DOI: 10.1200/JCO.2016.67.9712 21. Brandon E.L., Gu J.-W., Cantwell L., He Z., Wallace G., Hall J.E.. Obesity promotes melanoma tumor growth: Role of leptin. Cancer Biol. Ther. (2009.0) 8 1871-1879. DOI: 10.4161/cbt.8.19.9650 22. Karimi K., Lindgren T.H., Koch C.A., Brodell R.T.. Obesity as a risk factor for malignant melanoma and non-melanoma skin cancer. Rev. Endocr. Metab. Disord. (2016.0) 17 389-403. DOI: 10.1007/s11154-016-9393-9 23. Pandey V., Vijayakumar M.V., Ajay A.K., Malvi P., Bhat M.K.. Diet-induced obesity increases melanoma progression: Involvement of Cav-1 and FASN. Int. J. Cancer (2011.0) 130 497-508. DOI: 10.1002/ijc.26048 24. Veierød M.B., Thelle D.S., Laake P.. Diet and risk of cutaneous malignant melanoma: A prospective study of 50,757 Norwegian men and women. Int. J. Cancer (1997.0) 71 600-604. DOI: 10.1002/(SICI)1097-0215(19970516)71:4<600::AID-IJC15>3.0.CO;2-F 25. Clement E., Lazar I., Muller C., Nieto L.. Obesity and melanoma: Could fat be fueling malignancy?. Pigment Cell Melanoma Res. (2017.0) 30 294-306. DOI: 10.1111/pcmr.12584 26. Cauci S., Maione V., Buligan C., Linussio M., Serraino D., Stinco G.. BsmI (rs1544410) and FokI (rs2228570) vitamin D receptor polymorphisms, smoking, and body mass index as risk factors of cutaneous malignant melanoma in northeast Italy. Cancer Biol. Med. (2017.0) 14 302-318. DOI: 10.20892/j.issn.2095-3941.2017.0064 27. Li X., Liang L., Zhang M., Song F., Nan H., Wang L.-E., Wei Q., Lee J.E., Amos C.I., Qureshi A.A.. Obesity-related genetic variants, human pigmentation, and risk of melanoma. Hum. Genet. (2013.0) 132 793-801. DOI: 10.1007/s00439-013-1293-4 28. Risica P.M., Weinstock M.A., Rakowski W., Kirtania U., Martin R.A., Smith K.J.. Body Satisfaction: Effect on thorough Skin Self-Examination. Am. J. Prev. Med. (2008.0) 35 68-72. DOI: 10.1016/j.amepre.2008.03.017 29. Derossis A.M., Fey J.V., Cody H.S., Borgen P.I.. Obesity influences outcome of sentinel lymph node biopsy in early-stage breast cancer. J. Am. Coll. Surg. (2003.0) 197 896-901. DOI: 10.1016/j.jamcollsurg.2003.08.005 30. Chagpar A.B., Martin R.C., Scoggins C.R., Carlson D.J., Laidley A.L., El-Eid S.E., McGlothin T.Q., Noyes R.D., Ley P.B., McMasters K.M.. Factors predicting failure to identify a sentinel lymph node in breast cancer. Surgery (2005.0) 138 56-63. DOI: 10.1016/j.surg.2005.03.003 31. Bandera E.V., Fay S.H., Giovannucci E.. The use and interpretation of anthropometric measures in cancer epidemiology: A perspective from the World Cancer Research Fund international continuous update project. Int. J. Cancer (2016.0) 139 2391-2397. DOI: 10.1002/ijc.30248 32. van der Kooy K., Leenen R., Seidell J.C., Deurenberg P., Droop A., Bakker C.J.. Waist-hip ratio is a poor predictor of changes in visceral fat. Am. J. Clin. Nutr. (1993.0) 57 327-333. DOI: 10.1093/ajcn/57.3.327 33. Mccrory M.A., Gomez T.D., Bernauer E.M., Mol P.A.. Evaluation of a new air displacement plethysmograph for measuring human body composition. Med. Sci. Sports Exerc. (1995.0) 27 1686-1689. DOI: 10.1249/00005768-199512000-00016 34. Park Y.-W., Heymsfield S., Gallagher D.. Are dual-energy X-ray absorptiometry regional estimates associated with visceral adipose tissue mass?. Int. J. Obes. (2002.0) 26 978-983. DOI: 10.1038/sj.ijo.0801982 35. Wang Z., Aguilar E.G., Luna J.I., Dunai C., Khuat L.T., Le C.T., Mirsoian A., Minnar C.M., Stoffel K.M., Sturgill I.R.. Paradoxical effects of obesity on T cell function during tumor progression and PD-1 checkpoint blockade. Nat. Med. (2019.0) 25 141-151. DOI: 10.1038/s41591-018-0221-5 36. Kavazović I., Krapić M., Beumer-Chuwonpad A., Polić B., Wensveen T.T., Lemmermann N.A., van Gisbergen K.P., Wensveen F.M.. Hyperglycemia and Not Hyperinsulinemia Mediates Diabetes-Induced Memory CD8 T-Cell Dysfunction. Diabetes (2022.0) 71 706-721. DOI: 10.2337/db21-0209 37. McQuade J.L., Daniel C.R., Hess K.R., Mak C., Wang D.Y., Rai R.R., Park J.J., Haydu L.E., Spencer C., Wongchenko M.. Association of body-mass index and outcomes in patients with metastatic melanoma treated with targeted therapy, immunotherapy, or chemotherapy: A retrospective, multicohort analysis. Lancet Oncol. (2018.0) 19 310-322. DOI: 10.1016/S1470-2045(18)30078-0 38. Donnelly D., Bajaj S., Yu J., Hsu M., Balar A., Pavlick A., Weber J., Osman I., Zhong J.. The complex relationship between body mass index and response to immune checkpoint inhibition in metastatic melanoma patients. J. Immunother. Cancer (2019.0) 7 222. DOI: 10.1186/s40425-019-0699-5 39. Rutkowski P., Indini A., De Luca M., Merelli B., Mariuk-Jarema A., Teterycz P., Rogala P., Lugowska I., Cybulska-Stopa B., Labianca A.. Body mass index (BMI) and outcome of metastatic melanoma patients receiving targeted therapy and immunotherapy: A multicenter international retrospective study. J. Immunother. Cancer (2020.0) 8 e001117. DOI: 10.1136/jitc-2020-001117 40. Richtlinie des Gemeinsamen Bundesausschusses über die Früherkennung von Krebserkrankungen (Krebsfrüherkennungs-Richtlinie/KFE-RL) 41. Onkologie L.. S3-Leitlinie zur Diagnostik, Therapie und Nachsorge des Melanoms
--- title: Association of Serum Vaspin Concentration with Metabolic Disorders in Obese Individuals authors: - Łukasz Pilarski - Marta Pelczyńska - Anna Koperska - Agnieszka Seraszek-Jaros - Monika Szulińska - Paweł Bogdański journal: Biomolecules year: 2023 pmcid: PMC10046748 doi: 10.3390/biom13030508 license: CC BY 4.0 --- # Association of Serum Vaspin Concentration with Metabolic Disorders in Obese Individuals ## Abstract Vaspin, a molecule produced in visceral adipose tissue, seems to participate in the pathogenesis of metabolic disorders. The study aimed to determine the association of vaspin concentration with metabolic disorders in obese individuals. Forty obese patients and twenty normal-weight subjects underwent biochemical (fasting glucose, insulin, lipid profile, interleukin-6, hs-CRP, vaspin concentration), blood pressure, and anthropometric measurements. The HOMA-IR index was calculated. Serum vaspin concentrations in the obese group were significantly higher than in the control group (0.82 ± 0.62 vs. 0.43 ± 0.59; $p \leq 0.001$). Among the entire population, vaspin concentration was positively correlated with body weight, BMI, WHR, and the percentage and mass of adipose tissue. Positive correlations between vaspin concentration and triglyceride level, insulin concentration, and HOMA-IR value were found. Vaspin concentration was positively correlated with hs-CRP and IL-6 levels. In obese patients, positive correlations between vaspin concentration and the percentage of adipose tissue and hs-CRP level were demonstrated. Logistic regression analysis showed that increased BMI was the biggest factor stimulating vaspin concentrations (OR = 8.5; $95\%$ CI: 1.18–61.35; $$p \leq 0.0338$$). An elevated vaspin level may imply its compensatory role against metabolic disorders in obese patients. Thus, vaspin appears to be a useful diagnostic parameter for new therapeutic approaches in obesity-related complications. Nevertheless, due to the small sample size, further studies are needed to confirm our results. ## 1. Introduction Epidemiological studies have demonstrated that obesity is a constantly growing problem that has reached the status of a global epidemic. According to the World Health Organization (WHO), the number of obese people has tripled in the last 20 years. Nowadays, obesity is defined and characterized as a disease [1]. There are a number of causes leading to the development of obesity, including genetic and environmental factors. However, the main reasons for the increasing number of people with obesity are mainly poor eating habits and a sedentary lifestyle [2]. Obesity is a state of an excessive fat accumulation caused by a disruption of energy balance due to positive caloric intake, which boosts the risk of numerous diseases [3]. While genetic factors may be involved in the pathogenesis of obesity, environmental factors contribute mainly to the presence of obesity-related disorders [4]. Obesity is a main health risk factor. It is strongly associated with the development of different pathologies, such as insulin resistance (IR), which in turn play a fundamental role in the pathogenesis of obesity-related complications, including metabolic syndrome (MetS) components, i.e., type 2 diabetes (T2DM) and dyslipidemia. Moreover, obesity may escalate cardiovascular disease (CVD), low-grade inflammation, non-alcohol fatty liver disease (NAFLD), and some types of cancer (e.g., colorectal cancer) [4]. Finally, obesity-related cardiometabolic complications are associated with increased mortality [5]. Adipose tissue is a highly active endocrine organ that regulates energy homeostasis through the secretion of numerous bioactive molecules (adipokines) [6]. Altered adipose tissue function participates in metabolic disorders once fat accumulation has begun. An excess of adipose tissue causes adipocyte synthesis dysfunction, macrophage infiltration, and low-grade inflammation. Disorder in adipokine production accompanied by obesity leads to changes both in carbohydrate and lipid metabolism [1,2]. Although the link between obesity and disturbances in adipokine production have been recognized, further studies are needed to fully understand their mechanism of action and therapeutic value [4]. Visceral adipose tissue-derived Serpin (vaspin) was first described by Hida et al. as an insulin-sensitizing adipokine secreted from the white adipose tissues (WAT) of Otsuka Long–Evans Tokushima Fatty (OLETF) rats, an animal model for obesity and T2DM [7]. It was reported that both tissue expression and serum levels of vaspin paralleled the degree of obesity and IR. Administration of vaspin to obese mice fed a high-fat/high-sucrose diet improved both insulin sensitivity and glucose tolerance and normalized blood glucose. Vaspin administration resulted in the suppression of proinflammatory adipokines, such as tumor necrosis factor α (TNF-α), resistin, and leptin, while upregulating levels of adiponectin and glucose transporter type 4 (GLUT4) in the WAT of obese mice fed a high-fat/high-sucrose diet [7]. In humans, serum concentrations of vaspin range from 0.2 to 2.5 ng/mL [8]. Elevated serum vaspin levels in humans are correlated with body mass index (BMI) and IR, and low serum vaspin concentrations represent a risk factor for the progression of T2DM. A number of studies have confirmed higher serum vaspin concentrations in obese and T2DM patients [6,9,10]. The expression of vaspin is tissue-specific and the highest level is observed in WAT [6]. Physical activity, dietary modification, and medical therapy of obese or diabetes patients affect vaspin secretion [6]. A body of evidence has confirmed that serum vaspin concentrations decrease with weight reduction and lifestyle modifications in obese adults [8]. Due to its insulin-sensitizing functions during the hyperglycemic state and protective role in vasculature and adipose states, vaspin has been speculated as a beneficial adipokine and therapeutic candidate in metabolic disorders, thus providing the reasoning for undertaking this study [4]. This study aimed to determine the association of serum vaspin concentration with metabolic disorders in obese individuals. ## 2.1. Study Population The recruitment process was performed in the Hypertension and Metabolic Disorders Outpatient Clinic, Clinical Hospital of the Transfiguration of the Lord, Poznan, Poland. Of the recruited 60 participants, 40 were assigned to the study group of obese patients and 20 were assigned to the control group of normal-weight subjects. Both groups were comparable concerning age and sex. Participation in the study was voluntary. All participants received extensive information about the study and signed a written consent form. The study was approved by the Poznan University of Medical Science’s Bioethics Commission (approval no. $\frac{324}{14}$). The study was conducted in accordance with the Helsinki Declaration. The inclusion criteria for the study group were as follows: obesity defined by BMI equal to or greater than 30 kg/m2, age between 18–65 years, stable body mass during the last 4 weeks (±3 kg), and written consent to participate in the study. The control group consisted of healthy, normal-weight subjects with similar age and body mass stability restrictions. The exclusion criteria included: the presence of secondary obesity, type 2 diabetes, insufficiently controlled hypertension, clinically overt atherosclerotic disease, chronic kidney disease, clinically relevant liver dysfunction, acute or chronic clinically overt inflammation process, diagnosed neoplastic disease, alcohol abuse, or cigarette smoking. ## 2.2.1. Anthropometric Parameters Anthropometric measurements, including waist and hip circumference, were taken using a standard medical tape measure. Waist circumference was measured midway between the costal arch and upper iliac crest, and hip circumference was measured at the level of the greater trochanters. Body weight and height were measured using a RADWAG WPT $\frac{100}{200}$ OW electric scale. The measurements were performed in the morning after overnight fasting with patients dressed only in light clothes. Height, waist circumference, and hip circumference were all determined with an accuracy of 0.5 cm. BMI and WHR (waist-hip ratio) were calculated from the received data using the appropriate formulas, and body composition (i.e., body adipose tissue, lean body mass) was assessed using the bioimpedance method with Bioscan 920–2 device (Maltron International, Rayleigh, UK). Patients were advised to avoid consuming large amounts of fluid before the test and to discontinue intense physical exercise 12 h before measurement. ## 2.2.2. Blood Pressure Measurement Blood pressure measurements were taken 3 times in a seated position at 2 min intervals using an ESH-validated electronic sphygmomanometer (705IT, Omron Corporation, Kyoto, Japan), according to current European Society of Hypertension (ESH) and European Society of Cardiology (ESC) [11] recommendations. Average values of systolic (SBP) and diastolic (DBP) blood pressure were calculated from the three measurements. ## 2.2.3. Biochemical Parameters A 5 mL sample of blood was taken from every patient on an empty stomach (defined as 12 h after last meal), which was then centrifuged and frozen at a temperature of −80 °C. Serum vaspin concentration was assessed using the human visceral adipose-specific serine protease inhibitor (vaspin) ELISA Kit (QY-E02112; Qayee-Bio, Shanghai, China) according to the manufacturer’s guidelines. The plate coefficient of variation was less than $15\%$. The measurement of vaspin concentration was performed in duplicate and the mean concentration was given. Serum interleukin-6 (IL-6) concentration was measured using the human interleukin-6 (IL-6) ELISA Kit (QY-E04262; Qayee-Bio, Shanghai, China). The plate coefficient of variation was also less than $15\%$. The concentration of IL-6 was also measured twice and the mean concentration was taking into account. High sensitivity CRP was assessed using the human high-sensitivity C-reactive protein (hs-CRP) ELISA Kit (Shanghai Sunred Biological Technology Co., Ltd., Shanghai, China). The sensitivity of the hs-CRP test was 0.112 mg/L. Sample linear regression with the expected concentration gave a correlation coefficient R over 0.95. Other biochemical analyses were performed using standard commercial tests, including: fasting glucose, insulin, and lipid profile. HOMA-IR was calculated using the appropriate formula. ## 2.2.4. Statistical Analysis Statistical analyses were performed using the Statistica data analysis software system version 13, 2017 (TIBCO Software Inc., Tulsa, OK, USA). The results are presented as the average ± standard deviation (SD). Compatibility of the parameters’ distribution with normal distribution was checked using the Shapiro-Wilk test. Parameters compatible with normal distribution were compared using the Student’s t-test for independent samples. Parameters not compatible with normal distribution were compared using the Mann-Whitney U test. Correlations between variables obeying normal distribution were assessed using the Pearson’s linear correlation coefficient. Spearman’s rank correlation coefficient was used for the remaining variables. Correlations were adjusted for age and for age and BMI. Logistic regression analysis was used to investigate if vaspin concentration was associated with elevated or lower risk of obesity. Logistic regression was performed using MedCalc® Statistical Software version 20.027 (MedCalc Software Ltd., Ostend, Belgium). All differences were considered statistically significant at a level of $p \leq 0.05.$ The Bonferroni-Hochberg correction was applied. The sample size was determined according to the vaspin concentration, based on the study by Tarabeih et al. [ 12]. The mean vaspin vespin concentrations in the study (LBP-Duration) and control groups were 6.11 ± 0.074 and 5.83 ± 0.044 pg/mL, respectively. It was calculated that a sample size of at least 10 subjects in each group would yield at least $80\%$ power in detecting a significant difference. ## 3. Results A total of 40 people aged 43.3 ± 13.4 years were enrolled into the study group while 20 healthy participants aged 38.9 ± 14.7 years were recruited to the control group. The patients from the obese group had significantly higher body weights (by $35.9\%$, $p \leq 0.001$), BMI values (by $44.1\%$, $p \leq 0.001$), WHR (by $17.2\%$, $$p \leq 0.001$$), and percentages of body adipose tissue ($43.3\%$, $p \leq 0.001$) than the subjects from the control group (Table 1). Among the biochemical parameters, differences occurred in the concentration of triglycerides (by $113.3\%$, $p \leq 0.001$), hs-CRP ($68.7\%$, $p \leq 0.001$), and IL-6 ($53.4\%$, $p \leq 0.001$), which were higher in the obese group. A similar situation occurred with IR markers, with higher insulin levels ($52.9\%$, $$p \leq 0.004$$) and HOMA-IR values ($78.0\%$, $p \leq 0.001$). In contrast, levels of HDL cholesterol were significantly lower in the study group than in the control group (by $32.3\%$, $p \leq 0.001$). There were no differences in reference to total cholesterol and LDL levels, glucose concentration, and blood pressure values between groups. The serum vaspin concentrations in the obese group were significantly higher than in the control group (0.82 ± 0.62 vs. 0.43 ± 0.59; $p \leq 0.001$). Females and males did not differ in vaspin concentration. The detailed characteristics of the study population are shown in Table 1. In the entire population, vaspin concentration was positively correlated with body weight ($r = 0.452$; $$p \leq 0.003$$), BMI ($r = 0.558$; $p \leq 0.001$, Figure 1), WHR ($r = 0.447$; $$p \leq 0.003$$, Table 2), and the content of body adipose tissue (both percentage and mass; $r = 0.616$ and $r = 0.507$, respectively; $p \leq 0.001$, Figure 2). For the lipid metabolism parameters, a statistically significant positive correlation was found between vaspin concentration and triglyceride level ($r = 0.337$; $$p \leq 0.013$$). In contrast, no significant associations were found between vaspin concentration and total cholesterol, HDL, or LDL concentrations. We found a positive correlation between vaspin concentration and IR markers such as insulin level ($r = 0.341$; $$p \leq 0.013$$) as well as HOMA-IR value ($r = 0.382$; $$p \leq 0.022$$), with no associations with fasting glucose level. Moreover, vaspin concentration was positively correlated with hs-CRP ($r = 0.614$; $p \leq 0.001$) and IL-6 levels ($r = 0.457$; $$p \leq 0.003$$, Table 2). After adjusting for age, vaspin concentration was still correlated with BMI and WHR values, the amount of body adipose tissue, as well as insulin and hs-CRP concentrations. Meanwhile, after adjusting for age and BMI, vaspin concentration was only correlated with hs-CRP level (Table 2). In obese patients, positive correlations between vaspin concentration and the percentage of body adipose tissue ($r = 0.382$; $$p \leq 0.030$$) and hs-CRP level ($r = 0.428$; $$p \leq 0.002$$, Table 3) were demonstrated. No statistically significant relationships were found with the control group. The results of logistic regression showed that a one unit increase in vaspin concentration increased the risk of obesity (based on BMI) 8.5 times (OR = 8.5; $95\%$ CI: 1.18–61.35; $$p \leq 0.0338$$; Table 4). The statistical significance of this association was maintained after adjusting the model for age (OR = 8.33; $95\%$ CI: 1.15–60.21; $$p \leq 0.0338$$). ## 4. Discussion The present study documented higher vaspin concentrations in obese individuals than in normal-weight subjects. This study also showed that BMI value was an independent determinant of serum vaspin concentration. Moreover, several positive correlations between serum vaspin concentration and cardiometabolic risk-related parameters were observed in obese patients and the entire studied population. The research data confirmed the relationship between vaspin concentration and obesity. Similarly to our study, Taheri et al. demonstrated higher serum vaspin levels in normal-weight obese patients than in non-obese controls [13]. Yang et al., in a study involving 66 patients with type 2 diabetes (including 36 obese subjects) and 48 patients without diabetes (including 21 obese subjects), found higher vaspin concentrations in obese patients, both in the group with diabetes and that with normal carbohydrate metabolism [14]. Additionally, a meta-analysis of six studies evaluated the significantly higher serum vaspin levels in obese ($$n = 1826$$, vaspin level higher by 0.52 ng/mL, $95\%$ confidence interval [CI]: 0.10–0.93, $$p \leq 0.02$$) and T2DM patients ($$n = 1570$$, vaspin level higher by 0.36 ng/mL, $95\%$ CI: 0.23–0.49, $p \leq 0.00001$) that in non-obese healthy controls [6]. On the other hand, not all analyses have confirmed these reports. In a study conducted by Auguet et al. involving 71 women, including 40 morbidly obese (BMI > 40 kg/m2) and 31 normal body-weight (BMI < 25 kg/m2) patients, no significant difference in serum vaspin concentration was found. However, the authors reported significantly higher mRNA expression of this adipokine in visceral (VAT) and subcutaneous adipose tissue (SAT) in women with obesity [15]. It is worth mentioning that in contrast to our study, previous research [9,16,17] has shown sexual dysmorphism in relation to vaspin concentrations. Some authors attributed the increased vaspin level in females to high estrogen concentrations [18]. However, several studies found no gender differences regarding vaspin concentration [19,20,21]. It has been also indicated that sex differences are abrogated in type 2 diabetic patients [9]. Thus, the absence of sexual dimorphism regarding the discussed molecule may be associated with metabolic disturbances, such as impaired glucose tolerance or insulin sensitivity, which may explain the lack of gender difference in our study. As previously mentioned, vaspin is expressed mainly in adipose tissue (visceral and subcutaneous) in humans, although its expression is also noted in many other tissues, such as the liver, pancreas, skin, placenta, stomach, cerebrospinal fluid, hypothalamus, and ovaries [22]. Its level depends on different genetic and environmental factors, among which the most important seem to be body weight and the content of adipose tissue. Both vaspin mRNA and serum levels are associated with obesity and impaired insulin sensitivity. The level of vespin mRNA in adipose tissue is correlated with body mass increase [23] and is elevated in patients with obesity and T2DM [9,10]. Other factors that increase the expression of this adipokine in VAT and serum are insulin and leptin levels, the presence of IR, as well as exposure to a high-fat diet [24]. Gonzales et al. documented that age, gender, and nutritional status may also affect vaspin expression in WAT [25]. It is documented that vaspin may have positive effects on glucose and insulin metabolism, lipid profile, appetite control, and arteriosclerosis, thus counteracting obesity, IR, and inflammation [26]. In the already mentioned study conducted by Hida et al., it was shown that vaspin was poorly detectable in young OLETF rats and concentrations increased with age, weight, and insulin level, peaking at 30 weeks, when rats achieved the highest body weight and IR. On the other hand, the authors observed a reduction in vaspin concentration together with aggravation of T2DM and body weight loss in OLETF rats [7]. Other studies reported that central and peripheral vaspin administration led to a reduction in food intake [27] and increased insulin sensitivity in both db/db and C57BL6 mice [28]. Conversely, vaspin knockout mice consuming a high-fat/high-sucrose diet showed increased body weight, macrophage infiltration in adipose tissue, lipid accumulation in the liver, and aggravation of insulin sensitivity [29]. Thus, an elevated vaspin level in subjects with excessive body weight may imply the compensatory role of vaspin in obesity and metabolic dysfunction. The potential anorexigenic effects of vaspin may result from inhibition of a protease degrading a putative anti-orexigenic factor [30]. Moreover, it was reported that administration of vaspin into the arcuate nucleus (ARC) of the rat hypothalamus reduced hunger and food intake through the significant decrease in neuropeptide Y (NPY) and increase in proopiomelanocortin gene expression [31]. In turn, vaspin-induced antidiabetic effects include the promotion of islet cell secretion in the pancreas, protection of β cells from damage mediated by nuclear factor-kappa B (NF-κB), as well as reduction of hepatic glucose production [32]. In our entire study population, vaspin concentration was positively correlated with cardiometabolic risk-related parameters, such as body weight, BMI, and percentage of adipose tissue. Those correlations (besides body weight) were still significant after adjusting for age. On the other hand, due to intercorrelations between the obesity-related variables, after adjusting for age and BMI, the correlations between anthropometric and biochemical parameters and vaspin level were not statistically significant. Similarly to our results, Yang et al. showed positive relationships between serum vaspin concentration and BMI, fat percentage, and triglyceride level [14]. In another study, strong positive correlations between vaspin concentration and BMI and waist circumference were demonstrated in patients with sleep apnea [33]. On the other hand, Alizadeh et al. did not confirm this relationship, showing no correlations between serum vaspin level and body composition components (BMI, fat percentage) [34]. It is worthwhile to note that BMI increase was the biggest factor stimulating vaspin concentrations in obese individuals in the current study, showing that anthropometric measurements are useful tools in the diagnosis of excessive body weight. The present study showed positive correlations between vaspin concentrations and IR marker indicators such as fasting insulin and HOMA-IR in the entire study population. Studies on the relationship between vaspin level and carbohydrate metabolism parameters provide various information. In line with our study, Suleymanoglu et al. showed that vaspin level was positively correlated with fasting insulin level and HOMA-IR value [20]. Additionally, other authors demonstrated significant associations between vaspin concentration and fasting plasma insulin level as well as the insulin sensitivity index. Moreover, it has been indicated that the latter is an independent factor affecting the level of the discussed adipokine [14]. The positive correlations between vaspin level and insulin concentration, HOMA-IR index, and fasting glucose level were also confirmed in the study by Aliasghari et al. on a group of 83 patients with nonalcoholic fatty liver disease (NAFLD) [35]. On the contrary, not all studies have confirmed these results [17,36]. Interesting results were provided in a study by Wada et al. These authors showed a decrease in serum serum vaspin concentrations in OLETF rats together with the deterioration of diabetes control. The use of pioglitazone caused an increase in the concentration of this adipokine, and the administration of recombinant vaspin to OLETF rats improved their glucose tolerance and insulin sensitivity [37]. The exact effect of vaspin’s action on insulin signalling and metabolism in humans is not fully understood. It is indicated that vaspin may influence glycemic control by kallikrein 7, a human protease that cleaves insulin in vitro. Vaspin, through its serpin activity, seems to inhibit kallikrein 7, thus the antidiabetic effect of this adipokine may be related to decreased degradation of circulating insulin [28]. Moreover, vaspin may mediate insulin signalling via binding to glucose-related protein (GRP78), 7 KDa voltage-dependent anion channel. It was demonstrated that vaspin binds GRP78 in HepG2 cells. Additionally, the stimulation of H−4-II-E-C3 cells with recombinant vaspin led to activation of the protein kinase B (AKT) and the 5’AMP-activated protein kinase (AMPK) signalling pathways, which were disturbed by inhibition of GRP78 [29]. Recently, GRP78 has been shown to be highly expressed in adipose tissue both in humans and mice and levels increase with age, obesity, and diabetes. The overexpression of GRP78 seems to be related to hyperinsulinemia in adipocytes, thus the management of GRP78 expression may be a potential therapeutic target against disturbed carbohydrate metabolism [38]. Our study indicated a positive correlation between serum vaspin concentration and hs-CRP level both in the entire population and the study group. This correlation was still significant even after adjustment for age as well as age and BMI. Moreover, a positive correlation between vaspin concentration and IL-6 level was demonstrated in the whole study population. A significant positive relationship between the concentration of the discussed adipokine and hs-CRP level was also documented by other authors in children with metabolic syndrome [10] and women with polycystic ovary syndrome [39]. An anti-inflammatory effect of vaspin has been suggested, probably due to its ability to inhibit the expression of proinflammatory adhesion markers, such as intercellular adhesion molecule 1 (ICAM1), vascular cell adhesion molecule 1 (VCAM1), E-selectin, and monocyte chemoattractant protein 1 (MCP-1) [40]. Vaspin attenuates the expression of adhesion molecules in the AMPK-dependent pathway and inhibits the activity of NF-κB [41,42]. Vaspin reduces the formation of reactive oxygen species (ROS) and apoptosis induced by oxidative stress of mesenchymal stem cells (MSCs) [43,44]. Vaspin also increases the bioavailability of nitric oxide (NO) by increasing the expression of DDAH II (dimethylarginine dimethylaminohydrolase)–the enzyme responsible for the degradation of ADMA (asymmetric dimethylarginine), which has an inhibitory effect on endothelial NO synthase [45]. Furthermore, vaspin seems to modulate lipid metabolism. Its infusion decreased triglyceride and free fatty acid levels and promoted cholesterol efflux in macrophages due to the upregulation of ATP-binding cassette transporter A1 (ABCA1) [46,47]. The relationship between vaspin and lipid parameters was also documented in our study. It is important to note that our study has some limitations. The first limitation is the size of the study and control groups. To ensure a homogeneous study population, we applied numerous exclusion criteria. Moreover, the study included only Caucasians; therefore, the generalization of the results should be done with caution. Secondly, to assess body composition we used electrical bioimpedance together with anthropometric measurements instead of magnetic resonance imaging or the DEXA method. Nevertheless, all of the mentioned methods are widely accepted in clinical practice. Further, the assessment of IR was based on the value of the HOMA-IR index not the euglycemic clamp method due to its invasiveness and the duration of the measurement. In addition, this was an observational study and was thus unable to identify genetic variations between vaspin concentration and the analyzed metabolic parameters. Therefore, due to small sample size, the provided results should be considered as preliminary and further studies with a bigger sample size are needed to confirm our report as well as elucidate the molecular mechanisms underlying the relationship between vaspin level and cardiometabolic risk-related parameters. On the other hand, the key strengths of the study include a deep analysis of biochemical, physiologic, and anthropometric parameters and their association with metabolic disorders in obese individuals. Thus, our study provides insight into the role of vaspin in the pathogenesis of obesity-related complications. ## 5. Conclusions Obese individuals presented higher serum vaspin levels than normal-weight subjects. Several positive correlations between vaspin level and cardiometabolic risk-related parameters were found. Furthermore, BMI value turned out to be an independent determinant of serum vaspin concentration in obese individuals. Thus, an elevated vaspin level in subjects with excessive body weight may imply the compensatory role of the discussed adipokine against obesity and its complications. In conclusion, vaspin appears to be a useful diagnostic parameter for new therapeutic approaches in obesity-related disorders. Nevertheless, further research with a larger cohort is needed to confirm the results obtained in our study. ## References 1. Barazzoni R., Gortan Cappellari G., Ragni M., Nisoli E.. Insulin resistance in obesity: An overview of fundamental alterations. Eat. Weight Disord. (2018) 23 149-157. DOI: 10.1007/s40519-018-0481-6 2. Kojta I., Chacińska M., Błachnio-Zabielska A.. Obesity, bioactive lipids, and adipose tissue inflammation in insulin resistance. Nutrients (2020) 12. DOI: 10.3390/nu12051305 3. Pasco J.A., Holloway K.L., Dobbins A.G., Kotowicz M.A., Williams L.J., Brennan S.L.. Body mass index and measures of body fat for defining obesity and underweight: A cross-sectional, population-based study. BMC Obes. (2014) 1. DOI: 10.1186/2052-9538-1-9 4. Singh M., Benencia F.. Inflammatory processes in obesity: Focus on endothelial dysfunction and the role of adipokines as inflammatory mediators: We reviewed obesity-induced metabolic and immunological changes at the level of vasculature and emphasize on the importance of adipokines. Int. Rev. Immunol. (2019) 38 157-171. DOI: 10.1080/08830185.2019.1638921 5. Lin X., Li H.. Obesity: Epidemiology, Pathophysiology, and Therapeutics. Front. Endocrinol. (2021) 12 706978. DOI: 10.3389/fendo.2021.706978 6. Feng R., Li Y., Wang C., Luo C., Liu L., Chuo F., Li Q., Sun C.. Higher vaspin levels in subjects with obesity and type 2 diabetes mellitus: A meta-analysis. Diabetes Res. Clin. Pract. (2014) 106 88-94. DOI: 10.1016/j.diabres.2014.07.026 7. Hida K., Wada J., Eguchi J., Zhang H., Baba M., Seida A., Hashimoto I., Okada T., Yasuhara A., Nakatsuka A.. Visceral adipose tissue-derived serine protease inhibitor: A unique insulin-sensitizing adipocytokine in obesity. Proc. Natl. Acad. Sci. USA (2005) 102 10610-10615. DOI: 10.1073/pnas.0504703102 8. Teshigawara S., Wada J., Hida K., Nakatsuka A., Eguchi J., Murakami K., Kanzaki M., Inoue K., Terami T., Katayama A.. Serum Vaspin Concentrations Are Closely Related to Insulin Resistance, and rs77060950 at SERPINA12 Genetically Defines Distinct Group with Higher Serum Levels in Japanese Population. J. Clin. Endocrinol. Metabol. (2012) 97 E1202-E1207. DOI: 10.1210/jc.2011-3297 9. Youn B.S., Kloting N., Kratzsch J., Lee N., Park J.W., Song E.S., Ruschke K., Oberbach A., Fasshauer M., Stumvoll M.. Serum vaspin concentrations in human obesity and type 2 diabetes. Diabetes (2008) 57 372-377. DOI: 10.2337/db07-1045 10. Buyukinan M., Atar M., Can U., Pirgon O., Guzelant A., Deniz I.. The association between serum vaspin and omentin-1 levels in obese children with metabolic syndrome. Metab. Syndr. Relat. Disord. (2018) 16 76-81. DOI: 10.1089/met.2017.0133 11. Williams B., Mancia G., Spiering W., Agabiti Rosei E., Azizi M., Burnier M., Clement D.L., Coca A., de Simone G., Dominiczak A.. 2018 ESC/ESH Guidelines for the Management of Arterial Hypertension: The Task Force for the Management of Arterial Hypertension of the European Society of Cardiology and the European Society of Hypertension. J. Hypertens. (2018) 36 1953-2041. DOI: 10.1097/HJH.0000000000001940 12. Tarabeih N., Kalinkovich A., Shalata A., Livshits G.. Circulating Levels of Visceral Adipose Tissue-Derived Serine Protease Inhibitor (Vaspin) Appear as a Marker of Musculoskeletal Pain Disability. Diagnostics (2020) 10. DOI: 10.3390/diagnostics10100797 13. Taheri E., Hosseini S., Qorbani M., Mirmiran P.. Association of adipocytokines with lipid and glycemic profiles in women with normal weight obesity. BMC Endocr. Disord. (2020) 20. DOI: 10.1186/s12902-020-00648-8 14. Yang L., Chen S.J., Yuan G.Y., Wang D., Chen J.J.. Changes and clinical significance of serum vaspin levels in patients with type 2 diabetes. Genet. Mol. Res. (2015) 14 11356-11361. DOI: 10.4238/2015.September.25.2 15. Auguet T., Quintero Y., Riesco D., Morancho B., Terra X., Crescenti A., Broch M., Aguilar C., Olona M., Porras J.A.. New adipokines vaspin and omentin. Circulating levels and gene expression in adipose tissue from morbidly obese women. BMC Med. Genet. (2011) 12. DOI: 10.1186/1471-2350-12-60 16. Moradi S., Mirzaei K., Abdurahman A.A., Keshavarz S.A., Hossein-Nezhad A.. Mediatory effect of circulating vaspin on resting metabolic rate in obese individuals. Eur. J. Nutr. (2016) 55 1297-1305. DOI: 10.1007/s00394-015-0948-4 17. Von Loeffelholz C., Möhlig M., Arafat A.M., Isken F., Spranger J., Mai K., Randeva H.S., Pfeiffer A.F., Weickert M.O.. Circulating vaspin is unrelated to insulin sensitivity in a cohort of nondiabetic humans. Eur. J. Endocrinol. (2010) 162 507-513. DOI: 10.1530/EJE-09-0737 18. Tan B.K., Heutling D., Chen J., Farhatullah S., Adya R., Keay S.D., Kennedy C.R., Lehnert H., Randeva H.S.. Metformin decreases the adipokine vaspin in overweight women with polycystic ovary syndrome concomitant with improvement in insulin sensitivity and a decrease in insulin resistance. Diabetes (2008) 57 1501-1507. DOI: 10.2337/db08-0127 19. Chang M., Park H.S., Park C.Y., Song Y.S., Jang Y.J.. Association between serum vaspin concentrations and visceral adipose tissue in Korean subjects. Obesity (2010) 18 2105-2110. DOI: 10.1038/oby.2010.60 20. Suleymanoglu S., Tascilar E., Pirgon O., Tapan S., Meral C., Abaci A.. Vaspin and its correlation with insulin sensitivity indices in obese children. Diabetes Res. Clin. Pract. (2009) 84 325-328. DOI: 10.1016/j.diabres.2009.03.008 21. Handisurya A., Riedl M., Vila G., Maier C., Clodi M., Prikoszovich T., Ludvik B., Prager G., Luger A., Kautzky-Willer A.. Serum vaspin concentrations in relation to insulin sensitivity following RYGB-induced weight loss. Obes. Surg. (2010) 20 198-203. DOI: 10.1007/s11695-009-9882-y 22. Mlyczyńska E., Kieżun M., Kurowska P., Dawid M., Pich K., Respekta N., Daudon M., Rytelewska E., Dobrzyń K., Kamińska B.. New aspects of corpus luteum regulation in physiological and pathological conditions: Involvement of adipokines and neuropeptides. Cells (2022) 11. DOI: 10.3390/cells11060957 23. Klöting N., Berndt J., Kralisch S., Kovacs P., Fasshauer M., Schön M.R., Stumvoll M., Blüher M.. Vaspin gene expression in human adipose tissue: Association with obesity and type 2 diabetes. Biochem. Biophys. Res. Commun. (2006) 339 430-436. DOI: 10.1016/j.bbrc.2005.11.039 24. Kurowska P., Mlyczyńska E., Dawid M., Jurek M., Klimczyk D., Dupont J., Rak A.. Review: Vaspin (SERPINA12) expression and function in endocrine cells. Cells (2021) 10. DOI: 10.3390/cells10071710 25. Gonzalez C.R., Caminos J.E., Vazquez M.J., Garces M.F., Cepeda L.A., Angel A., González A.C., García-Rendueles M.E., Sangiao-Alvarellos S., López M.. Regulation of visceral adipose tissue-derived serine protease inhibitor by nutritional status, metformin, gender and pituitary factors in rat white adipose tissue. J. Physiol. (2009) 587 3741-3750. DOI: 10.1113/jphysiol.2009.172510 26. Jung H.N., Jung C.H.. The role of anti-inflammatory adipokines in cardiometabolic disorders: Moving beyond adiponectin. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222413529 27. Klöting N., Kovacs P., Kern M., Heiker J.T., Fasshauer M., Schön M.R., Stumvoll M., Beck-Sickinger A.G., Blüher M.. Central vaspin administration acutely reduces food intake and has sustained blood glucose-lowering effects. Diabetologia (2011) 5 1819-1823. DOI: 10.1007/s00125-011-2137-1 28. Heiker J.T., Klöting N., Kovacs P., Kuettner E.B., Sträter N., Schultz S., Kern M., Stumvoll M., Blüher M., Beck-Sickinger A.G.. Vaspin inhibits kallikrein 7 by serpin mechanism. Cell Mol. Life. Sci. (2013) 70 2569-2583. DOI: 10.1007/s00018-013-1258-8 29. Nakatsuka A., Wada J., Iseda I., Teshigawara S., Higashio K., Murakami K., Kanzaki M., Inoue K., Terami T., Katayama A.. Vaspin is an adipokine ameliorating ER stress in obesity as a ligand for cell-surface GRP78/MTJ-1 complex. Diabetes (2012) 61 2823-2832. DOI: 10.2337/db12-0232 30. Recinella L., Orlando G., Ferrante C., Chiavaroli A., Brunetti L., Leone S.. Adipokines: New potential therapeutic target for obesity and metabolic, rheumatic, and cardiovascular diseases. Front. Physiol. (2020) 11 578966. DOI: 10.3389/fphys.2020.578966 31. Brunetti L., Di Nisio C., Recinella L., Chiavaroli A., Leone S., Ferrante C., Orlando G., Vacca M.. Effects of vaspin, chemerin and omentin-1 on feeding behavior and hypothalamic peptide gene expression in the rat. Peptides (2011) 32 1866-1871. DOI: 10.1016/j.peptides.2011.08.003 32. Liu S., Li X., Wu Y., Duan R., Zhang J., Du F., Zhang Q., Li Y., Li N.. Effects of vaspin on pancreatic β cell secretion via PI3K/Akt and NF-κB signaling pathways. PLoS ONE (2017) 12. DOI: 10.1371/journal.pone.0189722 33. Pan Z., Zhuang X., Li X., Huang S., Zhang L., Lou F., Chen S., Ni Y.. Significance of vaspin in obstructive sleep apnea-hypopnea syndrome. Exp. Ther. Med. (2016) 11 841-845. DOI: 10.3892/etm.2016.2997 34. Alizadeh S., Mirzaei K., Mohammadi C., Keshavarz S.A., Maghbooli Z.. Circulating omentin-1 might be associated with metabolic health status in different phenotypes of body size. Arch. Endocrinol. Metab. (2017) 61 567-574. DOI: 10.1590/2359-3997000000269 35. Aliasghari F., Izadi A., Jabbari M., Imani B., Gargari B.P., Asjodi F., Ebrahimi S.. Are vaspin and omentin-1 related to insulin resistance, blood pressure and inflammation in NAFLD patients?. J. Med. Biochem. (2018) 37 470-475. DOI: 10.2478/jomb-2018-0006 36. Bilir B.E., Güldiken S., Tunçbilek N., Demir A.M., Polat A., Bilir B.. The effects of fat distribution and some adipokines on insulin resistance. Endokrynol. Pol. (2016) 67 277-2782. DOI: 10.5603/EP.a2016.0023 37. Wada J.. Vaspin: A novel serpin with insulin-sensitizing effects. Expert Opin. Investig. Drugs (2008) 17 327-333. DOI: 10.1517/13543784.17.3.327 38. Shin J., Toyoda S., Nishitani S., Fukuhara A., Kita S., Otsuki M., Shimomura I.. Possible Involvement of Adipose Tissue in Patients With Older Age, Obesity, and Diabetes With SARS-CoV-2 Infection (COVID-19) via GRP78 (BIP/HSPA5): Significance of Hyperinsulinemia Management in COVID-19. Diabetes (2021) 70 2745-2755. DOI: 10.2337/db20-1094 39. Cakal E., Ustun Y., Engin-Ustun Y., Ozkaya M., Kilinç M.. Serum vaspin and C-reactive protein levels in women with polycystic ovaries and polycystic ovary syndrome. Gynecol. Endocrinol. (2011) 27 491-495. DOI: 10.3109/09513590.2010.501874 40. Phalitakul S., Okada M., Hara Y., Yamawaki H.. Vaspin prevents TNF-a-induced intracellular adhesion molecule-1 via inhibiting reactive oxygen species dependent NF-kB and PKCh activation in cultured rat vascular smooth muscle cells. Pharmacol. Res. (2011) 64 493-500. DOI: 10.1016/j.phrs.2011.06.001 41. Jung C.H., Lee M.J., Kang Y.M., Lee Y.L., Yoon H.K., Kang S.W., Lee W.J., Park J.Y.. Vaspin inhibits cytokine-induced nuclear factor-kappa B activation and adhesion molecule expression via AMP-activated protein kinase activation in vascular endothelial cells. Cardiovasc. Diabetol. (2014) 13 41. DOI: 10.1186/1475-2840-13-41 42. Zieger K., Weiner J., Krause K., Schwarz M., Kohn M., Stumvoll M., Blüher M., Heiker J.T.. Vaspin suppresses cytokine induced inflammation in 3T3-L1 adipocytes via inhibition of NFκB pathway. Mol. Cell Endocrinol. (2018) 460 181-188. DOI: 10.1016/j.mce.2017.07.022 43. Qi D., Wang D., Zhang C., Tang X., He J., Zhao Y., Deng W., Deng X.. Vaspin protects against LPS-induced ARDS by inhibiting inflammation, apoptosis and reactive oxygen species generation in pulmonary endothelial cells via the Akt/GSK-3β pathway. Int. J. Mol. Med. (2017) 40 1803-1817. DOI: 10.3892/ijmm.2017.3176 44. Zhu X., Zhang L., Chen Y., Chen B., Huang H., Lv J., Hu S., Shen J.. Vaspin protects mouse mesenchymal stem cells from oxidative stress-induced apoptosis through the MAPK/p38 pathway. Mol. Cell Biochem. (2019) 462 107-114. DOI: 10.1007/s11010-019-03614-8 45. Jung C.H., Lee W.J., Hwang J.Y., Lee M.J., Seol S.M., Kim Y.M., Lee Y.L., Kim H.S., Kim M.S., Park J.Y.. Vaspin increases nitric oxide bioavailability through the reduction of asymmetric dimethylarginine in vascular endothelial cells. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0052346 46. Sato K., Shirai R., Yamaguchi M., Yamashita T., Shibata K., Okano T., Mori Y., Matsuyama T.A., Ishibashi-Ueda H., Hirano T.. Anti-Atherogenic Effects of Vaspin on Human Aortic Smooth Muscle Cell/Macrophage Responses and Hyper-lipidemic Mouse Plaque Phenotype. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19061732 47. Gao J.H., Zeng M.Y., Yu X.H., Zeng G.F., He L.H., Zheng X.L., Zhang D.W., Ouyang X.P., Tang C.K.. Visceral adipose tissue-derived serine protease inhibitor accelerates cholesterol efflux by up-regulating ABCA1 expression via the NF-κB/miR-33a pathway in THP-1 macropahge-derived foam cells. Biochem. Biophys. Res. Commun. (2018) 500 318-324. DOI: 10.1016/j.bbrc.2018.04.066
--- title: Remotely Programmable Deep Brain Stimulator Combined with an Invasive Blood Pressure Monitoring System for a Non-Tethered Rat Model in Hypertension Research authors: - Žilvinas Chomanskis - Vytautas Jonkus - Tadas Danielius - Tomas Paulauskas - Monika Orvydaitė - Kazimieras Melaika - Osvaldas Rukšėnas - Vaiva Hendrixson - Saulius Ročka journal: Brain Sciences year: 2023 pmcid: PMC10046755 doi: 10.3390/brainsci13030504 license: CC BY 4.0 --- # Remotely Programmable Deep Brain Stimulator Combined with an Invasive Blood Pressure Monitoring System for a Non-Tethered Rat Model in Hypertension Research ## Abstract The control circuits of blood pressure have a strong neural regulatory element important in the pathogenesis of essential drug-resistant hypertension. Targeting lower medullary neural control mechanisms of blood pressure by electrical stimulation could be beneficial, and therefore, a novel device is needed. This paper presents a remotely programmable deep brain stimulator with an invasive continuous blood pressure monitoring system in a non-tethered rat model. The device is designed for lower medullary deep brain stimulation research with minimal interference to a daily animal routine. Electrodes were implanted in the caudal ventrolateral medulla. Animal survivability, catheter patency rates, and device data drift were evaluated. Eight out of ten rats survived the surgery and testing period with no or mild temporary neurological compromise. The study revealed that carotid catheters filled with heparinized glycerol ensure better catheter patency rates and blood pressure transduction. There was no significant drift in the device’s pressure sensitivity during the experiment. To our knowledge, this is the first experimental study to show considerable animal survival after lower medullary implantation. Combining the ability to measure and monitor invasive blood pressure with a closed-loop brain pulse generator in a single device could be of potential value in future hemodynamic animal research. ## 1. Introduction Neuromodulation, in a broad sense, is various invasive and noninvasive techniques that utilize chemical agents, magnetic pulses, electric current, or light to modify the function of the nervous tissue [1,2,3,4,5,6,7,8,9,10]. Chemical stimulation, although targeting specific receptors, lacks a millisecond scale of precision that defines the normal activity of the nervous tissue [2]. Transcranial magnetic stimulation is a repetitive neuromodulatory technique that applies magnetic pulses noninvasively and is valuable therapy in resistant cases of major depression, anxiety, and phobias [3,4]. Optogenetics is the application of light to selectively target genetically engineered neurons that respond to various wavelengths of light [5]. Although a valuable and promising research tool that portends a new therapeutic era, optogenetics is currently not applied in clinical practice. Electrical stimulation is the primary technique of invasive neuromodulatory applications used in various human treatment protocols. Means of transferring electric current to the nervous tissue differ according to the stimulation site: deep brain stimulation (DBS), spinal cord stimulation, vagal nerve stimulation, peripheral nerve stimulation, and cortical stimulation [6,7,8,9,10]. Deep brain stimulation (DBS) was established as a modern, non-destructive neuromodulatory therapy for the treatment of patients suffering from Parkinson’s disease in the early 1990s [11]. DBS is a technique that stimulates deep-seated nuclei in the telencephalon, diencephalon, and mesencephalon. Since its introduction, this technique has become one of the most effective treatment options for patients with advanced Parkinson’s disease, essential tremor, or generalized dystonia [6]. With the current state-of-the-art DBS techniques, more selective and directed stimulation through the multiangled arrangement of contacts is possible [12]. Although a very effective therapy, the exact mechanism of action of DBS is still unsettled, and few suggested mechanisms are constantly debated in the literature, see the review by Chiken [13]. As a relatively new therapy, DBS clinical application boundaries are continually expanding [14,15,16]. Experimental DBS of various animal models is used to search for new brain targets for diseases already successfully treated by electrical stimulation [17]. In the same manner, animal DBS models are employed to explore the possibility of applying DBS in the treatment of diseases so far not established as amenable to treatment by electrical stimulation [18,19]. The most promising candidate diseases for treatment by DBS are primary arterial hypertension (PAH) and obesity, which have pathogeneses stemming from failed neural control [20,21,22,23]. Long-term PAH is a major risk factor for coronary heart disease, heart failure, or stroke, responsible for $13\%$ of deaths globally [24]. In 2015 high blood pressure affected approximately 1.13 billion people or $22\%$ of the global adult population [25,26]. Because the disease burden of PAH is becoming more severe at an accelerating pace, expanding treatment options would be of immense value. Contemporary PAH management options include lifestyle changes, pharmacological treatment, and non-pharmacological interventions. Non-pharmacological interventions mainly reserved for drug-resistant hypertension are carotid baroreceptor stimulation and percutaneous renal artery denervation [27,28]. Baroreceptor stimulation is applied to baroreceptors in the carotid sinus to inhibit the baroreflex and decrease the sympathetic tone reaching the heart and vasculature [27]. The percutaneous renal artery denervation procedure selectively ablates small sympathetic nerves around renal arteries to reduce salt and water retention [28]. Numerous scientific studies indirectly support the hypothetical potential of DBS to correct high blood pressure [27,28,29,30,31,32,33,34,35]. To provide a few examples, up to $44\%$ of PAH could be due to increased sympathetic nervous system tone, as is evidenced by studies assessing the concentration of catecholamines in plasma or norepinephrine spillover in urine [29]. Solid indications suggest that abnormal sympathetic renal innervation could be partly responsible for the pathogenesis of PAH [34]. The effectiveness of baroreflex stimulation and renal denervation procedures, both affecting nervous regulation of blood pressure circuitry, shows that the neurogenic component in the pathogenesis of hypertension could be a major factor determining disease severity and drug resistance [27,28,35]. Therefore, applying DBS in the treatment of patients suffering from drug-resistant hypertension could be beneficial. In non-survival experiments, applying electric current to the caudal ventrolateral medulla (CVLM) leads to profound hypotensive effects [36]. CVLM is a major relay center of the baroreflex arc that directly inhibits the rostral ventrolateral medulla, the main nucleus driving the sympathetic tone [37]. To our knowledge, long-term CVLM electrical stimulation in awake and freely moving rats hasn‘t been reported in the literature. Studies of brain stem stimulation have introduced the notion that long-term stimulation of the lower medullary region could lead to grave neurological compromise or even death [38,39]. Based on the facts mentioned above, our study aimed to develop and test in vivo, a remotely programmable brain stimulator combined with an invasive blood pressure monitoring system dedicated to DBS in the lower medullary region. In this paper, an emphasis on the survivability of animals, monitoring of neurological compromise, and patency rates of catheters was attempted. ## 2.1. Design of the Device The device consists of a wireless data transmission module and a measurement element (Figure 1). Wireless data transmission is based on the low-cost commercially available microchip (ESP8266 chip, Espressif Systems, Shanghai, China). A complete ESP8266 module with an integrated antenna, radio frequency (RF) circuit, and flash memory was used. Blood pressure measurement and brain stimulation were controlled by a microcontroller (LPC845, NXP Semiconductors). The microcontroller’s internal analog-to-digital converter (ADC) was used to record brain stimulation pulses in vivo. In contrast, the device’s external ADC and digital-to-analog converter (DAC) chips were used to measure blood pressure and generate brain stimulation pulses, respectively. The universal asynchronous receiver–transmitter (UART) interface was used for data and command exchange between the ESP8266 and LPC845. The power supply circuit generates 5 V and 3.3 V from a 3.7 V li-ion battery with a capacity of 2200 mAh. The blood pressure measurement circuit consists of a pressure sensor (BPS130, Bourns, Inc., Riverside, CA, USA) and external ADC (MCP3202, Microchip Technology Inc., Chandler, AZ, USA). A serial peripheral interface (SPI) was used for data exchange between LPC845 and the pressure measurement circuit. According to the sensor’s datasheet, the sensor’s zero drift and sensitivity drift should be minimal, and no recalibration process should be necessary for the usual physiological and pathological ranges of blood pressure [40]. The brain stimulation circuit is capable of generating an electric pulse of any arbitrary shape. The circuit consists of a voltage pulse generator and a voltage-to-current converter. Voltage pulse generation was performed by an LPC845 microcontroller and a DAC chip (MCP4921, Microchip Technology Inc., Chandler, AZ, USA). The voltage-to-current conversion was achieved by connecting brain tissue resistance to the negative feedback link of the operational amplifier (AD8542, Analog Devices Inc., Wilmington, MA, USA). This study used a charged-balanced, non-symmetrical pulse shape. The operating range and resolution of stimulation parameters and other important features of the device are summarized in Table 1. ## 2.2. Data Recording and Analysis The data was recorded and analyzed with a PC running a Linux-based operating system with 8 Gb of RAM and a processing power of 2.4 GHz. A custom software solution was built for data collection and visualization (see Figure 2). The programming language Python was used for communication with the sensor. This component was designed so that it could handle large amounts of data. ActiveMQ was used as a broker for message exchange. NodeRED was used to create dashboards for manual sensor control (recording period, stimulation wave, etc.). A sensor discovery and notification service are responsible for sending broadcast messages to all sensors and monitoring the health of the sensors. Grafana dashboards were used to visualize real-time and historical data, see Figure 3. Helper services such as data transfer and daily data transfer were written with Python. Since the amount of data is large enough to put pressure on the database (100 data points per second from one sensor), a historical database was used to reduce the load on real-time data. For in-depth analysis and simulations, historical data were used. A daily data transfer job was created for transferring data from real-time and historical databases. A helper service was also responsible for removing data from the real-time database. The data analysis component is responsible for pulse measurement, breathing rate measurement, and automatic sensor control. The pulse and breathing rates were measured by smoothing data using Gaussian kernels with different kernel bandwidths (20 for breathing rate extraction and 10 for pulse rate smoothing). The changes in blood pressure could also be monitored and stimulation parameters could be adjusted automatically in a closed-loop manner. ## 2.3. Device Peripherals The arterial catheters (Figure 4) were made of a polyethylene tube (Smiths Medical International Ltd., external diameter 0.96 mm, internal diameter 0.58 mm) with additional ports to replace lock solutions and sensor calibration. An anchor was fixed at 24–26 mm from the catheter tip and acted as a stopper and a securing device that positioned the catheter tip at the junction of the common carotid artery and aortic arch. Either 500 IU heparin/$99.7\%$ glycerol or 500 IU heparin/$50\%$ dextrose solutions were used for catheter lock solutions. Electrodes made of stainless-steel wire of 100 µm in diameter with polyamide insulation of 5 µm (Goodfellow Cambridge Ltd., Huntingdon, UK) were used. The electrodes were produced in a bipolar fashion with 200 µm overhang between the tips. The jackets for the rats were made from neoprene fabric with Velcro straps, and the device case was 3D printed (Ender-3 Pro, Creality) from polylactic acid plastic (Fiberology). ## 2.4. Procedure All study protocols and experiments followed current European Union animal research regulations. Experiments were approved by Lithuanian institutional authorities (State Food and Veterinary Service No. G2-128). The device’s performance was tested while experimenting on ten male Wistar rats aged 12–16 weeks, weighing 340–425 g, obtained from our in-house breeding colony (Vilnius University, Life Sciences Center, Lithuania). The rats were housed individually under controlled conditions with 12 h light/dark cycles, and drinks and food allowed ad libitum. Seven days before the implantation procedure, the animals were equipped with a jacket and a case simulating the device size to enable adaptation to restrained stress. Surgical interventions were performed under sevoflurane (Baxter) anesthesia (3–$4\%$ in a mixture of $100\%$ O2 administered through a nose cone). Catheter and electrode implantation procedures were performed in one session under a surgical microscope with ×6.4 magnification. Magnification allowed smaller surgical wound openings, better visualization and preservation of the vagal nerve and omohyoid muscle, as well as better control of electrode entrance. After the rats were anesthetized, their pain reflexes checked, and three applications of povidone-iodine solution (Egis Pharmaceuticals) given, two 1-cm-long median incisions in the vertex and suprasternal (anterior neck) regions were made. The device’s catheter filled with lock solution was inserted and tunneled subcutaneously between incisions through the left side of the neck. Caution was taken not to damage the left external jugular vein during the tunneling of the catheter. Approximately 0.5 cm of the left common carotid artery was bluntly dissected and exposed with two sutures proximally and distally. During dissection, the omohyoid muscle was left in a lateral position. Special care was taken while exposing the artery not to damage small vagal nerve branches that usually run over the carotid artery. The distal end of the artery was tied off, and a 2 mm longitudinal incision in the ventral wall of the carotid artery was made while tension was kept on the loose proximal suture. This technique allows almost bloodless catheter implantation and is the most challenging part of surgery to master. The catheter was inserted until the anchor reached the incision in the carotid artery, then the catheter was tied to the artery with two additional sutures. If the catheter’s intravascular part is too long, the catheter tip will occlude the aorta, and the animal will die. If the catheter is too short, it will become blocked by a clot and no pressure data will be gained. During the study, the observation was made that the length of the catheter’s intravascular part depended on the animal’s distance from the base of the tail to its nose. This could be a more accurate approximation than weighing the animal, as is usual practice in other laboratories. After implantation, the neck wound was closed with 5-0 sutures. After the placement of the catheter, animals were put in a stereotaxic apparatus (Narishige Scientific Instrument Lab). Arterial blood pressure was continuously recorded during the electrode placement procedure. The already made vertex incision was used for electrode implantation; thus, additional care should be taken while manipulating the electrode or high-speed drill not to damage the already inserted and tunneled catheter. Because survival surgery was attempted, an indirect transcerebellar route was chosen for the electrode implantation instead of direct open implantation where the cerebellum should be partially removed to expose the IV ventricle. The skull was positioned to be level between Bregma and Lambda. A trephination of approximately 3 mm in diameter was drilled on the occipital ridge region on the right side. Two additional holes were drilled in the skull for dental screws, anterior to the trephination and posterior to the lambdoid suture. The coordinates of CVLM were chosen in accordance with Paxinos and Goodchild and were the following: −4.40 mm anteroposteriorly, 2.10 mediolaterally, and 9.90 dorsoventrally from Lambda [41,42]. After advancing the electrode tip to 8.5 mm in the z-axis, a stimulation protocol was implemented. Blood pressure response to electrical stimulation was recorded at multiple sites while lowering the electrode in steps of 0.1 mm. The implantation site was verified by two consecutive stimulation trains of 3 s (100 μA, 50 Hz, 0.1 ms) that caused the arterial pressure to drop at least 20 mm Hg (Figure 5). At least 1 min was given before applying the stimulus at the next location. No attempt was made to lower the electrode below 10.5 mm in the z-axis. After the accurate placement of the electrode, a head post made of dental micro-hybrid composite (FlowX, ORBIS) was affixed. ## 2.5. Postoperative Period The postoperative and data collection period lasted continuously 24 h per day for two weeks. Close monitoring for abnormal postures and behaviors and administration of analgesia took place for 24 h after the procedure. Catheter patency was maintained by replacing lock solutions once per day with the known filling volume of the system. Blood pressure recordings were made with lock solutions in place, and no substitutions of heparinized saline were made. For lock solutions, either a heparinized $50\%$ dextrose solution or a heparinized $99.5\%$ glycerol solution was used. Five hundred IU of heparin per 1 mL of solution was added. Animals were closely monitored for neurological compromise by a staff veterinarian. Catheter patency status was determined as non-patent when there was evidence of loss of discernible blood pressure waves on the monitor. On the 8th day of the experiment, a chronic stimulation (50 μA, 50 Hz, 0.1 ms) was initiated that continued for the next seven days. The device pressure output (in mV) was calibrated to known-value pressure points of 0, 50, 100, 150, and 200 mm Hg using a custom pressure gauge kit. Calibrations were done immediately before implantation on the first day of the experiment, in the middle of the experiment on the 7th day, and at the end on the 14th day. ## 2.6. Data Analysis Data analysis was done with R and MS Excel. To assess the stability of the device measuring circuitry, the one-factor analysis of variance (ANOVA) was used to analyze the difference in calibrated blood pressure output (in mV) on the 1st, 7th, and 14th days of the experiment. Mean blood pressure was averaged by extracting data from the SQL database (PostgreSQL). The blood pressure data gained from the first four days were not used for the analysis as rats were recovering from surgery and blood pressure curves were unstable. Mean blood pressure was averaged on days 5–7 (“stimulation OFF” period) and days 8–14 (“stimulation ON” period). A dependent Student t-test was used to analyze the difference in mean blood pressure between the “Stimulation OFF” and “Stimulation ON” periods. Only data gained from rats that survived the whole experimentation period without neurological or behavioral compromise were used in the data analysis. Statistical significance was taken at $p \leq 0.05.$ Data were expressed as the mean ± standard deviation of the mean. ## 3.1. Survivability and Neurological Compromise Nine of the ten rats ($90\%$) used in the study survived the surgery and regained wakefulness. One rat ($10\%$) died during the procedure when the electrode had been advanced only 6 mm in the dorsoventral coordinate. The autopsy revealed an unexpectedly high-riding aortic arch and the catheter occluding the aorta. One rat regained wakefulness without evident neurological compromise and was hemodynamically stable, only to die eight hours later. The autopsy revealed a minor, 3 mm hemorrhagic lesion in the dorsal medulla in the path of the electrode. One rat had transient cerebellar symptoms: ataxic movements of the right side with no evident weakness that subsided in 48 h. The other seven rats ($70\%$) were free of any apparent neurological compromise during the entire postoperative period. ## 3.2. Blood Pressure Recording The mean blood pressure of rats during the “Stimulation OFF” period was higher (93.12 ± 5.33 mm Hg) than the mean blood pressure of rats during the “Stimulation ON” period (86.64 ± 6.47 mm Hg). A t-test for dependent samples showed that this difference was statistically significant, dF = 7, $$p \leq 0.004.$$ During the experiment, a clear distinction could be made between rats that had and had not responded to stimulation. A typical blood pressure pattern in animals that responded to stimulation is presented on the left side of Figure 6. Note the decrease in mean blood pressure when chronic stimulation on the 8th day was turned on. Note also, an acute rise in the mean blood pressure during the first three days after surgery, as this is related to the recovery and healing process postoperatively. Five out of eight rats ($62.5\%$) that survived two weeks showed a similar blood pressure pattern, while the three rats ($37.5\%$) did not respond to stimulation and had a blood pressure pattern similar to the one demonstrated on the right side of Figure 6. ## 3.3. Zero and Sensitivity Drift Device calibration data are illustrated in Figure 7. The ANOVA has shown that at all pressure points where the calibration of the device was performed, there is no significant difference between calibrated blood pressure output (in mV) and the day the calibration was done during the study, meaning that there is no significant device sensitivity drift. ## 3.4. Device Performance Recordings of brain stimulation pulses are illustrated in Figure 8. Plot (a) is a voltage recorded with an oscilloscope on a 1000-ohm resistor; the current can be calculated by dividing the voltage by the resistor value. Plots (b) and (c) were recorded by the ADC of the internal microcontroller, and current value was calculated as a voltage drop on the R7 resistor. Plot (c) was recorded during in vivo brain stimulation. Note the similarities of simulation pulses between three different methods of recording. ## 3.5. Catheter Patency Rates The mean duration of catheter patency filled with heparinized dextrose solution was 125.05 ± 22.54 min. The patency of the catheters filled with heparinized glycerol solution was stable for 24 h until daily catheter maintenance, and replacement of the lock solution was performed according to the study protocol. Non-patency of the glycerol-filled catheters was very rare. In all instances, it happened during the first day after the surgery when blood pressure perturbations were evident. As catheters filled with lock solutions outlasted the catheters filled with dextrose solutions by quite a margin, the application of statistical tests was not reasonable. ## 4. Discussion Although this is a preliminary/pilot study that seeks to test the device performance in vivo, the results of the CVLM stimulation are encouraging, as five rats responded to stimulation with a discernible hypotensive effect. This effect was less notable than in studies of acute, non-survival surgery designs [36,43]. It is known that anesthesia facilitates depressor pathways in the brain [44,45]. Although O’Callaghan et al. showed considerable hypotensive effect while stimulating periaqueductal gray matter (PAG) in anesthetized rats, they failed to show the same results when the rats were awake [45]. In the literature, we have not found any successful chronic CVLM stimulation studies that show the survivability of the animals after the implantation and, at the same time, demonstrate the effectiveness of stimulation. Studies involving stimulation or recording with lower medullary electrodes have usually been performed in a surgical manner that does not assure survival [46,47,48]. A few publications were found presenting research concerning the stimulation of the brain stem with electrodes in survival surgery, but none with the position of the electrodes in such a posterior location [49,50,51]. This could be due to a notion that lower medullary lead implantation could be dangerous and prone to severe neurological complications in the region of the ‘no man’s land of neurosurgery’ [38,39]. This is the first study to show considerable survivability of animals with stimulating electrodes implanted in the CVLM. CVLM DBS would be difficult to translate to human trials. The lowest point of the human brainstem axis where electrodes were inserted was the locus coeruleus region and parabrachial complex, both in the lower pons [52,53]. Those successful implantations were done in the pre-DBS era by courageous neurologists and neurosurgeons and are not possible nowadays due to a lack of reproducible findings and a drift to more superior targets that could be easier to reach without risking severe complications [54]. Second, the technology of DBS may not have advanced enough to allow such a delicate structure as the CVLM to be targeted. Clinical data exploring the DBS treatment effect on drug-resistant hypertension comes from a few studies targeting PAG [20,55,56,57]. From a physiological theoretical perspective, CVLM, although a difficult-to-reach target, has two inherent advantages over PAG. First, CVLM has no known direct neuroendocrine, nociceptive, or integrative behavioral function. Its primary function is to relay information from NTS to the RVLM in a reflex manner [45]. Second, being the lowest inhibitory center of the autonomic nervous system, CVLM should influence different pathogenetic circuitry of arterial hypertension. No matter the cause of hypertension, be it obesity-induced, salt-sensitive, or hypertension caused by sensitization of autonomic pathways, CVLM stimulation should be able to lower arterial blood pressure effectively [58]. An analysis of blood pressure calibration data did not show significant sensitivity drift at any calibration point during the experiment. This is in accordance with the datasheet for the BPS130 sensors [40]. The device is stable and can be used for further research. Direct blood pressure in laboratory animals can be recorded by an implantable radio telemetry transducer or an external transducer attached through a swivel to a catheter system [59]. Laboratories use many custom-made pulse generator systems worldwide [60,61,62]. This device combines both functions: wireless blood pressure telemetry and an electric pulse generator with the possibility of closed-loop stimulation. To our knowledge, this is the first device that combines blood pressure telemetry and a brain stimulator in one device that is sufficiently small to mount on the animal. Although appearing relatively robust, no animal demonstrated restrained behaviors (changes in postures, eating, or drinking habits) during the study period (Figure 4). It is a common practice that studies presenting custom stimulators fail to acknowledge the testing part of the created device or the testing being only rudimentarily performed with 1000-ohm resistors [60,61]. This device was tested in vivo to check whether cerebral biopotentials as high as 100 µV could interfere with introducing an electric current into neural tissue. Testing did not show that the device worked differently in vivo compared to conditions in which a resistor was used. ( Figure 8). Catheter patency rates showed results that were comparable to those demonstrated by Luo et al. [ 63]. In this research paper, the patency of the catheter was determined visually by assessing whether flushing of the system was still possible. In this study, the patency of the catheters was determined by evaluating whether discernible blood pressure waves were evident. In this regard, catheters filled with heparinized glycerol solutions were patent for up to 24 h when daily replacement of lock solutions was done following the study protocol. All blood pressure recordings were done while the catheter system was filled with lock solutions, acting not only to prevent clotting but as a transfer agent of pressure force to the sensor. In our study, blood pressure waves were comparable to high-fidelity blood pressure recordings in other studies [64]. Even without a normal pressure wave signal, determination of mean arterial blood pressure is possible as long as a discernible pressure wave is evident, indicating that catheter patency is maintained [59]. ## 5. Conclusions The device presented in this study is stable and provides reproducible measurements of blood pressure while at the same time stimulating targets in the medullary region. The equipment can be used as a novel solution for testing various targets that could be implicated in blood pressure control in animal models. More studies with groups, including sham controls, are needed to demonstrate the effect of CVLM stimulation on blood pressure. Future studies should also include not otherwise healthy Wistar rats, but instead rats prone to hypertension, such as spontaneously hypertensive rats. Furthermore, the effect of bilateral CVLM stimulation should also be tested as this could cause a reinforced effect on pressure change. Future studies could even include other targets in the lower medullary region, such as the rostral ventrolateral medulla. As a cardioaccelerator center, this particular target could be used to treat patients prone to vasovagal syncope. Further, new opportunities could be explored for treating conditions such as a vegetative state after severe traumatic brain injury or other brain tissue insult. ## References 1. Wang J., Chen Z.. Neuromodulation for Pain Management. Adv. Exp. Med. Biol. (2019.0) 1101 207-223. DOI: 10.1007/978-981-13-2050-7_8 2. Deer T.R., Pope J.E., Hanes M.C., McDowell G.C.. Intrathecal Therapy for Chronic Pain: A Review of Morphine and Ziconotide as Firstline Options. Pain Med. (2019.0) 20 784-798. DOI: 10.1093/pm/pny132 3. Burke M.J., Fried P.J., Pascual-Leone A.. Transcranial magnetic stimulation: Neurophysiological and clinical applications. Handb. Clin. Neurol. (2019.0) 163 73-92. DOI: 10.1016/B978-0-12-804281-6.00005-7 4. Borgomaneri S., Battaglia S., Garofalo S., Tortora F., Avenanti A., di Pellegrino G.. State-Dependent TMS over Prefrontal Cortex Disrupts Fear-Memory Reconsolidation and Prevents the Return of Fear. Curr. Biol. (2020.0) 30 3672-3679. DOI: 10.1016/j.cub.2020.06.091 5. Kalanithi P.S., Henderson J.M.. Optogenetic neuromodulation. Int. Rev. Neurobiol. (2012.0) 107 185-205. DOI: 10.1016/B978-0-12-404706-8.00010-3 6. Krauss J.K., Lipsman N., Aziz T., Boutet A., Brown P., Chang J.W., Davidson B., Grill W.M., Hariz M.I., Horn A.. Technology of deep brain stimulation: Current status and future directions. Nat. Rev. Neurol. (2021.0) 17 75-87. DOI: 10.1038/s41582-020-00426-z 7. Rock A.K., Truong H., Park Y.L., Pilitsis J.G.. Spinal Cord Stimulation. Neurosurg. Clin. N. Am. (2019.0) 30 169-194. DOI: 10.1016/j.nec.2018.12.003 8. González H.F.J., Yengo-Kahn A., Englot D.J.. Vagus Nerve Stimulation for the Treatment of Epilepsy. Neurosurg. Clin. N. Am. (2019.0) 30 219-230. DOI: 10.1016/j.nec.2018.12.005 9. Erosa S.C., Moheimani R.S., Oswald J.C., Castellanos J.P., Abraham M.E., Schuster N.M.. Peripheral Nerve Stimulation for Chronic Pain and Migraine: A Review. Phys. Med. Rehabil. Clin. N. Am. (2022.0) 33 379-407. DOI: 10.1016/j.pmr.2022.01.007 10. Tronnier V., Rasche D.. Epidural and subdural stimulation. Handb. Clin. Neurol. (2013.0) 116 343-351. DOI: 10.1016/B978-0-444-53497-2.00028-0 11. Limousin P., Pollak P., Benazzouz A., Hoffmann D., Le Bas J.F., Broussolle E., Perret J.E., Benabid A.L.. Effect of parkinsonian signs and symptoms of bilateral subthalamic nucleus stimulation. Lancet (1995.0) 345 91-95. DOI: 10.1016/S0140-6736(95)90062-4 12. Steigerwald F., Matthies C., Volkmann J.. Directional Deep Brain Stimulation. Neurotherapeutics (2019.0) 16 100-104. DOI: 10.1007/s13311-018-0667-7 13. Chiken S., Nambu A.. Mechanism of Deep Brain Stimulation: Inhibition, Excitation, or Disruption?. Neuroscientist (2016.0) 22 313-322. DOI: 10.1177/1073858415581986 14. Mallet L., Polosan M., Jaafari N., Baup N., Welter M.L., Fontaine D., du Montcel S.T., Yelnik J., Chéreau I., Arbus C.. Subthalamic nucleus stimulation in severe obsessive-compulsive disorder. N. Engl. J. Med. (2008.0) 359 2121-2134. DOI: 10.1056/NEJMoa0708514 15. Halpern C.H., Wolf J.A., Bale T.L., Stunkard A.J., Danish S.F., Grossman M., Jaggi J.L., Grady M.S., Baltuch G.H.. Deep brain stimulation in the treatment of obesity. J. Neurosurg. (2008.0) 109 625-634. DOI: 10.3171/JNS/2008/109/10/0625 16. Malone D.A., Dougherty D.D., Rezai A.R., Carpenter L.L., Friehs G.M., Eskandar E.N., Rauch S.L., Rasmussen S.A., Machado A.G., Kubu C.S.. Deep brain stimulation of the ventral capsule/ventral striatum for treatment-resistant depression. Biol. Psychiatry (2009.0) 65 267-275. DOI: 10.1016/j.biopsych.2008.08.029 17. Breit S., Lessmann L., Benazzouz A., Schulz J.B.. Unilateral lesion of the pedunculopontine nucleus induces hyperactivity in the subthalamic nucleus and substantia nigra in the rat. Eur. J. Neurosci. (2005.0) 22 2283-2294. DOI: 10.1111/j.1460-9568.2005.04402.x 18. Sani S., Jobe K., Smith A., Kordower J.H., Bakay R.A.. Deep brain stimulation for treatment of obesity in rats. J. Neurosurg. (2007.0) 107 809-813. DOI: 10.3171/JNS-07/10/0809 19. Oterdoom D.L.M., Lok R., van Beek A.P., den Dunnen W.F.A., Emous M., van Dijk J.M.C., van Dijk G.. Deep Brain Stimulation in the Nucleus Accumbens for Binge Eating Disorder: A Study in Rats. Obes. Surg. (2020.0) 30 4145-4148. DOI: 10.1007/s11695-020-04697-9 20. O’Callaghan E.L., Hart E.C., Sims-Williams H., Javed S., Burchell A.E., Papouchado M., Tank J., Heusser K., Jordan J., Menne J.. Chronic Deep Brain Stimulation Decreases Blood Pressure and Sympathetic Nerve Activity in a Drug- and Device-Resistant Hypertensive Patient. Hypertension (2017.0) 69 522-528. DOI: 10.1161/HYPERTENSIONAHA.116.08972 21. Ems R., Garg A., Ostergard T.A., Miller J.P.. Potential Deep Brain Stimulation Targets for the Management of Refractory Hypertension. Front. Neurosci. (2019.0) 13 93. DOI: 10.3389/fnins.2019.00093 22. Formolo D.A., Gaspar J.M., Melo H.M., Eichwald T., Zepeda R.J., Latini A., Okun M.S., Walz R.. Deep Brain Stimulation for Obesity: A Review and Future Directions. Front. Neurosci. (2019.0) 13 323. DOI: 10.3389/fnins.2019.00323 23. Hsu T.I., Nguyen A., Gupta N., Godbole N., Perisetla N., Hatter M.J., Beyer R.S., Bui N.E., Jagan J., Yang C.. Effectiveness of Deep Brain Stimulation in Treatment of Anorexia Nervosa and Obesity: A Systematic Review. World Neurosurg. (2022.0) 168 179-189. DOI: 10.1016/j.wneu.2022.09.114 24. 24. World Health Organization Global Health Risks: Mortality and Burden of Disease Attributable to Selected Major Risks1st ed.World Health OrganizationGeneva, Switzerland2009V. Global Health Risks: Mortality and Burden of Disease Attributable to Selected Major Risks (2009.0) V 25. Zhou B., Bentham J., Di Cesare M., Bixby H., Danaei G., Cowan M.J., Paciorek C.J., Singh G., Hajifathalian K., Bennett J.E.. Worldwide trends in blood pressure from 1975 to 2015: A pooled analysis of 1479 population-based measurement studies with 19.1 million participants. Lancet (2017.0) 389 37-55. DOI: 10.1016/S0140-6736(16)31919-5 26. Lim S.S., Vos T., Flaxman A.D., Danaei G., Shibuya K., Adair-Rohani H., Amann M., Anderson H.R., Andrews K.G., Aryee M.. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: A systematic analysis for the Global Burden of Disease Study 2010. Lancet (2012.0) 380 2224-2260. DOI: 10.1016/S0140-6736(12)61766-8 27. Yoruk A., Bisognano J.D., Gassler J.P.. Baroreceptor Stimulation for Resistant Hypertension. Am. J. Hypertens. (2016.0) 29 1319-1324. DOI: 10.1093/ajh/hpw074 28. Denker M.G., Cohen D.L.. Resistant Hypertension and Renal Nerve Denervation. Methodist Debakey Cardiovasc. J. (2015.0) 11 240-244. DOI: 10.14797/mdcj-11-4-240 29. Mann S.J.. Neurogenic hypertension: Pathophysiology, diagnosis and management. Clin. Auton. Res. (2018.0) 28 363-374. DOI: 10.1007/s10286-018-0541-z 30. King A.J., Novotny M., Swain G.M., Fink G.D.. Whole body norepinephrine kinetics in ANG II-salt hypertension in the rat. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2008.0) 294 1262-1267. DOI: 10.1152/ajpregu.00819.2007 31. Osborn J.W., Fink G.D.. Region-specific changes in sympathetic nerve activity in angiotensin II-salt hypertension in the rat. Exp. Physiol. (2010.0) 95 61-68. DOI: 10.1113/expphysiol.2008.046326 32. Armitage J.A., Burke S.L., Prior L.J., Barzel B., Eikelis N., Lim K., Head G.A.. Rapid onset of renal sympathetic nerve activation in rabbits fed a high-fat diet. Hypertension (2012.0) 60 163-171. DOI: 10.1161/HYPERTENSIONAHA.111.190413 33. Grassi G.. Sympathomodulatory Effects of Antihypertensive Drug Treatment. Am. J. Hypertens. (2016.0) 29 665-675. DOI: 10.1093/ajh/hpw012 34. DiBona G.F., Esler M.. Translational medicine: The antihypertensive effect of renal denervation. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2010.0) 298 245-253. DOI: 10.1152/ajpregu.00647.2009 35. Mahfoud F., Schlaich M.P., Lobo M.D.. Device Therapy of Hypertension. Circ. Res. (2021.0) 128 1080-1099. DOI: 10.1161/CIRCRESAHA.121.318091 36. Blessing W.W., Reis D.J.. Inhibitory cardiovascular function of neurons in the caudal ventrolateral medulla of the rabbit: Relationship to the area containing A1 noradrenergic cells. Brain Res. (1982.0) 253 161-171. DOI: 10.1016/0006-8993(82)90683-7 37. Guyenet P.G., Stornetta R.L., Souza G.M.P.R., Abbott S.B.G., Brooks V.L.. Neuronal Networks in Hypertension: Recent Advances. Hypertension (2020.0) 76 300-311. DOI: 10.1161/HYPERTENSIONAHA.120.14521 38. Mayorov D.N., Head G.A.. Influence of rostral ventrolateral medulla on renal sympathetic baroreflex in conscious rabbits. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2001.0) 280 577-587. DOI: 10.1152/ajpregu.2001.280.2.R577 39. Kawase T.. Light up the “no-man’s land” on the brain stem. Keio J. Med. (1995.0) 44 133-139. DOI: 10.2302/kjm.44.133 40. Bourns.com 41. Paxinos G., Watson C.. The Rat Brain in Stereotaxic Coordinates (2009.0) 42. Goodchild A.K., Moon E.A.. Maps of cardiovascular and respiratory regions of rat ventral medulla: Focus on the caudal medulla. J. Chem. Neuroanat. (2009.0) 38 209-221. DOI: 10.1016/j.jchemneu.2009.06.002 43. Sved A.F., Blessing W.W., Reis D.J.. Caudal ventrolateral medulla can alter vasopressin and arterial pressure. Brain Res. Bull. (1985.0) 14 227-232. DOI: 10.1016/0361-9230(85)90087-5 44. Bradley D.J., Pascoe J.P., Paton J.F., Spyer K.M.. Cardiovascular and respiratory responses evoked from the posterior cerebellar cortex and fastigial nucleus in the cat. J. Physiol. (1987.0) 393 107-121. DOI: 10.1113/jphysiol.1987.sp016813 45. O’Callaghan E.L., McBryde F.D., Patel N.K., Paton J.F.R.. Examination of the periaqueductal gray as a site for controlling arterial pressure in the conscious spontaneously hypertensive rat. Auton. Neurosci. (2022.0) 240 102984. DOI: 10.1016/j.autneu.2022.102984 46. Sun W., Panneton W.M.. The caudal pressor area of the rat: Its precise location and projections to the ventrolateral medulla. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2002.0) 283 768-778. DOI: 10.1152/ajpregu.00184.2002 47. Ciriello J.. Caudal ventrolateral medulla mediates baroreceptor afferent inputs to subfornical organ angiotensin II responsive neurons. Brain Res. (2013.0) 1491 127-135. DOI: 10.1016/j.brainres.2012.10.064 48. Cooper C.M., Farrand A.Q., Andresen M.C., Beaumont E.. Vagus nerve stimulation activates nucleus of solitary tract neurons via supramedullary pathways. J. Physiol. (2021.0) 599 5261-5279. DOI: 10.1113/JP282064 49. Chen C.Y., Kuo T.B., Hsieh I.T., Yang C.C.. Electrical stimulation of the rostral ventrolateral medulla promotes wakefulness in rats. Sleep Med. (2013.0) 14 1076-1084. DOI: 10.1016/j.sleep.2013.06.011 50. Lertwittayanon W., Devinsky O., Carlen P.L.. Cardiorespiratory depression from brainstem seizure activity in freely moving rats. Neurobiol. Dis. (2020.0) 134 104628. DOI: 10.1016/j.nbd.2019.104628 51. Robinson T.E.. Electrical Stimulation of the Brain Stem in Freely Moving Rats: I. Effects on Behavior, Physiology & Behavior (1978.0) 223-231 52. Feinstein B., Gleason C.A., Libet B.. Stimulation of locus coeruleus in man. Preliminary trials for spasticity and epilepsy. Stereotact. Funct. Neurosurg. (1989.0) 52 26-41. DOI: 10.1159/000099484 53. Young R.F., Tronnier V., Rinaldi P.C.. Chronic stimulation of the Kölliker-Fuse nucleus region for relief of intractable pain in humans. J. Neurosurg. (1992.0) 76 979-985. DOI: 10.3171/jns.1992.76.6.0979 54. Elias G.J.B., Loh A., Gwun D., Pancholi A., Boutet A., Neudorfer C., Germann J., Namasivayam A., Gramer R., Paff M.. Deep brain stimulation of the brainstem. Brain (2021.0) 144 712-723. DOI: 10.1093/brain/awaa374 55. Green A.L., Hyam J.A., Williams C., Wang S., Shlugman D., Stein J.F., Paterson D.J., Aziz T.Z.. Intra-operative deep brain stimulation of the periaqueductal grey matter modulates blood pressure and heart rate variability in humans. Neuromodulation (2010.0) 13 174-181. DOI: 10.1111/j.1525-1403.2010.00274.x 56. Green A.L., Wang S., Owen S.L., Xie K., Liu X., Paterson D.J., Stein J.F., Bain P.G., Aziz T.Z.. Deep brain stimulation can regulate arterial blood pressure in awake humans. Neuroreport (2005.0) 16 1741-1745. DOI: 10.1097/01.wnr.0000183904.15773.47 57. Green A.L., Wang S., Bittar R.G., Owen S.L., Paterson D.J., Stein J.F., Bain P.G., Shlugman D., Aziz T.Z.. Deep brain stimulation: A new treatment for hypertension?. J. Clin. Neurosci. (2007.0) 14 592-595. DOI: 10.1016/j.jocn.2006.04.015 58. DeLalio L.J., Sved A.F., Stocker S.D.. Sympathetic Nervous System Contributions to Hypertension: Updates and Therapeutic Relevance. Can. J. Cardiol. (2020.0) 36 712-720. DOI: 10.1016/j.cjca.2020.03.003 59. Kurtz T., Griffin K., Bidani A., Davisson R., Hall J.. Recommendations for Blood Pressure Measurement in Humans and Experimental Animals. Hypertension (2005.0) 45 299-310. DOI: 10.1161/01.HYP.0000150857.39919.cb 60. Ewing S.G., Lipski W.J., Grace A.A., Winter C.. An inexpensive, charge-balanced rodent deep brain stimulation device: A step-by-step guide to its procurement and construction. J. Neurosci. Methods (2013.0) 219 324-330. DOI: 10.1016/j.jneumeth.2013.08.003 61. Fluri F., Mützel T., Schuhmann M.K., Krstić M., Endres H., Volkmann J.. Development of a head-mounted wireless microstimulator for deep brain stimulation in rats. J. Neurosci. Methods (2017.0) 291 249-256. DOI: 10.1016/j.jneumeth.2017.08.024 62. Forni C., Mainard O., Melon C., Goguenheim D., Kerkerian-Le Goff L., Salin P.. Portable microstimulator for chronic deep brain stimulation in freely moving rats. J. Neurosci. Methods (2012.0) 209 50-57. DOI: 10.1016/j.jneumeth.2012.05.027 63. Luo Y.S., Luo Y.L., Ashford E.B., Morin R.R., White W.J., Fisher T.F.. Comparison of catheter lock solutions in rats. Proceedings of the 51st Annual Meeting of the American Association for Laboratory Animal Science 64. Wang Y., Cong Y., Li J., Li X., Li B., Qi S.. Comparison of invasive blood pressure measurements from the caudal ventral artery and the femoral artery in male adult SD and Wistar rats. PLoS ONE (2013.0) 8. DOI: 10.1371/journal.pone.0060625
--- title: Effects of Triheptanoin on Mitochondrial Respiration and Glycolysis in Cultured Fibroblasts from Neutral Lipid Storage Disease Type M (NLSD-M) Patients authors: - Nelida Inés Noguera - Daniela Tavian - Corrado Angelini - Francesca Cortese - Massimiliano Filosto - Matteo Garibaldi - Sara Missaglia - Ariela Smigliani - Alessandra Zaza - Elena Maria Pennisi journal: Biomolecules year: 2023 pmcid: PMC10046759 doi: 10.3390/biom13030452 license: CC BY 4.0 --- # Effects of Triheptanoin on Mitochondrial Respiration and Glycolysis in Cultured Fibroblasts from Neutral Lipid Storage Disease Type M (NLSD-M) Patients ## Abstract Neutral lipid storage disease type M (NLSD-M) is an ultra-rare, autosomal recessive disorder that causes severe skeletal and cardiac muscle damage and lipid accumulation in all body tissues. In this hereditary pathology, the defective action of the adipose triglyceride lipase (ATGL) enzyme induces the enlargement of cytoplasmic lipid droplets and reduction in the detachment of mono- (MG) and diglycerides (DG). Although the pathogenesis of muscle fiber necrosis is unknown, some studies have shown alterations in cellular energy production, probably because MG and DG, the substrates of Krebs cycle, are less available. No tests have been tried with medium-chain fatty acid molecules to evaluate the anaplerotic effect in NLSD cells. In this study, we evaluated the in vitro effect of triheptanoin (Dojolvi®), a highly purified chemical triglyceride with seven carbon atoms, in fibroblasts obtained from five NLSD-M patients. Glycolytic and mitochondrial functions were determined by Seahorse XF Agylent Technology, and cellular viability and triglyceride content were measured through colorimetric assays. After the addition of triheptanoin, we observed an increase in glycolysis and mitochondrial respiration in all patients compared with healthy controls. These preliminary results show that triheptanoin is able to induce an anaplerotic effect in NLSD-M fibroblasts, paving the way towards new therapeutic strategies. ## 1. Introduction Neutral lipid storage disease type M (NLSD-M) is an ultra-rare, recessive disorder that causes severe skeletal and cardiac muscle damage with lipid accumulation in virtually all tissues of the body, because of the mutations in the PNPLA2 gene-coding adipose triglyceride lipase (ATGL). To date, no therapy is available for this very disabling disease, and the pathogenesis is still largely unknown. A decrease in the lipolytic action of the ATGL enzyme induces the enlargement of cytoplasmic lipid droplets (LDs) and a reduction in mono- and diglycerides for energy production in the cell. The cytoplasmic accumulation of the lipids alone does not totally explain the muscle atrophy seen in patients [1]. With this assumption, some studies have been carried out to evaluate whether altered functioning in the aerobic mitochondrial metabolism was also involved in the pathogenesis of the disease [2]. Mitochondria generate up to $90\%$ of the energy in the cell, producing adenosine triphosphate (ATP) in the β-oxidation process, through the metabolism of fatty acids (FAs). This pathway also produces molecules that are used as cellular structural components for post-translational modifications of proteins and in signaling cascades [3]. In other muscle disorders due to defects in the lipid metabolism, such as very-long-chain acyl-CoA dehydrogenase deficiency (VLCAD), long-chain 3-hydroxyacyl-CoA dehydrogenase deficiency (LCHAD), trifunctional protein deficiency (TFP), and carnitine palmitoyl transferase II deficiency (CPT II), medium-branched chain fatty acids were successfully utilized as anaplerotic treatments. The MCT diet plus carnitine supplementation in these disorders stimulates citric acid cycle function and ATP production, enhances gluconeogenesis and urea cycle function, and maintains intra-mitochondrial homeostasis [4]. Triheptanoin (Dojolvi®) is a synthetic medium-chain triglyceride (MCT) developed by Ultragenyx Pharmaceutical Inc., Food-and-Drug-Administration approved, for use in the treatment of inherited metabolic disorders [5]. Triheptanoin is a triglyceride of three odd-chain fatty acids (heptanoate, C7), cleaved by intestinal lipases, and heptanoate is absorbed by the gastrointestinal tract [6]. Intracellularly, C7 crosses the mitochondrial membranes without an active transport system and can enter the β-oxidation cycle. The rationale to use triheptanoin in patients with NLSD-M is based on the hypothesis that energy deficiency in fatty acid oxidation may be exacerbated by the depletion of catalytic intermediates in the TCA and would, thus, benefit from the anaplerotic effect of triheptanoin, as well as to its ability to bypass metabolic block induced by ATGL deficiency. Vockley et al. demonstrated that triheptanoin treatment is associated with reduced major clinical events, such as rhabdomyolysis, cardiomyopathy, and hypoglycemia in patients with long-chain fatty acid oxidation disorders [7]. However, in studies on mice, using long-term supplementation of triheptanoin induces de novo biosynthesis and elongation of fatty acids in both WT and VLCAD-/- mice, altering the hepatic and cardiac fatty acid composition to no physiological profiles, characterized by a strong reduction in polyunsaturated fatty acids and a marked increase in monounsaturated species [8]. These results suggest that the time of application and the dose should be individually adapted to meet the energy demand of the patient. Our study explores the potential beneficial effect of triheptanoin on metabolism on NLSD-M cells in vitro. ## 2.1. Patients Two women and three men affected by NLSD-M, from the Italian NLSD Group [9], aged 50 y and 68 y, underwent skin biopsy after informed consent was provided. Clinical and genetic data of these patients are briefly summarized in Table 1. Still, ambulatory patients were considered less severely affected than non-ambulatory patients, and patients with weakness in only a few muscles were considered to have very mild disease. Control fibroblasts were obtained after informed consent, from skin biopsies of 3 subjects at different ages (12, 58, and 73 years old), undergoing orthopedic surgery. Controls were tested for increases in CK. Myopathic signs and symptoms were excluded by neurological evaluation. Fibroblasts obtained from controls were tested for ATGL gene mutations. The study was conducted in accordance with the Helsinki Declaration. ## 2.2. Characterization of Metabolism via Seahorse XF Agylent Technology Fibroblasts from the NLSD-M patients and the normal subjects were cultured in DMEM high-glucose medium (Euroclone, MI, Italy), $10\%$ fetal bovine serum (FBS) (GIBCO-BRL), 20 mM Hepes, 100 U/mL penicillin, and 100 µg/mL streptomycin (GIBCO-BRL). Cultures were maintained at 37 °C in a $5\%$ CO2 humidified incubator. Fibroblasts were treated with triheptanoin at 25 µM for one week before analysis. At the time of the experiment, 25 µM of triheptanoin or DMSO was added. Using a Seahorse Bioscience XFe96 analyzer (Agilent Technologies, Santa Clara, CA, USA), mitochondrial and glycolytic function were assessed, and extracellular flux assay kits were used to measure oxygen consumption (OCR) and glycolytic flux (ECAR), as previously described [10,11]. Briefly, fibroblasts were seeded and incubated overnight at a density of 10 × 104 cells/well using cell culture microplates, four wells for each condition (XFe Seahorse, Agilent Technology, Santa Clara, CA, USA). Before the start of the test, the cells were incubated in XF-DMEM medium, pH 7.4 (Agilent Technology, Santa Clara, CA, USA) at 37 °C without CO2 for 1 h. To evaluate mitochondrial function, the oxygen consumption rate (OCR) was measured using the Seahorse Bioscience XF Cell Mito Stress Test (Agilent Technology, Santa Clara, CA, USA). Mitochondrial oxidative phosphorylation (OXPHOS) was analyzed under basal conditions, in the presence of 2 µM oligomycin, 1 µM carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP), and 0.5 mM rotenone/antimycin A (R/A). Oligomycin inhibits ATP synthase (V complex), as it decreases the flow of electrons through the electron transport change (ETC), resulting in a reduction in mitochondrial respiration. This reduction in OCR is linked to cellular ATP production. FCCP is a decoupling agent that interrupts the potential of the mitochondrial membrane. Rotenone, a complex I inhibitor, and antimycin A, a complex III inhibitor, shut down mitochondrial respiration, enabling the calculation of non-mitochondrial respiration. As a result, the oxygen consumption of complex IV reaches a maximum and can be used to calculate respiratory reserve capacity, defined as the difference between maximum respiration and baseline respiration. Respiratory reserve capacity is a measure of the cell’s ability to respond to increased energy demand or under stress. The extracellular acidification rate (ECAR) caused by the conversion of glucose into pyruvate, and subsequently into lactate, is indicative of glycolysis. The Glycolysis Rate Assay measured glycolytic rates for basal conditions and compensatory glycolysis following mitochondrial inhibition after the injection of 0.5 mM R/A. The final injection is 2-deoxy-glucose (2-DG), a glucose analogue, which inhibits glycolysis through competitive binding with glucose hexokinase. The resulting reduction in ECAR confirms that the ECAR produced in the experiment is due to glycolysis. ## 2.3. MTT Assay Thus, 2 × 104 cells for each NLSD-M line were washed with phosphate-buffered saline (DPBS) 24 h after seeding and then fed with DMEM supplemented with 12.5 or 25 μM triheptanoin. Triheptanoin was diluted in DMSO. Cells grown in the medium supplemented only with DMSO were designated as controls. At the end of the short treatment (24 h, 48 h, and 72 h), the Thiazolyn Blue Tetrazolium Bromide (MTT, Merck, Darmstadt, Germany) assay was used to quantify cell viability and proliferation according to data sheet. Absorbance was measured with VICTOR® Nivo™ Reader. ( PerkinElmer, La Jolla, CA, USA). Each experimental sample was tested in triplicate. ## 2.4. Triglyceride Intracellular Content NLSD-M and control fibroblasts were cultured in DMEM supplemented with 12.5 or 25 μM triheptanoin. The cellular triacylglycerol (TAG) accumulation was quantified before and after 9-day treatment using Triglyceride Quantification Colorimetric kit (Biovision Diagnostics, LLC, Collinsville, USA) according to the instructions. In brief, 1 × 106 cells were homogenized in 1 mL solution containing $5\%$ NP-40 in water and then incubated in a reaction mix for 1 h in dark conditions. The absorbance was measured at 570 nm with VICTOR® Nivo™ Reader (PerkinElmer, La Jolla, California, USA). ## 2.5. Immunofluorescence Analysis 2 × 105 NLSDM fibroblasts were seeded on coverslips in an Earle’s MEM culture medium and allowed to adhere for 12 h. Then, the culture medium was supplemented with 12.5 or 25 μM triheptanoin. After 9 days, the cells were fixed on coverslips using $3\%$ paraformaldehyde, rinsed with distilled water, and incubated with Nile Red (NR) solution for 20 min in the dark. NR staining solution was freshly prepared in DPBS (1:100 v/v) from a saturated solution (1 mg/mL) in DMSO. Images were obtained using a Leica DM5000B microscope equipped with a 40× objective (NR staining: excitation at 450–490 nm) and analyzed using the public domain Java image processing program ‘WCIF ImageJ 1.35j’ (developed by W. Rasband, NIH, Bethesda, MD, USA). This software allows us to isolate components with the same wavelength and to measure and quantify different parameters such as area, numbers of selected units (LDs, in this case), and pixels per cell; 15 inches [2] was chosen as threshold value for LD area to discard fluorescent emissions due to impurity. ## 2.6. Statistical Analysis Data were analyzed using GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA, USA). Statistical analysis was performed using the Mann–Whitney test and Student’s t-test. Statistical significance was established at $p \leq 0.05.$ ## 3.1. Mitochondrial Respiration Basal and maximal respiration, proton leak, ATP, and spare respiratory capacity were determined as attributes of oxidative phosphorylation (OXPHOS). Cellular basal respiration represents the energetic demand of the cell under baseline conditions; maximal respiration measures maximal oxygen consumption rate that cells can achieve; spare respiratory capacity measures the difference between the ATP produced by OXPHOS at basal and maximal activity. These parameters were significantly increased after treatment with triheptanoin in the five patients analyzed, indicating that triheptanoin ameliorates the capability of NLSD-M fibroblasts to respond to an energetic demand (Figure 1A and Table 2). The improvement in mitochondrial respiration allowed for an increase in ATP production in fibroblasts from patients NLSD 1, NLSD 3, and NLSD 5. The other two patients did not show an enhancement in ATP production, probably because of a slight increase in proton leak levels (Figure 1A and Table 2). To produce energy, the electrons from substrate oxidation are passed through the respiratory chain on the inner mitochondrial membrane (IMM), and the flux is maintained by proton pumping against gradient from the inner mitochondrial matrix to the intermembrane space via respiratory complexes I, III, and IV. All three control samples did not show variations in any attributes of OXPHOS analyzed (Figure 1B and Table 2). Interestingly, basal OXPHOS mean values are normalized after triheptanoin treatment in NLSD-M patients’ fibroblasts, whereas there is no difference in normal controls’ fibroblasts. Triheptanoin treatment also improved maximal respiration, spare respiratory capacity, and ATP production. The trend was pronounced but the differences were not significant, probably because of the low number of samples analyzed (Figure 1C and Table 3). Moreover, excluding the NLSD 4 patient whose molecular defect completely abolishes ATGL enzymatic activity, the difference is significant in maximal respiration and in SRC (Figure 1D), suggesting that the treatment might enhance residual enzymatic activity. ## 3.2. Glycolytic Respiration Glucose in cells is converted to pyruvate and then to lactate in the cytoplasm, or to CO2 and water in the mitochondria. The conversion of glucose to lactate results in a net production of protons into the extracellular medium, which is detected by the XF Analyzer as extracellular acidification rate (ECAR). The rates of glycolysis were determined as a percentage increase in ECAR after the addition of the complex I and complex III mitochondrial inhibitors rotenone/antimycin and of 2-Deoxy Glucose, an inhibitor of glycolysis to confirm pathway specificity. Compensatory glycolysis is the rate of glycolysis in cells following the addition of mitochondrial inhibitors, which effectively block oxidative phosphorylation and drive compensatory changes in the cell to use glycolysis to meet the cells’ energy demands. Consistently, we found that triheptanoin treatment promoted the activation of the aerobic glycolysis pathway in NLSD fibroblasts. Indeed, we observed a significant basal glycolysis increase in NLSD 2, 3, 4, and 5, while NLSD1 presented an increase in the compensatory glycolysis values (Figure 2A,C and Table 4). These effects were not observed in controls (Figure 2B,C and Table 4). ## 3.3. Cellular Viability and TAG Accumulation Study To investigate the effect of triheptanoin supplementation on proliferation, a time course of 24 h, 48 h, and 72 h was performed on cells from NLSD 2, NLSD3, NLSD4, and NLSD5 patients. After 48 h, cell viability significantly increased in the NLSD 2 cell line cultured with 12.5 μM triheptanoin (p ≤ 0.05), in comparison with untreated cells. After 72 h, NLSD 2 and 3, treated with 12.5 or 25 μM triheptanoin, showed significantly higher viability than those of untreated cells (p ≤ 0.05). For NLSD 4 and 5 fibroblasts, no significant differences were observed (Figure 3A). Finally, intracellular TAG content was detected before and after 9-day treatment. The results showed that triheptanoin did not decrease neutral lipid abnormal storage. On the contrary, NLSD-M cells slightly increased the TAG amount stored in cytoplasmic lipid droplets (Figure 3B). Immunofluorescence analysis of LDs in NLSD-M fibroblasts confirmed data obtained by TAG evaluation. After treatment, the cells did not show a decrease in LD number and size in comparison with untreated cells (Figure 3C). ## 4. Discussion To date, there are no studies examining the cellular oxidative metabolic activity of patients with NLSD. The goal of this study was to evaluate if triheptanoin can ameliorate the glycolytic and oxidative metabolism when administered to fibroblasts of patients with NLSD-M. The method used to analyze the oxidative metabolism of affected fibroblasts has been widely used in other studies, mainly for oncological pathologies [7,8,10], but has never been used for the study of NLSD. The results demonstrate that triheptanoin ameliorates the metabolic performances in NLSD-M fibroblasts. Basal respiration represents a variable percentage of the maximal respiratory capacity and is characteristic of each cell type [12,13]. Our results showed that treatment with triheptanoin could normalize the basal levels of NLSD-M fibroblasts. When necessary to enable energetic adaptation, mitochondrial respiration can suddenly increase to the maximum levels to synthesize more ATP. To achieve this, cells use the spare respiratory capacity (SRC). This process is tightly controlled by the nature and flow of nutrients that can be oxidized in the mitochondrial matrix by the tricarboxylic acid (TCA) cycle [8]. SRC is increased in all NLSD-M patients after treatment with triheptanoin, indicating that acetyl-CoA and Propionyl-CoA, the metabolized products of triheptanoin, provide an anaplerotic effect by replenishing deficient TCA cycle intermediates in NLSD-M. SRC represents a particularly robust functional parameter to evaluate mitochondrial reserve. It characterizes the mitochondrial capacity to meet extra energy requirements and represents a determination of mitochondrial fitness [13]. In NLSD-M, both inter- and intra-familial variability in clinical phenotypes can be observed. As physical endurance training and starvation can improve SRC levels [14], different degrees of clinical severity could partially depend on lifestyle, which is known to involve muscle activity, and dietary regimen [6]. The importance of anaplerotic reactions for the regulation of amino acid, glucose, and FA metabolism is unchallenged. Our results suggest that FA metabolism is interrupted in NLSD-M patients due to impaired anaplerosis of the TCA cycle, as no long-chain FA can be used. Triheptanoin derivatives may function as an alternative source of anaplerotic substrates for the TCA cycle, bypassing the block. To produce energy, the electrons from substrate oxidation generated a gradient proton across the inner mitochondrial membrane, but oxidative phosphorylation is incompletely ‘coupled’, since protons can ‘leak’ across the inner membrane balancing the gradient, without ATP synthesis. Leaked protons link reactive oxygen molecules to produce H2O and thermic dispersion to regulate physiological processes, such as thermogenesis and ROS reduction [15]. The slight increase in proton leak levels observed in NLSD-M patients after treatment with triheptanoin could be beneficial for patients because it contribute to reducing ROS production in the cells. The balance between glycolysis and oxidative phosphorylation is believed to be critical for maintaining cellular bioenergetics. We observed that in NLSD-M fibroblasts, triheptanoin also stimulates glycolysis. During starvation, when the rates of lipolysis are highest, a major fraction (up to $30\%$) of the free fatty acids generated from triglyceride breakdown is re-esterified back to triglyceride in adipose tissue [16,17]. This process requires a source of 3-glycerol phosphate, which is generally supplied by glucose via glycolysis. It is possible that these mechanisms are involved in the stimulation of glycolysis by triheptanoin in NLSD-M. Data obtained after a short triheptanoin treatment revealed that two different NLSD-M cell lines were able to grow faster than untreated cells. These fibroblasts were collected from two siblings, compound heterozygotes for two different missense mutations (p.L56R and p.I193F) [6]. A functional study performed to clarify the pathogenic effect of these variations showed that both missense mutations caused a partial loss of ATGL function [18]. In this case, supplementation with triheptanoin can provide an additional fat source that can be utilized to increase cellular energy production. The greater amount of energy could balance the partial decrease in TAG hydrolysis and ameliorate cell viability. On the contrary, in NLSD 4 and 5 patients, no beneficial effects on cell viability were observed. Genetic analysis of NLSD 4 subjects displayed an ATGL frameshift mutation (c.695delT), which determines no protein production [6]. Although triheptanoin supplementation can increase energy level, it could not be sufficient to compensate for the total loss of ATGL function. No functional study has been performed to clarify the impact of ATGL mutation on protein function in NLSD 5. It could be hypothesized that, also in this case, the treatment with triheptanoin is not sufficient to balance the effect of gene mutation on lipase activity. Finally, triheptanoin treatment did not show any beneficial effect on lipid accumulation. As expected, the amount of TGs remained constant after the triheptanoin addition, because it does not exert any lipolytic function on neutral lipids stored inside LDs. The severe clinical expression of myopathy due to ATGL gene mutation is probably caused by various pathogenic mechanisms. We postulated that an altered cellular metabolism can participate in myopathic damage. We hypothesized that a reduction in the availability of mono- and diglycerides, due to the impairment of ATGL lipase function, could remove metabolites from the Krebs cycle and beta-oxidation. Our findings are in line with previous studies, suggesting that triheptanoin in NLSD-M produces acetyl-CoA and propionyl-CoA. Propionyl-CoA can be further converted into succinyl-CoA, an anaplerotic substrate for the tricarboxylic acid cycle, supporting mitochondrial energy production. The limitations of the study are the small sample size obtained from patients suffering with this ultra-rare disease. The results of this experiment cannot be translated directly into humans but constitute an interesting prerequisite for understanding the pathology and may indicate a therapeutic strategy. ## References 1. Akman H.O., Davidzon G., Tanji K., MacDermott E.J., Larsen L., Davidson M.M., Haller R.G., Szczepaniak L.S., Lehman T.J., Hirano M.. Neutral lipid storage disease with subclinical myopathy due to a retrotransposal insertion in the PNPLA2 gene. Neuromuscul. Disord. (2010) 20 397-402. DOI: 10.1016/j.nmd.2010.04.004 2. Ørngreen M.C., Laforêt P., Preisler N., Andersen G., Vissing J.. Blocked Muscle Fat Oxidation During Exercise in Neutral Lipid Storage Disease. Arch. Neurol. (2012) 69 530. DOI: 10.1001/archneurol.2011.631 3. Tucci J., Chen T., Margulis K., Orgel E., Paszkiewicz R.L., Cohen M.D., Oberley M.J., Wahhab R., Jones A.E., Divakaruni A.S.. Adipocytes Provide Fatty Acids to Acute Lymphoblastic Leukemia Cells. Front. Oncol. (2021) 11 665763. DOI: 10.3389/fonc.2021.665763 4. Shirley M.. Triheptanoin: First Approval. Drugs (2020) 80 1595-1600. DOI: 10.1007/s40265-020-01399-5 5. Vockley J., Burton B., Berry G., Longo N., Phillips J., Sanchez-Valle A., Chapman K., Tanpaiboon P., Grunewald S., Murphy E.. Effects of triheptanoin (UX007) in patients with long-chain fatty acid oxidation disorders: Results from an open-label, long-term extension study. J. Inherit. Metab. Dis. (2020) 44 253-263. DOI: 10.1002/jimd.12313 6. Pennisi E.M., Arca M., Bertini E., Bruno C., Cassandrini D., D’Amico A., Garibaldi M., Gragnani F., Maggi L.. Neutral Lipid Storage Diseases: Clinical/genetic features and natural history in a large cohort of Italian patients. Orphanet J. Rare Dis. (2017) 12 90. DOI: 10.1186/s13023-017-0646-9 7. Albanesi J., Noguera N., Banella C., Colangelo T., De Marinis E., Leone S., Palumbo O., Voso M., Ascenzi P., Nervi C.. Transcriptional and Metabolic Dissection of ATRA-Induced Granulocytic Differentiation in NB4 Acute Promyelocytic Leukemia Cells. Cells (2020) 9. DOI: 10.3390/cells9112423 8. Banella C., Catalano G., Travaglini S., Pelosi E., Ottone T., Zaza A., Guerrera G., Angelini D.F., Niscola P., Divona M.. Ascorbate Plus Buformin in AML: A Metabolic Targeted Treatment. Cancers (2022) 14. DOI: 10.3390/cancers14102565 9. Quattrocchi A., Quattrocchi A., Maiorca C., Maiorca C., Billi M., Billi M., Tomassini S., Tomassini S., De Marinis E., De Marinis E.. Genetic lesions disrupting calreticulin 3′-untranslated region in JAK2 mutation-negative polycythemia vera. Am. J. Hematol. (2020) 95 25911. DOI: 10.1002/ajh.25911 10. Schuh R.A., Jackson K.C., Khairallah R.J., Ward C.W., Spangenburg E.E.. Measuring mitochondrial respiration in intact single muscle fibers. Am. J. Physiol. Integr. Comp. Physiol. (2012) 302 R712-R719. DOI: 10.1152/ajpregu.00229.2011 11. Horan M.P., Pichaud N., Ballard J.W.O.. Review: Quantifying Mitochondrial Dysfunction in Complex Diseases of Aging. J. Gerontol. Ser. A (2012) 67 1022-1035. DOI: 10.1093/gerona/glr263 12. Hill B.G., Benavides G.A., Lancaster J.R., Ballinger S., Dell’Italia L., Zhang J., Darley-Usmar V.M.. Integration of cellular bioenergetics with mitochondrial quality control and autophagy. Biol. Chem. (2012) 393 1485-1512. DOI: 10.1515/hsz-2012-0198 13. Marchetti P., Fovez Q., Germain N., Khamari R., Kluza J.. Mitochondrial spare respiratory capacity: Mechanisms, regulation, and significance in non-transformed and cancer cells. FASEB J. (2020) 34 13106-13124. DOI: 10.1096/fj.202000767R 14. Larsen S., Rabøl R., Stride N., Helge J.W., Dela F., Hey-Mogensen M.. The influence of age and aerobic fitness: Effects on mitochondrial respiration in skeletal muscle. Acta Physiol. (2012) 205 423-432. DOI: 10.1111/j.1748-1716.2012.02408.x 15. Divakaruni A.S., Brand M.D.. The Regulation and Physiology of Mitochondrial Proton Leak. Physiology (2011) 26 192-205. DOI: 10.1152/physiol.00046.2010 16. Elia M., Zed C., Neale G., Livesey G.. The energy cost of triglyceride-fatty acid recycling in nonobese subjects after an overnight fast and four days of starvation. Metabolism (1987) 36 251-255. DOI: 10.1016/0026-0495(87)90184-3 17. Patel D., Kalhan S.. Glycerol Metabolism and Triglyceride-Fatty Acid Cycling in the Human Newborn: Effect of Maternal Diabetes and Intrauterine Growth Retardation. Pediatr. Res. (1992) 31 52-58. DOI: 10.1203/00006450-199201000-00010 18. Missaglia S., Tasca E., Angelini C., Moro L., Tavian D.. Novel missense mutations in PNPLA2 causing late onset and clinical heterogeneity of neutral lipid storage disease with myopathy in three siblings. Mol. Genet. Metab. (2015) 115 110-117. DOI: 10.1016/j.ymgme.2015.05.001
--- title: Efficient Pause Extraction and Encode Strategy for Alzheimer’s Disease Detection Using Only Acoustic Features from Spontaneous Speech authors: - Jiamin Liu - Fan Fu - Liang Li - Junxiao Yu - Dacheng Zhong - Songsheng Zhu - Yuxuan Zhou - Bin Liu - Jianqing Li journal: Brain Sciences year: 2023 pmcid: PMC10046767 doi: 10.3390/brainsci13030477 license: CC BY 4.0 --- # Efficient Pause Extraction and Encode Strategy for Alzheimer’s Disease Detection Using Only Acoustic Features from Spontaneous Speech ## Abstract Clinical studies have shown that speech pauses can reflect the cognitive function differences between Alzheimer’s Disease (AD) and non-AD patients, while the value of pause information in AD detection has not been fully explored. Herein, we propose a speech pause feature extraction and encoding strategy for only acoustic-signal-based AD detection. First, a voice activity detection (VAD) method was constructed to detect pause/non-pause feature and encode it to binary pause sequences that are easier to calculate. Then, an ensemble machine-learning-based approach was proposed for the classification of AD from the participants’ spontaneous speech, based on the VAD Pause feature sequence and common acoustic feature sets (ComParE and eGeMAPS). The proposed pause feature sequence was verified in five machine-learning models. The validation data included two public challenge datasets (ADReSS and ADReSSo, English voice) and a local dataset (10 audio recordings containing five patients and five controls, Chinese voice). Results showed that the VAD Pause feature was more effective than common feature sets (ComParE: 6373 features and eGeMAPS: 88 features) for AD classification, and that the ensemble method improved the accuracy by more than $5\%$ compared to several baseline methods ($8\%$ on the ADReSS dataset; $5.9\%$ on the ADReSSo dataset). Moreover, the pause-sequence-based AD detection method could achieve $80\%$ accuracy on the local dataset. Our study further demonstrated the potential of pause information in speech-based AD detection, and also contributed to a more accessible and general pause feature extraction and encoding method for AD detection. ## 1. Introduction Alzheimer’s disease (AD) is an irreversible disease and there is little physicians can do when patients progress to advanced stages [1]. According to estimates from the World Alzheimer’s Disease Report 2019 and Alzheimer’s Disease International, there are over 50 million people living with dementia in the world, and the projected estimates for 2050 reach above 150 million [2]. Early detection holds great value for AD patients, as patients that are diagnosed early and seek treatment can be helped by numerous interventions to maintain their current state or delay cognitive decline [3]. Deterioration in speech and language production is among the first signs of the disease [4], and people with AD exhibit language impairment long before they are diagnosed [5]. In clinical practice, neuropsychological assessments are often used to initially screen for AD, evaluating their cognitive status through the patients’ manner of speaking and content [6]. However, this evaluation method relies on subjective assessment by human experts, which is difficult to quantify and meet the needs of AD patients for home testing. Several researchers have focused on the use of spontaneous speech to detect AD. Yuan et al. [ 7] obtained the best accuracy of $89.6\%$ using text-based features and acoustic features. Agbavor et al. [ 8] developed an end-to-end AD detection method with average area under the curve of 0.846. Although these methods have achieved good performance, they are still prone to voice-to-text transcriptional accuracy issues. For patients who are not native English speakers, and elderly people with dialect accents, the generalization ability of text-based feature AD detection methods may be limited. Compared to text-based feature approaches, using only acoustic features allows for extracting direct information from the speech itself, which is more robust to the native language of the subject. Luz [9] used spontaneous speech data from the “cookie theft” picture description task in the Pitt database, extracted statistical and nominal acoustic features, and achieved a $68\%$ accuracy using a Bayesian classifier. Eyben et al. [ 10] designed a unified feature set, Computational Paralinguistics ChallengE (ComParE), based on the features and lessons learned from the 2009–2012 challenge, which has general applicability in paralinguistic information extraction. Eyben et al. [ 11] proposed a standard acoustic parameter set, the extended Geneva minimalistic acoustic parameter set (eGeMAPS), such that the research results obtained in various areas of automatic speech analysis could be properly compared. Moreover, some researchers explored acoustic features for AD diagnosis from the perspective of digital signal processing, such as higher-order spectral features [12], fractal features [13], and wavelet-packet-based features [14]. Speech pauses can reflect the speaker’s cognitive function. Patients with neurodegenerative diseases have decreased cognitive function and often exhibit language impairment, such as frontotemporal lobar degeneration [15], Lewy body spectrum disorders [16], primary progressive aphasia [17] and AD [18]. Patients with AD experience disfluency, a high number of pauses, and repetition in speech, which is attributed to lexical retrieval difficulties due to cognitive decline [19]. Therefore, pauses are often used to analyze the speech and language of AD patients in order to assess their cognitive level [7,20,21]. Pauses mainly include filled and silent (unfilled) pauses. English has two common filled pauses: “uh” and “um” [7]. The term “silent pause” represents a temporal region in which a speaker does not utter a word, phrase, or sentence during spontaneous speech [22]. Researchers detected AD by calculating features such as the location, duration and frequency of the two types of pauses [23,24,25]. However, in some AD-related classification tasks, silent pauses performed better than filled pauses [23,26]. Moreover, silent pauses are easier to obtain from speech signals and more convenient for AD detection in daily conversation. In addition, silent pauses can be used to detect other acquired language disorders [27,28]. Therefore, this paper focuses on the study of silent pauses (hereafter, silent pauses are referred to as pauses). Pauses have been shown to be effective in the detection of AD. Vincze et al. [ 23] acquired speech recordings by stimulating participants’ memory systems through three connected speech tasks, and used the PRAAT software for language analysis. The results show that the length, number, and rate of pauses differed between AD patients and healthy controls. Yuan et al. [ 7] likewise studied the variability of speech pauses between AD and healthy controls, classifying pauses into three discrete categories (under 0.5 s, 0.5–2 s, and over 2 s) based on their duration, with three types of punctuation (“,”, “.”, and “…”), marking the pauses in different groups and encoding them into word sequences to form pause-encoded transcripts. They used the ERNIE model for classification and achieved a high accuracy on the ADReSS test set. However, the approach remains a method of speech and language fusion, relying on speech transcription. Moreover, the discrete representation of pause features may not necessarily reflect all cognitive impairment information contained in the pause. In order to provide a simple, widely available screening method for AD, this paper further explores the use of pauses in AD detection and proposes a voice activity detection (VAD)-based pausing feature recognition strategy, which can automatically generate pausing sequences in patients’ spontaneous speech. We compare the performance of the constructed pause features, with the public feature set on different machine-learning classifiers. The effectiveness of VAD Pause was verified on a local dataset. The main contributions of this study are listed as follows: [1] We introduce an acoustic feature named the VAD Pause, extracted from speech signals on the newly shared ADReSS and ADReSSo English datasets, and the VAD Pause features are tested on a local Chinese dataset. [2] A machine-learning-based ensemble approach using only acoustic features is proposed for AD classification. [3] Statistical results revealed that VAD Pause features are higher in classification accuracy than the acoustic feature sets, and the ensemble method has higher classification value than the public feature-set-based method. ## 2.1. System Framework The framework of the proposed method is shown in Figure 1. First, the participant was asked to perform a picture description task (“Cookie Theft” picture), during which their voice was recorded. Second, the speech signal was pre-processed to standardize it for next processing. Then, two types of features were extracted from speech signals. The first was the VAD-based pause feature proposed by the research, and the other was the feature from the common feature set. We compare the effectiveness of features on different classifiers. At the same time, in order to integrate the advantages of different types of features, we also designed a feature ensemble method based on the major voting strategy to obtain a better AD detection performance. The following will be introduced in detail from dataset descriptions, information preprocessing, feature extraction, ensemble classification methods, and evaluation methods. ## 2.2.1. Public Dataset To evaluate the performance of our method, publicly available English audio datasets from references [29,30] were used for training and testing (ADReSS Challenge and ADReSSo Challenge). The challenges were organized by Reacher Saturnino Luz, Fasih Haider, and Sofia de la Fuente Garcia from the University of Edinburgh, and Davida Fromm and Brian MacWhinney of Carnegie Mellon University. They targeted a difficult automatic prediction problem of societal and medical significance, namely the detection of cognitive impairment and Alzheimer’s dementia. The competitions provide standard datasets of spontaneous speech, defining a shared task by which different spontaneous speech-based AD detection approaches can be compared. The datasets consist of a set of recordings of picture descriptions produced by patients with an AD diagnosis and cognitively normal subjects, who were asked to describe the “Cookie Theft” picture from the Boston Diagnostic Aphasia Examination [31,32]. The ADReSS Challenge dataset consists of recordings of 156 participants, while the ADReSSo Challenge dataset contains speech clips of 237 participants, both of which have been balanced with respect to age and gender to eliminate potential confounding and bias. For the data used in the classification task, competitions were classified as both mild cognitive impairment (MCI) and the dementia stage as AD, with matched controls as non-AD. The details of the ADReSS and the ADReSSo Challenge dataset are as follows:Dataset ADReSS: $70\%$ of the ADReSS2020 data were used as the training set, and $30\%$ of the data were used as the test set;Dataset ADReSSo: $70\%$ of the ADReSSo2021 data were used as the training set, and $30\%$ of the data were used as the test set. Table 1 shows the composition and distribution of the datasets we used. ## 2.2.2. Local Dataset A total of 10 audio signals (Chinses) were recorded in this study, including 5 in the AD group, and 5 in the control group (CN). Patients were recruited from the Nanjing Brain Hospital in the Jiangsu Province, China. Cases assessed by experienced neurologists were based on the criteria for the clinical diagnosis of AD, which were established by the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer’s Disease and Related Disorders Association (ADRDA) workgroup in 1984 [33]. The Mini-Mental State Examination (MMSE) were adopted. A professional conducted the MMSE measurements prior to the picture description experiment. The inclusion criteria varied according to the educational level of the patients. The inclusion criteria for illiterate patients were MMSE ≤ 17 and meeting the criteria for NINCDS-ADRDA. The inclusion criteria for primary school patients were MMSE ≤ 20 and meeting the criteria for NINCDS-ADRDA. The inclusion criteria for middle school and above patients were MMSE ≤ 24 and meeting the criteria for NINCDS-ADRDA. The exclusion criteria included: [1] speech impairment; [2] a history of severe stroke, extensive multiple cerebral infarction, critical cerebral infarction or severe white matter lesions; [3] previous psychiatric or psychological disorders; [4] other neurological disorders that cause cognitive impairment, e.g., Lewy body dementia, Parkinson’s disease, hydrocephalus, vascular cognitive; [5] other diseases that can cause dementia, such as severe anemia, thyroid disease, syphilis, HIV infection, etc.; [ 6] combined with serious heart, liver, kidney and other medical diseases, or combined with serious hypertension, diabetes and other complications; [7] unable to cooperate with the completion of cranial MRI or CT scan. The regional review board approved the use of human participants in this study. Patients signed written informed consent forms before participation. The study was approved by the ethics committee of The First Affiliated Hospital with Nanjing Medical University, Nanjing, China, in accordance with the Helsinki Declaration. Referring to the challenges, participants were asked to describe the “Cookie Theft” picture. Speech signals were recorded using the SONY recording pen (ICD-TX660, Nanjing City, China). We collected 10 voice messages ranging in length from 25 s to 183 s. The audio recording information details are summarized in Table 2. ## 2.3. Preprocessing Speech signal preprocessing was divided into three steps: audio format conversion, sample rate normalization, and channel number conversion. Based on the format conversion algorithm, the audio was unified into the ‘.wav’ format at first. Then, to normalize the audio, the audio sampling rate was unified to 44,100 Hz. Finally, we detected the number of channels of the audio, and converted them to mono audio. ## 2.4. Feature Extraction Two kinds of acoustic features used in this study were introduced in detail. One was a pause feature based on the speech signal called VAD Pause, while the other was two commonly used acoustic feature sets, ComParE and eGeMAPS. We extracted the public feature sets for comparison with the VAD Pause, and fused them to propose an ensemble method for detecting AD. ## 2.4.1. VAD Pause Feature Initially, we sectioned a voice recording into audio frames, and then used WebRTC VAD to detect each audio frame. Ultimately, a piece of audio datum was marked as the voiced and non-voiced frame according to the scale of audio frame, and this sequence was regarded as the pause feature of the speech. As shown in Figure 2, we set the frame duration to 0.03 s, such that the original 0.12 s of audio data turned into the VAD Pause feature, which was a sequence with four labels. Zero represents the non-voiced frame, and one represents the voiced frame. The Google WebRTC VAD method was used to identify whether the audio frame was in a voiced or silent state. The detection procedure was as follows: First, according to the correspondence between human speech pronunciation regulation and acoustic frequency, an audio frame was divided into six sub-bands: 80–250 Hz, 250–500 Hz, 500 Hz–1 kHz, 1–2, 2–3, 3–4 kHz. Next, the energy of the six sub-bands was calculated and denoted as E1, E2, E3, E4, E5, E6, and the total energy of the audio frame Et. Then, if the total energy of the audio frame Et was greater than the energy threshold Tm, we proceeded to the following step: For each sub-band, the probability that it belonged to the voiced state was calculated based on the voiced GMM:[2]P(Ei\|H1)=wv112πσv1e(−(Ei−μv1)22σv12)+wv212πσv2e(−(Ei−μv2)22σv22), where Ei ($i = 1$, 2, …, 6) denotes the energy of the sub-band; ws1, ws2 and wv1, wv2 denote the weights for silent and voiced mixture Gaussian distributions, respectively; μs1, μs2, and μv1, μv2 denote the respective means; σs1, σs2, and σv1, σv2 denote the respective standard deviations. [3]Li(Ei)=log2(P(Ei\|H1)P(Ei\|H0))[4]Lt=∑$i = 16$KiLi(Ei), where Ki denotes the weight each sub-band. [5]Fvad={1,Li>Tτ‖Lt>Ta0, else, where Tτ denotes the threshold value of the sub-band log-likelihood ratio; Ta denotes the threshold value of the total log-likelihood ratio. ## 2.4.2. Common Acoustic Feature Sets We used an open-source software package openSMILE toolkit, widely employed for emotion and affect recognition in speech [34], for acoustic feature set extraction of speech fragments. Following is a brief description of the two feature sets constructed in this manner: ComParE: The ComParE 2013 [10] feature set includes energy, spectral, Mel-Frequency Cepstral Coefficients (MFCC), and logarithmic harmonic-to-noise ratio, voice quality features, Viterbi smoothing for F0, spectral harmonicity, and psychoacoustic spectral sharpness. Furthermore, statistical functions were applied to these features, bringing the total to 6373 features for every speech segment. eGeMAPS: The eGeMAPS [11] feature set contains the F0 semitone, jitter, shimmer, loudness, spectral flux, MFCC, F1, F2, F3, alpha ratio, Hammarberg index, and slope V0 features. Statistical functionals were also computed, for a total of 88 features per speech segment. ## 2.5. Ensemble Classification and Voting A novel fusion method was introduced based on the mentioned features and classifiers, as illustrated in Figure 3. Two-step classification experiments were conducted to detect AD. First, the signal was segmented into four s-long sequences after preprocessing and use of the segment-level classification, where classifiers were trained and tested to predict whether a speech segment was uttered by an AD or non-AD patient. Then, we calculated the majority vote from the segment-level results of classifiers, and returned a class label for each subject. The classify model included five classic machine-learning methods, namely: linear discriminant analysis (LDA), decision trees (DT), k-nearest neighbor (KNN), support vector machines (SVM with a linear kernel and a sequential minimal optimization solver), and tree bagger (TB). In order to focus on the effect of speech features and avoid the interference of classifier factors, the traditional machine-learning model, with relatively mature research and strong interpretability, was used as the classifier of this paper. When the validity of the proposed pause feature sequence was determined, such sequence could be directly used as input for training the depth neural network classifier. It could also be further fused with other semantic-based AD detection features to obtain better classification performance. ## 2.6.1. Classification Metrics Depending on whether the predicted label matched with the true label, the outputs of a binary classification algorithm fell into one of the four categories: true positives (TP), false positives (FP), false negatives (FN), and true negatives (TN). The classification metrics are defined as follows:[6]Accuracy=TP+TNTP+TN+FP+FN [7]Precision=TPTP+FP [8]Recall=TPTP+FN [9]F1score=2(Precision)(Recall)Precision+Recall ## 2.6.2. Statistical Analysis To explore whether the feature was truly effective from the perspective of data analysis, the effects of classifiers and features on classification accuracy were tested with a two-way ANOVA (feature (ComParE, eGeMAPS, VAD Pause, ensemble) and classifier (LDA, DT, KNN, SVM, TB) as factors. If there was no interaction between the factors, further multivariate analysis could be performed to analyze the inter-group differences of a single factor, while if the interaction between the factors appeared significant, multiple analysis was further conducted to determine which level of combination of classifiers and features had the highest average classification accuracy. ## 3. Results The results are presented in four parts. First, an intuitive comparison of pause features in AD and non-AD populations is provided. Second, quantitative comparison results of our method are shown in two datasets. Then, a statistical analysis of the classification methods was conducted to demonstrate the validity of the proposed pause feature and the ensemble method. Finally, the VAD Pause features were tested on our own collected Chinese dataset. ## 3.1. Comparison of VAD Pauses in AD and Non-AD Subjects To investigate whether the speech of the AD and controls differed in pauses, we visualized the original speech recordings and VAD Pause features to analyze and confirm the difference from a visual point of view. We selected ten audio clips (half of the speakers were AD subjects, and half were non-AD) of about 1 min length from the test set of the ADReSS2020 dataset. Then, we intercepted the first 24 s of data and extracted their VAD Pause features from them. Figure 4 shows the raw voice waveform (A,B) and VAD Pause features of the voice recordings (C,D). Figure 4 shows some differences in the speech between the AD and non-AD subjects. From the comparison, the subjects with AD had more pauses in every group, and exhibited poor speech coherence compared to healthy controls. The speech recordings of healthy controls contained more voiced frames. Although pauses were present, the speech was more continuous than in the AD group. ## 3.2. Quantitative Results in Classic Machine-Learning Methods We applied the VAD Pause to five machine-learning methods employed previously by [29,30], namely LDA, DT, KNN, SVM, and TB, for AD and non-AD classification. The classification results of all methods on the datasets ADReSS and ADReSSo, are reported in Figure 5 and Figure 6, respectively. For a fair comparison, the ComParE with the best effect used in the literature [29] was selected and tested on the dataset ADReSS, while the eGeMAPS used in the literature [30] was selected and tested on the dataset ADReSSo. All classifiers used the same parameters as those in the literature, and each classification was performed with five runs and shuffling the data order. The average results of the operations are shown in the figures. Figure 5 shows the classification results of the proposed VAD Pause feature and ComParE, using five machine-learning methods on dataset ADReSS. It can be observed that the VAD Pause had a better classification effect and smaller error bars than the ComParE on three machine-learning methods KNN, SVM, and TB. Our VAD Pause with the TB had the best effect, reaching $65.4\%$ accuracy, $67.9\%$ F1 score, $63.4\%$ precision, and $73.3\%$ recall. This indicates better performance than the acoustic feature-based baseline accuracy of $62.5\%$ obtained by [29]. Figure 6 shows the classification results of the proposed VAD Pause feature and eGeMAPS, using five machine-learning methods on the dataset ADReSSo. It can be observed that the VAD Pause feature proposed in this study had a better classification performance and smaller error bars than the eGeMAPS on the same dataset using LDA, KNN, DT, and TB. Our VAD Pause with the TB had the best effect, reaching $65.6\%$ accuracy, $62.3\%$ F1 score, $68.3\%$ precision, and $58.3\%$ recall. Thus, it performed better than the acoustic feature-based baseline accuracy of $64.8\%$ obtained by [30]. Although the VAD Pause gains advantages compared to ComParE and the eGeMAPS, a method that can improve the classification results is required. Thus, we introduced the ensemble procedure. We observed that the TB performs best (in terms of accuracy and stability), and the results are shown in Table 3. Our findings indicate that combining the public feature set (ComParE and eGeMAPS) with the VAD Pause improves the classification results. We further observe the variance from Figure 5 and Figure 6. The proposed ensemble method (dark green) improves the mean and reduces variance overestimates. The classification performance of the presented ensemble method was also compared against other AD detection approaches on the test sets in the literature, shown in Table 3. Our proposed ensemble method achieved the best performance among methods without the deep-learning (DL) model. ## 3.3. Statistical Analysis of Classification Methods To verify whether our method achieves significant improvement, statistical analysis was used to compare the classification results. Figure 7 illustrates the overall effect of different features or the ensemble method, using five machine-learning classifiers on the two datasets. In contrast to the public feature set, the proposed feature and method have a higher average recognition rate, with less recognition dispersion in testing. Thus, the proposed feature and method exhibit better generalization ability. The two-way ANOVA reveals that both classifiers and features contribute to the differences noted in the classification accuracy ($p \leq 0.05$). The interaction between them is also very significant on the datasets ADReSS and ADReSSo ($p \leq 0.05$). Multiple analysis is further conducted to determine the level of combination of classifiers and features that had the highest average classification accuracy, as the interaction between classifiers and features appeared significant on the datasets ADReSS and ADReSSo. The results of the multiple analysis (Figure 8A,B) show that slightly different results are obtained from different datasets. These combinations achieved the best results on both sets of data: the ensemble method with DT, the ensemble method with SVM, the VAD Pause feature with TB, and the ensemble method with TB. The simple effects analyses confirmed that the proposed ensemble method and VAD Pause-based method have significantly higher classification accuracies than the public feature set-based method, which is consistent with our results in Section 3.2. ## 3.4. Experimental Results on a Local Dataset In aiming to test the effectiveness of VAD Pause features in practical applications, we initially collected ten Chinese speech recordings for experimentation. The ADReSS and ADReSSo English datasets were mixed to serve as the training set after the overlapping data were removed. Then, the five classifiers were fed separately for training, and finally the classification results were tested separately on the local Chinese dataset. As shown in Table 4, the VAD Pause feature obtained good classification results on the local Chinese dataset. The classification was basically correct for all AD patients, but there was potential for improvement in the classification of healthy individuals. To further explore the causes of classification errors, the raw waveforms and VAD Pause features of the local Chinese speech dataset are shown in Figure 9. As seen in Figure 9, there were more and longer pauses in the speech of AD patients compared to healthy people in the local dataset, which is consistent with our previous results discussed in Figure 4. It is normal that 1, 3, and 5, which have a relatively high number of pauses in healthy people, have the possibility of being misclassified, which may be related to the existence of pauses in or between utterances in Chinese itself. Additionally, the native language of the participating subjects in the local dataset was Chinese with interspersed dialects, while the public dataset was English. Our proposed features have significant discriminative power on the test subjects of different languages, which shows the advantage of AD recognition based on speech features only. ## 4. Discussion In this paper, we tried to mine the AD symptom information contained in the speech signal itself from the pause perspective, and proposed a VAD-Pause-based method for AD detection that can be applied to easily and conveniently screen for AD in the future. The method was tested for AD classification on the English public datasets and our own local Chinese dataset. We explored the effect of applying the VAD Pause and machine-learning methods, and a novel ensemble method, as well as possible independence/interdependence of their association with the AD classification. The results confirm that AD subjects use more pauses than healthy controls, and that the VAD Pause and ensemble method have higher classification values and better generalization ability than public acoustic feature sets. The effectiveness and generalization ability of the VAD Pause for AD detection constitute the most intriguing results. VAD Pause performance can be attributed to the difference in pauses in speech between the two groups. Previous studies reported that in speech production, disfluencies, such as hesitations and speech errors, are correlated with cognitive functions, such as cognitive load, arousal, and working memory [45,46]. Semantic verbal fluency and phonological verbal fluency tests are widely used in the diagnosis of AD, and they are reliable indicators of language deterioration in the early detection of AD [47]. Another study suggested that AD patients require more effort to speak than healthy individuals: namely, patients speak more slowly with longer pauses [48]. Yuan et al. [ 7] studied the function of pauses from the perspective of manual transcription, and concluded that AD subjects used more and longer pauses than healthy people. All of these results seem to suggest that pauses have a potential status in the distinction between the AD and non-AD. Furthermore, the VAD Pause may be a valid feature that represents pauses well. Figure 4 and Figure 9 show that generalizing the difference between the two for original recordings, which hold a significant amount of information, is not a trivial task, whereas the VAD Pause feature proposed in this study shows the difference visually. Notably, our result is representative of the performance of AD detection when using only audio recordings, without transcription. Several reasons account for the advantages of this practice. First, it is more practical to detect AD using speech directly. When a physician conducts a clinical interview, they usually evaluate the patient directly based on their voice, rather than transcribing the words. When the program detects AD directly via speech, it takes less time than when needing to transcribe it into text. Moreover, in this study, we use pauses as detection features, while transcription had an effect on the detection of pauses. A prior study showed that transcription errors impacted findings related to the usefulness of prosodic features in parsing [49]. Several state-of-the-art approaches rely on this manually generated text for feature extraction, and their performance may vary depending on whether the transcription is automated [3]. If we use speech directly for feature extraction and AD detection, the influence of transcription can be avoided, and the error generated in the process may be reduced. Comparisons with other methods using the same data attract particular interest. Table 3 shows that methods using acoustic features for AD detection can be divided into two categories: those that use DL and those that do not. Our ensemble method achieves the best results among methods that do not rely on DL [29,30,39,44]. In the dataset ADReSS, the pre-trained VGGish model and Uni-CRNN are used in the method proposed by [35]. An accuracy of $72.9\%$ was achieved with this method. In addition, among the DL-related approaches, some studies use neural networks as classifiers, while others focus on using them to extract acoustic embeddings. Cummins et al. [ 36] and Rohanian et al. [ 37] combined public acoustic features with neural networks and achieved an accuracy of 70.8 and $66.6\%$, respectively. Acoustic embeddings as speech features started to attract the attention of numerous researchers, and have gained good performance in AD detection [38,40,41,42,43]. There appears to be a trade-off in accuracy and convenience. Methods using DL are more accurate, but methods without DL are more convenient. A better approach could involve using VAD Pause and acoustic embeddings to represent speech information. Due to time cost and practicality, we will further explore this idea in future. We showed that both the features and classifiers used in this study contribute to the classification performance, as the results of the two-way ANOVA were significant in all two datasets. Moreover, in the datasets ADReSS and ADReSSo, the factor contribution to the highest average classification accuracy was intertwined. Statistically, the ensemble method with DT, the ensemble method with SVM, the VAD Pause feature with TB, and the ensemble method with TB have similarly high accuracies. Based on the analysis of these results, we recommend the ensemble method with TB, as it achieves stable and high accuracy, while several other methods are likewise available (ensemble method with DT, ensemble method with SVM, VAD Pause feature with TB). Additionally, we did not explore the effect of using DL networks for classification. Yet, usually DL classification requires a significant amount of data and time, such that it is necessary to optimize and improve the algorithm, which will be a future direction of our study. Silent pauses have been implemented to other populations beyond AD. Some researchers have investigated the use of silent pause features for disease detection. Mignard et al. [ 50] investigated fluency disorders in Parkinson’s patients by the pause ratio. Potagas et al. [ 51] used speech rate, articulation rate, pause frequency, and pause duration as analytical indicators, and the results showed that silent pauses can be used as complementary biomarkers for PPA. Imre et al. [ 52] conducted a study on the temporal speech characteristics of elderly patients with type 2 diabetes (T2DM). The healthy cognition participants in the T2DM group showed higher duration rate of silent pause and total pause, and a higher average duration of silent pauses and total pauses compared to the group without T2DM group. These methods were mainly carried out based on the calculation of the frequency and duration of silent pauses, etc. Compared to previous publications, the method we propose to encode speech into a sequence of pauses can characterize the temporal sequence of pauses in speech more accurately, while the data processing is simpler and easy to implement and repeat. Spontaneous speech analysis plays an important role in the study of acquired language disorders. Ditthapron et al. [ 27] used smartphones to passively capture changes in acoustic characteristics of spontaneous speech for continuous traumatic brain injury monitoring. Spontaneous speech can also be used for research on depression [53] and aphasia [28,54]. Thus, our proposed spontaneous speech-based approach has the potential to be used in other clinical populations with acquired language disorders. In future work, we consider investigating the feasibility of applying this method to other populations. ## 5. Conclusions We proposed a pause/non-pause feature sequence (VAD Pause) encoded using only speech, and investigated its effectiveness when applied to distinguish AD patients from healthy subjects. Its classification effect was tested on both public datasets and a local dataset. We further introduced an ensemble method for AD classification from spontaneous speech and investigated the impact of features and classifiers on the results in detail, further demonstrating the superiority of the ensemble method through a comprehensive comparison of classification results of different datasets (two English datasets and one Chinese dataset). The results of the study suggest that our method can be partly immune to AD detection errors due to the language environment, which is of greater value for the widespread screening of AD. In future work, the proposed AD screening method will be considered extended to the detection of other acquired speech disorders, with appropriate modifications. We will also collect more local data to further demonstrate the reliability and potential of our proposed method. ## References 1. Derby C.A.. Trends in the public health significance, definitions of disease, and implications for prevention of Alzheimer’s disease. Curr. Epidemiol. Rep. (2020.0) 7 68-76. DOI: 10.1007/s40471-020-00231-8 2. 2. Alzheimer’s Disease International World Alzheimer Report 2019: Attitudes to DementiaAlzheimer’s Disease InternationalLondon, UK2019. World Alzheimer Report 2019: Attitudes to Dementia (2019.0) 3. Mahajan P., Baths V.. Acoustic and language based deep learning approaches for Alzheimer’s dementia detection from spontaneous speech. Front. Aging Neurosci. (2021.0) 13 623607. DOI: 10.3389/fnagi.2021.623607 4. Mueller K.D., Koscik R.L., Hermann B.P., Johnson S.C., Turkstra L.S.. Declines in connected language are associated with very early mild cognitive impairment: Results from the Wisconsin registry for Alzheimer’s prevention. Front. Aging Neurosci. (2018.0) 9 437. DOI: 10.3389/fnagi.2017.00437 5. Mesulam M., Wicklund A., Johnson N., Rogalski E., Léger G.C., Rademaker A., Weintraub S., Bigio E.H.. Alzheimer and frontotemporal pathology in subsets of primary progressive aphasia. Ann. Neurol. (2008.0) 63 709-719. DOI: 10.1002/ana.21388 6. Meghanani A., Anoop C., Ramakrishnan A.G.. Recognition of alzheimer’s dementia from the transcriptions of spontaneous speech using fasttext and cnn models. Front. Comput. Sci. (2021.0) 3 624558. DOI: 10.3389/fcomp.2021.624558 7. Yuan J., Cai X., Bian Y., Ye Z., Church K.. Pauses for detection of Alzheimer’s disease. Front. Comput. Sci. (2021.0) 2 624488. DOI: 10.3389/fcomp.2020.624488 8. Agbavor F., Liang H.. Artificial Intelligence-Enabled End-To-End Detection and Assessment of Alzheimer’s Disease Using Voice. Brain Sci. (2023.0) 13. DOI: 10.3390/brainsci13010028 9. Luz S.. Longitudinal monitoring and detection of Alzheimer’s type dementia from spontaneous speech data. Proceedings of the 2017 IEEE 30th International Symposium on Computer-Based Medical Systems (CBMS) 45-46 10. Eyben F., Weninger F., Gross F., Schuller B.. Recent developments in opensmile, the munich open-source multimedia feature extractor. Proceedings of the 21st ACM International Conference on Multimedia 835-838 11. Eyben F., Scherer K.R., Schuller B.W., Sundberg J., André E., Busso C., Devillers L.Y., Epps J., Laukka P., Narayanan S.S.. The Geneva minimalistic acoustic parameter set (GeMAPS) for voice research and affective computing. IEEE Trans. Affect. Comput. (2015.0) 7 190-202. DOI: 10.1109/TAFFC.2015.2457417 12. Nasrolahzadeh M., Rahnamayan S., Haddadnia J.. Alzheimer’s disease diagnosis using genetic programming based on higher order spectra features. Mach. Learn. Appl. (2022.0) 7 100225. DOI: 10.1016/j.mlwa.2021.100225 13. Lopez-de-Ipiña K., Alonso-Hernández J., Solé-Casals J., Travieso-González C.M., Ezeiza A., Faundez-Zanuy M., Calvo P.M., Beitia B.. Feature selection for automatic analysis of emotional response based on nonlinear speech modeling suitable for diagnosis of Alzheimer’s disease. Neurocomputing (2015.0) 150 392-401. DOI: 10.1016/j.neucom.2014.05.083 14. Nasrolahzadeh M., Haddadnia J., Rahnamayan S.. Multi-objective optimization of wavelet-packet-based features in pathological diagnosis of alzheimer using spontaneous speech signals. IEEE Access (2020.0) 8 112393-112406. DOI: 10.1109/ACCESS.2020.3001426 15. Ash S., Moore P., Vesely L., Gunawardena D., McMillan C., Anderson C., Avants B., Grossman M.. Non-fluent speech in frontotemporal lobar degeneration. J. Neurolinguist. (2009.0) 22 370-383. DOI: 10.1016/j.jneuroling.2008.12.001 16. Ash S., Xie S.X., Gross R.G., Dreyfuss M., Boller A., Camp E., Morgan B., O’Shea J., Grossman M.. The organization and anatomy of narrative comprehension and expression in Lewy body spectrum disorders. Neuropsychology (2012.0) 26 368. DOI: 10.1037/a0027115 17. Wilson S.M., Henry M.L., Besbris M., Ogar J.M., Dronkers N.F., Jarrold W., Miller B.L., Gorno-Tempini M.L.. Connected speech production in three variants of primary progressive aphasia. Brain (2010.0) 133 2069-2088. DOI: 10.1093/brain/awq129 18. Lindsay H., Tröger J., König A.. Language impairment in alzheimer’s disease—Robust and explainable evidence for ad-related deterioration of spontaneous speech through multilingual machine learning. Front. Aging Neurosci. (2021.0) 228 642033. DOI: 10.3389/fnagi.2021.642033 19. Pistono A., Jucla M., Barbeau E.J., Saint-Aubert L., Lemesle B., Calvet B., Köpke B., Puel M., Pariente J.. Pauses during autobiographical discourse reflect episodic memory processes in early Alzheimer’s disease. J. Alzheimer’s Dis. (2016.0) 50 687-698. DOI: 10.3233/JAD-150408 20. Yuan J., Xu X., Lai W., Liberman M.. Pauses and pause fillers in Mandarin monologue speech: The effects of sex and proficiency. Proceedings of the Speech Prosody 2016 1167-1170 21. Shea C., Leonard K.. Evaluating measures of pausing for second language fluency research. Can. Mod. Lang. Rev. (2019.0) 75 216-235. DOI: 10.3138/cmlr.2018-0258 22. Ogata J., Goto M., Itou K.. The use of acoustically detected filled and silent pauses in spontaneous speech recognition. Proceedings of the 2009 IEEE International Conference on Acoustics, Speech and Signal Processing 4305-4308 23. Vincze V., Szatlóczki G., Tóth L., Gosztolya G., Pákáski M., Hoffmann I., Kálmán J.. Telltale silence: Temporal speech parameters discriminate between prodromal dementia and mild Alzheimer’s disease. Clin. Linguist. Phon. (2021.0) 35 727-742. DOI: 10.1080/02699206.2020.1827043 24. Pistono A., Pariente J., Bézy C., Lemesle B., Le Men J., Jucla M.. What happens when nothing happens? An investigation of pauses as a compensatory mechanism in early Alzheimer’s disease. Neuropsychologia (2019.0) 124 133-143. DOI: 10.1016/j.neuropsychologia.2018.12.018 25. Pastoriza-Domínguez P., Torre I.G., Diéguez-Vide F., Gomez-Ruiz I., Gelado S., Bello-López J., Ávila-Rivera A., Matias-Guiu J.A., Pytel V., Hernández-Fernández A.. Speech pause distribution as an early marker for Alzheimer’s disease. Speech Commun. (2022.0) 136 107-117. DOI: 10.1016/j.specom.2021.11.009 26. Gayraud F., Lee H.-R., Barkat-Defradas M.. Syntactic and lexical context of pauses and hesitations in the discourse of Alzheimer patients and healthy elderly subjects. Clin. Linguist. Phon. (2011.0) 25 198-209. DOI: 10.3109/02699206.2010.521612 27. Ditthapron A., Lammert A.C., Agu E.O.. Continuous TBI Monitoring From Spontaneous Speech Using Parametrized Sinc Filters and a Cascading GRU. IEEE J. Biomed. Health Inform. (2022.0) 26 3517-3528. DOI: 10.1109/JBHI.2022.3158840 28. Lfab C., Abb C., Lba D., Jfb C.. Speech timing changes accompany speech entrainment in aphasia—ScienceDirect. J. Commun. Disord. (2021.0) 90 106090. PMID: 33611108 29. Luz S., Haider F., de la Fuente S., Fromm D., MacWhinney B.. Alzheimer’s dementia recognition through spontaneous speech: The ADReSS challenge. arXiv (2020.0). DOI: 10.48550/arXiv.2004.06833 30. Luz S., Haider F., de la Fuente S., Fromm D., MacWhinney B.. Detecting cognitive decline using speech only: The ADReSSo Challenge. arXiv (2021.0). DOI: 10.48550/arXiv.2104.09356 31. Becker J.T., Boiler F., Lopez O.L., Saxton J., McGonigle K.L.. The natural history of Alzheimer’s disease: Description of study cohort and accuracy of diagnosis. Arch. Neurol. (1994.0) 51 585-594. DOI: 10.1001/archneur.1994.00540180063015 32. Goodglass H., Kaplan E., Weintraub S.. BDAE: The Boston Diagnostic Aphasia Examination (2001.0) 33. Jack Jr C.R., Albert M., Knopman D.S., McKhann G.M., Sperling R.A., Carillo M., Thies W., Phelps C.H.. Introduction to revised criteria for the diagnosis of Alzheimer’s disease: National Institute on Aging and the Alzheimer Association Workgroups. Alzheimer’s Dement. J. Alzheimer’s Assoc. (2011.0) 7 257. DOI: 10.1016/j.jalz.2011.03.004 34. Eyben F., Wöllmer M., Schuller B.. Opensmile: The munich versatile and fast open-source audio feature extractor. Proceedings of the 18th ACM International Conference on Multimedia (2010.0) 1459-1462 35. Koo J., Lee J.H., Pyo J., Jo Y., Lee K.. Exploiting multi-modal features from pre-trained networks for Alzheimer’s dementia recognition. arXiv (2020.0). DOI: 10.21437/Interspeech.2020-3153 36. Cummins N., Pan Y., Ren Z., Fritsch J., Nallanthighal V.S., Christensen H., Blackburn D., Schuller B.W., Magimai-Doss M., Strik H.. A comparison of acoustic and linguistics methodologies for Alzheimer’s dementia recognition. Proceedings of the Interspeech 2020 2182-2186 37. Rohanian M., Hough J., Purver M.. Multi-modal fusion with gating using audio, lexical and disfluency features for Alzheimer’s dementia recognition from spontaneous speech. arXiv (2021.0) 38. Pappagari R., Cho J., Moro-Velazquez L., Dehak N.. Using State of the Art Speaker Recognition and Natural Language Processing Technologies to Detect Alzheimer’s Disease and Assess its Severity. Proceedings of the Interspeech 2020 2177-2181 39. Edwards E., Dognin C., Bollepalli B., Singh M.K., Analytics V.. Multiscale System for Alzheimer’s Dementia Recognition Through Spontaneous Speech. Proceedings of the Interspeech 2020 2197-2201 40. Balagopalan A., Novikova J.. Comparing Acoustic-based Approaches for Alzheimer’s Disease Detection. arXiv (2021.0) 41. Pan Y., Mirheidari B., Harris J.M., Thompson J.C., Jones M., Snowden J.S., Blackburn D., Christensen H.. Using the Outputs of Different Automatic Speech Recognition Paradigms for Acoustic-and BERT-Based Alzheimer’s Dementia Detection Through Spontaneous Speech. Proceedings of the Interspeech 2021 3810-3814 42. Pérez-Toro P.A., Bayerl S.P., Arias-Vergara T., Vásquez-Correa J.C., Klumpp P., Schuster M., Nöth E., Orozco-Arroyave J.R., Riedhammer K.. Influence of the Interviewer on the Automatic Assessment of Alzheimer’s Disease in the Context of the ADReSSo Challenge. Proceedings of the Interspeech 2021 3785-3789 43. Pappagari R., Cho J., Joshi S., Moro-Velázquez L., Zelasko P., Villalba J., Dehak N.. Automatic Detection and Assessment of Alzheimer Disease Using Speech and Language Technologies in Low-Resource Scenarios. Proceedings of the Interspeech 2021 3825-3829 44. Chen J., Ye J., Tang F., Zhou J.. Automatic detection of Alzheimer’s disease using spontaneous speech only. Proceedings of the Interspeech 2021 3830 45. Daneman M.. Working memory as a predictor of verbal fluency. J. Psycholinguist. Res. (1991.0) 20 445-464. DOI: 10.1007/BF01067637 46. Arciuli J., Mallard D., Villar G.. “Um, I can tell you’re lying”: Linguistic markers of deception versus truth-telling in speech. Appl. Psycholinguist. (2010.0) 31 397-411. DOI: 10.1017/S0142716410000044 47. Laws K.R., Duncan A., Gale T.M.. ‘Normal’semantic–phonemic fluency discrepancy in Alzheimer’s disease? A meta-analytic study. Cortex (2010.0) 46 595-601. DOI: 10.1016/j.cortex.2009.04.009 48. López-de-Ipiña K., Alonso J.-B., Travieso C.M., Solé-Casals J., Egiraun H., Faundez-Zanuy M., Ezeiza A., Barroso N., Ecay-Torres M., Martinez-Lage P.. On the selection of non-invasive methods based on speech analysis oriented to automatic Alzheimer disease diagnosis. Sensors (2013.0) 13 6730-6745. DOI: 10.3390/s130506730 49. Tran T., Toshniwal S., Bansal M., Gimpel K., Livescu K., Ostendorf M.. Parsing speech: A neural approach to integrating lexical and acoustic-prosodic information. arXiv (2017.0). DOI: 10.48550/arXiv.1704.07287 50. Mignard P., Cave C., Lagrue B., Meynadier Y., Viallet F.. Silent pauses in Parkinsonian patients during spontaneous speech and reading: An instrumental study. Rev. De Neuropsychol. (2001.0) 11 39-63 51. Potagas C., Nikitopoulou Z., Angelopoulou G., Kasselimis D., Laskaris N., Kourtidou E., Constantinides V.C., Bougea A., Paraskevas G.P., Papageorgiou G.. Silent Pauses and Speech Indices as Biomarkers for Primary Progressive Aphasia. Medicina (2022.0) 58. DOI: 10.3390/medicina58101352 52. Imre N., Balogh R., Gosztolya G., Tóth L., Hoffmann I., Várkonyi T., Lengyel C., Pákáski M., Kálmán J.. Temporal Speech Parameters Indicate Early Cognitive Decline in Elderly Patients With Type 2 Diabetes Mellitus. Alzheimer Dis. Assoc. Disord. (2022.0) 36 148. DOI: 10.1097/WAD.0000000000000492 53. Lu X., Shi D., Liu Y., Yuan J.. Speech depression recognition based on attentional residual network. Front. Biosci.-Landmark (2021.0) 26 1746-1759. DOI: 10.52586/5066 54. Le D., Licata K., Provost E.M.. Automatic Quantitative Analysis of Spontaneous Aphasic Speech. Speech Commun. (2018.0) 100 1-12. DOI: 10.1016/j.specom.2018.04.001
--- title: Osseointegration of Titanium Implants in a Botox-Induced Muscle Paralysis Rat Model Is Sensitive to Surface Topography and Semaphorin 3A Treatment authors: - Jingyao Deng - D. Joshua Cohen - Michael B. Berger - Eleanor L. Sabalewski - Michael J. McClure - Barbara D. Boyan - Zvi Schwartz journal: Biomimetics year: 2023 pmcid: PMC10046785 doi: 10.3390/biomimetics8010093 license: CC BY 4.0 --- # Osseointegration of Titanium Implants in a Botox-Induced Muscle Paralysis Rat Model Is Sensitive to Surface Topography and Semaphorin 3A Treatment ## Abstract Reduced skeletal loading associated with many conditions, such as neuromuscular injuries, can lead to bone fragility and may threaten the success of implant therapy. Our group has developed a botulinum toxin A (botox) injection model to imitate disease-reduced skeletal loading and reported that botox dramatically impaired the bone formation and osseointegration of titanium implants. Semaphorin 3A (sema3A) is an osteoprotective factor that increases bone formation and inhibits bone resorption, indicating its potential therapeutic role in improving osseointegration in vivo. We first evaluated the sema3A effect on whole bone morphology following botox injections by delivering sema3A via injection. We then evaluated the sema3A effect on the osseointegration of titanium implants with two different surface topographies by delivering sema3A to cortical bone defect sites prepared for implant insertion and above the implants after insertion using a copper-free click hydrogel that polymerizes rapidly in situ. Implants had hydrophobic smooth surfaces (PT) or multiscale biomimetic micro/nano topography (SLAnano). Sema3A rescued the botox-impaired bone formation. Furthermore, biomimetic Ti implants improved the bone-to-implant contact (BIC) and mechanical properties of the integrated bone in the botox-treated rats, which sema3A enhanced. This study demonstrated the value of biomimetic approaches combining multiscale topography and biologics in improving the clinical outcomes of implant therapy. ## 1. Introduction The success rate of dental, spinal, or orthopedic implants has always been challenging for compromised patients, such as the elderly, smokers, diabetics, and osteoporosis [1,2]. To increase implant success, research on implants with modifications such as complex micro-/nano-topography, wettability, and altered surface chemistry has been ongoing for many years [3,4,5]. Titanium and its alloys are the most common materials for implant therapy in the bone due to their superior corrosion resistance, easy processing, mechanical support properties, and biocompatibility [6,7]. The integration of an implant with surrounding bone is a highly structured biological process consisting of protein adsorption, immune cell modulation, mesenchymal cell recruitment and osteoblast differentiation, primary bone formation and vascularization, bone remodeling, and mature bone formation [4]. This process, termed osseointegration, is defined operationally as the degree of bone-to-implant contact (BIC) [4]. Many studies aim to create implant surfaces that mimic the physical structure of bone surfaces following resorption by osteoclasts [8,9]. The unique structure of the resorbed bone surface during bone remodeling, including micro-scale, submicron-scale, and nano-scale features, favors subsequent bone formation, and implant surfaces that possess these features exhibit enhanced osseointegration and long-term implant success [4,5,10,11]. This has been demonstrated in rodent models with compromised bone quality, including osteoporosis and diabetes [12,13,14,15]. Peri-implant bone formation depends on the ability of bone marrow stromal cells (MSCs) to colonize implant surfaces and differentiate into osteoblasts [4]. Osteoprogenitor cells exhibit osteoblast differentiation markers and produce local factors in a well-orchestrated manner when responding to the biomimetic implant surface properties that are characterized by physical features similar to the surface of osteoclast-resorbed bone [16]. This results in a microenvironment favoring osteogenesis [17,18], anti-inflammatory tissue regeneration [19,20,21], and new blood vessel formation [22,23], as well as modulating osteoclast resorptive activities [24]. This osteogenic microenvironment is dynamic and underlying mechanisms are not fully understood. However, studies using animal models indicate that supplementing the normal production of various local factors using an exogenous biologic approach can aid or enhance the surface-mediated osteogenic microenvironment [25,26]. One indicator of the effect of the osteogenic surface on MSC differentiation is the upregulation of osteocalcin (OCN) [3], a marker of a well-differentiated osteoblast. This is partly due to the production of the osteoinductive protein, bone morphogenetic protein 2 (BMP2) [27]. In addition to BMP2, MSCs respond to biomimetic Ti surface features with the increased production of a nerve-derived factor, semaphorin 3A (sema3A) [28]. Sema3A, originally identified as an axon guidance molecule, is osteoprotective [29,30,31,32]. It can increase osteoblastic differentiation while regulating osteoclast resorptive activities [32]. The treatment of MSCs cultured on Ti disks with a grit-blasted, acid-etched surface with sema3A further enhanced the effect of the biomimetic topography on osteoblastic differentiation, including the production of OCN and osteoprotegerin (OPG) [28]. These observations suggest that sema3A may function similarly in vivo. Several studies have shown that Ti implants with a biomimetic surface topography result in improved osseointegration in animal models that exhibit a compromised bone quality compared to implants with a smooth topography created by machining [33,34,35,36]. Most of these studies have used rats or mice with an induced diabetic phenotype or an osteoporotic phenotype resulting from ovariectomy [12,13,33,34,35,36]. A compromised bone quality can also result from disuse or mechanical unloading situations, which can be caused by long-term bed rest, neuromuscular injuries, space flight, or spinal cord injuries [15,37,38,39]. To assess whether Ti implant surface features would improve osseointegration under conditions where the bone was mechanically unloaded, we established a rat model using the neurotoxin botulinum toxin A (botox), which inhibits the release of acetylcholine from the neuromuscular junctions, causing paralysis. We found that osseointegration was drastically inhibited in the absence of mechanical stress [40]. We previously showed that the local delivery of human recombinant sema3A improved the osseointegration of Ti implants with a biomimetic surface topography placed transcortically in diabetic rat femurs, which have a similar bone phenotype to mechanically unloaded bone with respect to the loss of trabeculae in the metaphysis [40,41]. This suggested that sema3A might also improve implant integration in mechanically unloaded bone. Accordingly, we used the botox injection model to evaluate the effectiveness of biomimetic surface topography in the absence of mechanical loading and to assess if the local delivery of the nerve-derived factor sema3A would improve osseointegration to the level found in healthy animals. The study determined whether sema3A treatment is sufficient to stimulate osseointegration with a smooth Ti implant to the levels observed when using a Ti surface with a multiscale biomimetic topography. We then assessed if the addition of sema3A is able to enhance the osseointegration of a Ti implant with a biomimetic surface in the botox-compromised model. ## 2.1. Implant Manufacturing Titanium Implants machined from grade 4 titanium rods to be 2.5 mm in diameter, 3.5 mm in length, and 0.8 mm in pitch were customized to fit in a rat femur by Institut Straumann AG (Basel, Switzerland). The machined implants were designated “pre-treatment” (PT). The PT implants were blasted with 250–500 μm Al2OH3 grit and acid-etched in a mixture of HCl and H2SO4, resulting in a complex microrough topography (SLA), and then processed under nitrogen and stored in $0.9\%$ sterile saline, resulting in a hydrophilic surface that had nanoscale features hydrophilic (modSLA). PT and modSLA implants were sterilized using γ irradiation. In order to obtain hydrophobic implants that had both the SLA microroughness and the added nanoscale features found on the modSLA surfaces, the modSLA implants were removed from the sterile saline package in a biological safety cabinet under sterile conditions and aged for at least 1 month to generate the SLAnano surfaces, which were repackaged in aluminum foil. The physical and chemical properties of the PT and SLAnano surfaces have been described in detail [42]. We previously reported that the osseointegration of modSLA implants, which have a hydrophilic surface, is impaired in the botox-compromised rat model [40]. In the present study, our goal was to focus on surface topography without the confounding variable of wettability, and the PT and SLAnano surfaces are both hydrophobic. Thus, the present experimental design enabled us to focus on the contribution of biomimetic multiscale topography on osseointegration in this mechanical unloading model and to assess the contribution of sema3A to the process. ## 2.2. Hydrogel Preparation We used a copper-free click hydrogel as the sema3A delivery vehicle. We have successfully used click hydrogels to deliver biologics, drugs, and antibodies to bone defect sites, with no evidence of toxicity [41,43,44]. Swelling is minimal, and following an initial burst release, the payload is released at a steady rate, with degradation occurring over a 14-day period [41,43,44]. The techniques used to prepare the hydrogel were adapted from previous work [41,43,44]. Briefly, the copper-free click-based chemistry was used to combine two aqueous solutions that underwent in situ chemical crosslinking to create the quickly polymerizing hydrogel. A thiol-Michel addition reaction involving PEG-dithiol and dibenzocyclooctyne maleimide (DBCO-maleimide) was used to create a poly-ethylene glycol (PEG) crosslinker that has been functionalized with dibenzocyclooctyne (DBCO). When combined with an azide-functionalized acylate polymer, the DBCO-functionalized precursor created an in situ crosslinked hydrogel. Reversible addition-fragmentation chain transfer (RAFT) polymerization, which allows for the precise control of azide functionality, was used to create PEG-N3 from azide functionalized and non-functionalized PEG methacrylate monomers to produce a water-soluble, non-fouling multivalent azide functionalized polymer. The components were synthesized at a commercial facility under Good Laboratory Practice controls (Syngene International Limited, Bangalore, India) according to our requirements and shipped lyophilized to our laboratory. Before use, the components were stored at −80 °C after being reconstituted in sterile 1X PBS (ThermoFisher Scientific, Waltham, MA, USA). The hydrogels were formed by combining PEG-N3 ($50\%$; w:v) and PEG-DBCO ($12.5\%$; w:v) at a 1:2 (v/v) ratio. ## 2.3. Animals and Surgical Procedures The Institutional Animal Care and Use Committee at Virginia Commonwealth University approved all the animal procedures. The National Institutes of Health’s guide for the care and use of laboratory animals was followed in all experiments. Animals were kept in an AALAC-approved animal facility in indoor housing with a 12h/12h light/dark cycle and individually ventilated, solid-bottomed polysulfone cages that allow for temperature and humidity adjustment within ranges appropriate for the animals. For all animal procedures, $5\%$ isoflurane gas with O2 was used to induce anesthesia and kept at $2.5\%$ after. Animals recovered consciousness on a water-circulating warming pad before returning to the vivarium. For surgical procedures, 1 mg/kg of sustained-release buprenorphine (ZooPharm, Windsor, CO, USA) was administered pre-operatively and subcutaneously to provide a minimum of 72 h of postoperative analgesia. ## 2.3.1. Therapeutic Effect of Sema3a on the Botox-Induced Compromised Bone Phenotype In total, 29 male Sprague Dawley rats (SD) weighing 300–325 g (Charles River Laboratories, Wilmington, MA, USA) were randomly divided into 4 groups: control (veh, $$n = 4$$), control with sema3A injections (veh+sema3A, $$n = 8$$), botox injections (BTX, $$n = 8$$), and botox injections with sema3A injections (BTX+sema3A, $$n = 9$$). One extra rat was prepared and randomly assigned to the BTX+sema3A group. Botulinum toxin type A (onabotulinumtoxinA; BOTOX®, Allergan, Inc. Irvine, CA, USA [botox]) was dissolved in $0.9\%$ saline (10 units/mL) [40]. On day 1 and day 25, for the BTX and BTX+sema3A groups, the right hindlimbs were injected intramuscularly with a total of 8 units of botox distributed as 2 units into the following locations: paraspinal muscles, quadriceps, the hamstrings, and the calf muscles. The contralateral legs were the internal controls (Figure 1a). On day 21 and day 28, recombinant human sema3A was reconstituted in $0.9\%$ sterile saline (100 μg/mL, R&D Systems, Minneapolis, MN, USA), and 6 μg of sema3A (100 μg/mL in 60 μL) were injected into the periosteal layer of the distal end of the third trochanter on the right femurs for veh+sema3A and BTX+sema3A groups, and the same amount of $0.9\%$ sterile saline was injected to the rest of the groups (Figure 1a). On day 38, the rats were humanely euthanized by CO2 inhalation and cervical dislocation. Femurs were harvested in 1XPBS and further evaluated by microCT and 3-point bending fracture analysis, which is described in Section 2.4.2. We opted to use recombinant human sema3A instead of recombinant rat sema3A for two reasons. We knew that human sema3A could enhance surface-mediated osteoblast differentiation of human MSCs in vitro, as well as the production of factors associated with osteogenesis. Moreover, human sema3A restored the osseointegration of Ti implants in a type 2 diabetic rat model to normal levels, demonstrating that it was bioactive in vivo [28,41]. Therefore, the concentrations of the same human recombinant sema3A were adopted for this study. ## 2.3.2. Effect of Surface Topography on Response to Sema3A In total, 49 male SD rats weighing 300–325 g (Charles River) were randomly divided into 6 groups: control rats with PT implants (Control+PT, $$n = 8$$), control rats with SLAnano implants (Control+SLAnano, $$n = 8$$), botox-injected rats with PT implants (BTX+PT, $$n = 8$$), botox-injected rats with PT implants and sema3A injections (BTX+PT+sema3A, $$n = 8$$), botox-injected rats with SLAnano implants (BTX+SLAnano, $$n = 8$$), and botox-injected rats with SLAnano implants and sema3A injections (BTX+SLAnano+sema3A, $$n = 8$$). One animal from the BTX+SLAnano group was withdrawn from the study as it met the humane endpoint; thus, BTX+SLAnano had $$n = 7$$ for the subsequent tissue analysis. On day 1 and day 28, the same dose of botox was injected into the same muscle groups to BTX+PT, BTX+PT+sema3A, BTX+SLAnano, and BTX+SLAnano+sema3A right hindlimbs. On day 21, all rats were prepared for implant insertion and hydrogel loading surgeries by shaving and cleaning the right hindlimbs with $70\%$ ethanol and $2\%$ chlorhexidine. The implant insertion sites were produced by sequentially drilling a defect with increasing diameter dental drill bits (Ø1.0 mm, Ø1.6 mm, Ø2.0 mm, and Ø2.2 mm) to a depth of 3.5 mm in the distal metaphysis of the femur after separating the adjacent muscles and periosteum. Recombinant human sema3A (R&D Systems) was reconstituted with the PEG-DBCO crosslinker solution. The hydrogels were formed by combining 5.33 μL of PEG-N3 and 10.66 μL of PEG-DBCO with or without 6 μg of sema3A to the designated groups using separate pipettors to pipette two components into the holes simultaneously. This resulted in 6 μg of sema3A/hydrogel. Threaded PT implants or SLA implants were inserted into the holes after gelation. Cover screws were added to cap the implants. Then, the hydrogels were delivered on top of the implants again with or without sema3A with the same techniques (Figure 1b). Rats were recovered from anesthesia on a water-circulating warming pad and weighed weekly. On day 49, all rats were humanely euthanized, and femurs were harvested in 1X PBS for further analysis. ## 2.4.1. Micro-Computed Tomography Femurs were isolated and prepared for microCT scanning (SkyScan 1173, Bruker, Kontich, Belgium) within 24 h of harvest without fixation to evaluate the bone phenotype and peri-implant bone growth. For the first study, to evaluate the effect of sema3A on bone morphology, both distal and proximal ends of the femurs were scanned at a resolution of 1120 × 1120 pixels (isotropic voxel size of 15.82 μm) using a 1.0 mm aluminum filter, at an exposure of 250 ms, with scanning energies of 85 kV and 94 μA [40]. A standard Feldkamp reconstruction was conducted by NRecon Software (Bruker) with a beam hardening correction of $20\%$, and no smoothing was applied. The quantitative trabecular morphometric parameters, including bone volume/total volume (BV/TV), trabecular number, trabecular thickness, and total porosity, were evaluated for the first animal study. The quantitative cortical morphometric parameters were determined, including BV/TV, the total porosity, and cortical thickness. For the second animal study, bone-to-implant contact (BIC) was evaluated for each implant by scanning the metaphysis of the distal femurs using a 0.25 mm brass filter, at an exposure of 420 ms, with scanning energies of 120 kV and 66 μA. After reconstruction, total BIC, BIC in the marrow space, and cortical bone BIC were determined using previously described methods [40]. ## 2.4.2. Mechanical Testing In the first experiment, a 3-point bending study was performed using a BOSE ElectroForce 3200 Series III axis (TA Instruments, New Castle, DE, USA). Bones were positioned so that the femur’s sema3A injection site was in the middle of two support struts that were facing up. The load cell was attached using a triangular prism-pointed testing mount. An axial compressive displacement rate of 0.1 mm per second was used to achieve axial displacement until failure. In the second experiment, mechanical torque to failure was used to determine the overall implant mechanical integrity using a Bose ElectroForce 3200 Series III Axial-Torsion mechanical testing system equipped with a 445 N/5.7 N m load/torque transducer. Before loading the samples, the load cell was zeroed, and a 0.1 Hz filter was used to cancel the background noise. Femurs were mounted on customized polylactic acid holders with polyurethane adhesive as described [15,40]. The implant mount was customized to fit the implant after removing the cover screws on the top. The sides of the bone holder were then clamped between two flat specimen clamps once the implant had been firmly secured to the mount and was perpendicular to the axis of rotation. The implant was subsequently removed from the surrounding bone by rotating the femurs at 0.1°/ s while rising at a pace of 0.8 mm/360° simultaneously. After zeroing the initial load and displacement or initial torque and rotation radian, the mechanical examination was done by creating load vs. displacement graphs for 3-point bending analyses and torque vs. radian graphs for torsional analyses. SLM-Shape Language (Modeling version 1.14, MATLAB, MathWorks, Natick, MA, USA) fitted the curve to a bilinear model to separate the linear region from the toe region to eliminate the initial gap potentially between the mount and the implant. The curve was then evaluated for the maximum load at failure (N), stiffness (N m), and toughness at failure (millijoules) and normalized to the cross-sectional area calculated in the microCT analysis for each leg 3-point bending analyses. For the implant osseointegration analysis, it was possible to compute the torque at failure (Nm), torsional stiffness (linear region slope, N m/radians), and torsional energy (area under the curve, millijoules) from the torque vs. degree graphs. The information is displayed as the treatment (right leg) over the control (left leg) as published [40]. ## 2.5. Statistical Analysis A power analysis was performed using an alpha of 0.05 and a power of $80\%$ (delta = 5, sigma = 3, $m = 1$), which revealed that a minimum of $$n = 7$$ per group was required for the study to be statistically significant. An in vivo assessment was done between contralateral legs and treatment legs by Wilcoxon matched-pairs signed rank test (α = 0.05) represented by an asterisk () using GraphPad Prism (GraphPad, La Jolla, CA, USA), and a one-way analysis of variance to compare between groups with Tukey’s post hoc test using JMP statistical software (SAS Institute Inc., Cary, North Carolina). A two-way ANOVA was used to compare differences among groups with two independent variances for removal of the torque mechanical test analysis using GraphPad Prism. ## 3.1. Botox Compromised the Trabecular and Cortical Bone Phenotype at the Distal Metaphysis of the Femurs The trabecular bone and cortical bone phenotype at the distal ends of the femur near the implant insertion site were analyzed by microCT (Figure 2a), and the representative images are shown in Figure 2b–i. The development of a compromised bone phenotype induced by botox injections was demonstrated qualitatively by reduced the trabecular bone formation (Figure 2h) compared to both vehicle groups (Figure 2b,f) and its contralateral leg (Figure 2d). This was further confirmed quantitatively, including a lower BV/TV (Figure 2j), higher total porosity (Figure 2k), and lower trabecular thickness and number (Figure 2l,m). The addition of sema3A did not have any significant effect on increasing the trabecular bone formation in both healthy rats (veh+sema3A, Figure 2j–m) and botox-injected rats (BTX+sema3A, Figure 2j–m). The cortical bone was also affected by a botox injection. Cortical BV/TV was reduced (Figure 2n), the cortical bone total porosity was increased (Figure 2o), and the cortical thickness was reduced (Figure 2p), indicating that botox decreased the cortical bone formation. Additionally, sema3A did not affect the cortical bone formation in healthy or botox-injected rats at the distal metaphysis. The raw data without normalization are also presented in Figure S1. ## 3.2. Sema3A Burst Release Had a Therapeutic Effect on the Botox-Compromised Cortical Bone at Its Injection Sites The cortical bone phenotype was evaluated at the sema3A injected sites at the distal side of the third trochanter and at the mid-diaphysis (Figure 3a) and was compared to the contralateral legs. The difference caused by a botox injection on the cortical bones was hard to distinguish in the qualitative images (Figure 3b–i). However, microCT showed that botox reduced BV/TV at the trochanter, and sema3A injections at that site had no effect (Figure 3j). Botox increased the total porosity (Figure 3k) and decreased the cortical thickness (Figure 3l). The Sema3A injections restored the total porosity to normal levels in the botox-treated rats but had no effect on the cortical thickness. Similarly, botox injections reduced BV/TV (Figure 3m), increased the total porosity (Figure 3n), and decreased the cortical thickness (Figure 3o) at the mid diaphysis in comparison to the contralateral legs, and the injection of sema3A had no effect. Overall, botox injections affected the whole bone phenotype by compromising both the trabecular bone and cortical bone formation, and the effect of sema3A on rescuing the compromised bone phenotype was localized and specific to its injection sites. The raw data without normalization are presented in Figure S2. ## 3.3. Biomimetic Surface Topography Improved Osseointegration, and this was Enhanced by Sema3A Treatment PT and SLAnano implants were inserted into the metaphysis of the distal femurs as described in the methods. The representative images are shown in Figure 4a–c for SLAnano implants. Botox injections caused a reduction in trabecular bone in the bone marrow compartment, regardless of whether the rats were treated with sema3A (Figure 4a–c). Botox reduced the total BIC and cortical BIC compared to the vehicle control groups (Figure 4a,d,f), but botox injections did not affect BIC in the bone marrow compartment (Figure 4e). Even though it was not significantly different, marrow BIC was $18\%$ less in the BTX group than in the control group. The addition of sema3A to the BTX group had a mean of $25\%$ for marrow BIC for SLAnano, which was higher (not statistically significant) than both marrow BIC in the botox ($18.22\%$) and control group ($22.95\%$) (Table S1). The addition of sema3A eliminated the difference in total BIC between the healthy and botox-injected rats, mainly due to significantly higher cortical BIC after a sema3A treatment in the BTX group (Figure 4d,f). PT implants had qualitatively fewer bone trabeculae associated with them than were present around SLAnano implants (compare Figure 4a,g). The PT implants did not alter the botox-compromised bone phenotype (Figure 4h,i). The total BIC for PT implants in botox-treated animals was significantly lower than in the control groups, and the BIC mainly contributed to the decrease in the bone marrow space. This was different from SLAnano implants, in which the decreased total BIC was mainly contributed by lower cortical BIC in the botox-treated rats. Sema3A did not show any effects on improving BIC for PT implants, but there was a therapeutic effect on improving BIC for SLAnano implants under botox-compromised conditions (Figure 4d,f). ## 3.4. Ti Surfaces with a Multiscale Biomimetic Topography Improve Osseointegration for Mechanical Unloading Situations Regardless of Sema3A Treatment Mechanical analysis of hindlimbs by three-point bending (Figure 5a) showed no differences between vehicle and botox treatment with or without sema3A treatment for the maximum load (Figure 5b), stiffness (Figure 5c), and toughness (Figure 5d). Mechanical torque to failure was used to quantify the material properties of the newly formed bone around PT and SLAnano implants. In healthy rats, SLAnano implants robustly increased the maximum load (Figure 5e), torsional stiffness (elastic modulus) (Figure 5f), and yield point (Figure 5g), which was consistent with the higher amount of trabecular bone observed in representative microCT images (Figure 4a,g). Compared to PT, the use of SLAnano in BTX-compromised rats increased the integrated bone mechanical properties more than three-fold (Table S1). Botox injections reduced the maximum load, torsional stiffness, and yield point for rough titanium implants regardless of sema3A treatment ($56\%$ reduction) (Figure 5e–g). At the same time, there was no difference in the mechanical properties of bone integrated into PT implants when comparing botox-injected rats with healthy rats (Figure 5e–g). Additionally, sema3A did not affect the mechanical properties of bone attached to either type of implant. Overall, our data showed that titanium implants with a biomimetic surface topography demonstrated the clinical advantages of increasing osseointegration for mechanically unloaded situations compared to smooth titanium implants. The addition of sema3A increased the amount of bone attached to the implants while not affecting the mechanical properties. ## 4. Discussion This study confirmed our previous observations that mechanically unloaded bone in botox-treated rat femurs exhibits a compromised phenotype characterized by reduced trabeculae in the metaphysis and reduced cortical bone in the diaphysis. Consistent with our previous study [40], we did not observe any botox-related toxicity issues in terms of disrupting normal eating, high-stress levels, or other concerns. The results of the present study showed that botox treatment reduced the mechanical stability of transcortical Ti implants. The effect was greatest for implants that lacked a biomimetic surface topography. Our results also show that treatment with sema3A via injection did not mitigate the effects of mechanical unloading resulting from botox injection. However, if sema3A was delivered to the treatment site in a biodegradable Cu-free click hydrogel, it was able to mitigate the impact of botox. Interestingly, this ability to overcome the negative impact of botox was limited to sites receiving implants with biomimetic surface topography. Pathologies, where the muscle function is chronically disrupted, have been proven to affect skeletal health [45,46,47]. In conditions such as bed rest and spinal cord injury, bone loss is rapid and acute, ranging from $5\%$ to $25\%$, depending on the skeletal sites and injury severity [48]. Rodent models representing these clinically mechanically unloaded or muscle disuse conditions, including tail suspension, cast immobilization, intramuscular botox injections, and tendon resection, exhibit bone loss, which might potentially jeopardize bone regeneration [49,50,51,52]. Our results showed that botox injections dramatically decreased trabecular and cortical bone in femurs at three different locations: the distal metaphysis, the mid-shaft, and the proximal side. The muscle paralysis induced by botox injections compromised the whole bone phenotype, as noted previously [53,54,55], whereas the contralateral legs were unaffected. Our previous work also showed that not only was the bone phenotype affected by botox injections, but the regenerative ability of bone tissues with implants that had a hydrophilic biomimetic topography was also impaired to a greater extent compared to neurectomy [40]. Osseointegration is a complex biological event consisting of stem cell recruitment, primary bone formation, bone remodeling, and mature bone formation [56]. Improvements in osseointegration can be approached by improving net bone formation during primary bone formation or by balancing bone formation and bone resorption during the remodeling phase, both of which are interrupted by diseases such as osteoporosis and diabetes [57,58,59,60]. In the current study, a nerve-derived factor, sema3A, was used to evaluate its therapeutic potential in improving bone formation. Sema3A has been shown to increase osteoblastic differentiation, inhibit osteoclast resorption in vitro, and improve bone formation in animal models, including osteoporotic rabbits and mice and diabetic rats [61,62,63]. Our data indicate that sema3A successfully improves botox-induced cortical bone loss to a similar level as the healthy rats with only two burst releases to the periosteum of the third trochanter. Our data also showed that the effect of sema3A was extremely localized. This may have been due to the limited residency of sema3A after the injection of the periosteum. We did not test this clinically important question in the present study. If the effect of sema3A is limited to the injection site, it can be used locally in areas of regeneration without affecting the bone distal to the area of regeneration. We used a rapidly polymerizing Cu-free click chemistry hydrogel to achieve local sema3A delivery to the implant insertion site. We have used this hydrogel to deliver a variety of factors to bone sites without observing any evidence of toxicity to the surrounding tissues [43,44]. The use of our Cu-free click hydrogel for the delivery of sema3A into rat bone sites was investigated in our previous study, including sema3A release kinetics [41]. The rapid in situ polymerization of the hydrogel demonstrated minimal swelling and remained cohesively intact under physiological conditions [41,43,44]. As anticipated, sema3A delivered via the hydrogel was biologically active and restored bone-to-implant contact in the cortical region where the factor was injected. It was retained locally and did not affect BIC in the marrow cavity. Importantly, its effect was evident only when the implant had a biomimetic surface topography. Previous studies have demonstrated that making use of the Ti implants’ physical surface features can encourage peri-implant bone growth and osseointegration in several challenging conditions [12,13,14,15,42]. The roughness produced by the grit-blasting and acid-etching processes results in craters varying from 30 to 100 μm, overlaid with pits in the range of 1 to 3 μm. SLAnano has additional mesoscale and nanoscale features. This complex multi-scale topography contributes to better bone development and osseointegration by mimicking the natural structure of osteoclast resorption pits on the normal bone surface [8,64]. Here, we investigated the contributions of surface topography for improving osseointegration in this compromised model and showed that implants with a microscale/mesoscale/nanoscale structure significantly improved regenerated bone quality by improving the maximum load the bone can bear before failure, increasing the recovery ability of the bones at certain loads, and the higher endurance of loads before permanent damage occurs. The biomimetic surface topography did not overcome the negative botox effect on the mechanical properties to any greater extent than we observed previously with modSLA implants, which had the same topography but were hydrophilic rather than hydrophobic [40]. However, our data do indicate that biomimetic surface topography contributes to the success of additive approaches, potentially by providing an osteogenic microenvironment that can be further enhanced pharmacologically [28]. The observation that sema3A treatment further enhanced BIC on multiscale biomimetic surfaces but not on smooth surfaces was consistent with in vitro observations [28]. Sema3A increased the production of osteoprotegerin by MSCs cultured on Ti surfaces with a biomimetic topography. Sema3A is a coupling factor that can increase bone formation and decrease bone resorption [29] and osteoprotegerin is a decoy factor that can inhibit osteoclast differentiation [65]. These observations suggest that sema3A increased BIC by increasing bone formation or decreasing bone resorption, achieving net bone formation. The present data also suggest that the osteogenic factors generated by cells on biomimetic multiscale topographies work in concert with exogenous sema3A, whereas cells on smooth surfaces either do not produce these factors or produce them at concentrations that are not sufficient for a synergistic effect, especially in compromised bone conditions. The use of botox in this study was to create a mechanically unloaded situation that mimics clinical conditions, such as patients with neuromuscular injuries and spinal cord injuries or are recovering from prolonged bed rest or long-term use of wheelchairs, as well as patients who have experienced microgravity. The negative impact of botox on the osseointegration of Ti implants was greater than observed in a neurectomy model that also mechanically unleaded the femoral bone [40]. However, our findings show that data impaired osseointegration could be improved by surface modifications to mimic the natural bone environment in combination with sema3A, although sema3A did not show the advantages of improving either the whole bone mechanical properties or mechanical properties of regenerated bone around implants at the periods we checked, even though our data showed that sema3A increased bone formation and BIC. The results of the present study also might indicate a therapeutic method to improve osseointegration in patients with compromised bone regeneration. Further studies emphasizing the concentration and delivery of sema3A can be modified for clinical use to improve osseointegration more than rescue. ## 5. Conclusions The biomimetic concept of providing surface multiscale topography to resemble the natural bone structure is a promising tool for enhancing osseointegration in compromised bone-like disuse conditions, especially when surface modifications are combined with local factors produced by surface-cultured osteoblastic lineage cells. Titanium implant surfaces with a multiscale micro/nano texture exhibited the advantage of increasing the mechanical properties of integrated bone in both healthy and botox-compromised rats. Moreover, with the addition of sema3A, the deleterious effect of botox on osseointegration was restored to healthy levels. Clinically, it is important to understand the detrimental effect of disuse on the bone morphology and success rates of Ti implants. The findings have inspired clinicians to consider biomimetic approaches involving surface modifications and biologics combinations for improving implant outcomes in compromised patients. ## References 1. Moy P.K., Medina D., Shetty V., Aghaloo T.L.. Dental Implant Failure Rates and Associated Risk Factors. Int. J. Oral. Maxillofac. Implant.* (2005) 20 569-577 2. Olivares-Navarrete R., Raines A.L., Hyzy S.L., Park J.H., Hutton D.L., Cochran D.L., Boyan B.D., Schwartz Z.. Osteoblast Maturation and New Bone Formation in Response to Titanium Implant Surface Features Are Reduced with Age. J. Bone Miner. Res. (2012) 27 1773-1783. DOI: 10.1002/jbmr.1628 3. Lotz E.M., Olivares-Navarrete R., Berner S., Boyan B.D., Schwartz Z.. Osteogenic Response of Human MSCs and Osteoblasts to Hydrophilic and Hydrophobic Nanostructured Titanium Implant Surfaces. J. Biomed. Mater. Res. A (2016) 104 3137-3148. DOI: 10.1002/jbm.a.35852 4. Boyan B.D., Lotz E.M., Schwartz Z.. Roughness and Hydrophilicity as Osteogenic Biomimetic Surface Properties. Tissue Eng. Part A. (2017) 23 1479-1489. DOI: 10.1089/ten.tea.2017.0048 5. Gittens R.A., Olivares-Navarrete R., Schwartz Z., Boyan B.D.. Implant Osseointegration and the Role of Microroughness and Nanostructures: Lessons for Spine Implants. Acta Biomater. (2014) 10 3363-3371. DOI: 10.1016/j.actbio.2014.03.037 6. Elias C.N., Fernandes D.J., Resende C.R.S., Roestel J.. Mechanical Properties, Surface Morphology and Stability of a Modified Commercially Pure High Strength Titanium Alloy for Dental Implants. Dent. Mater. (2015) 31 e1-e13. DOI: 10.1016/j.dental.2014.10.002 7. Geetha M., Singh A.K., Asokamani R., Gogia A.K.. Ti Based Biomaterials, the Ultimate Choice for Orthopaedic Implants—A Review. Prog. Mater. Sci. (2009) 54 397-425. DOI: 10.1016/j.pmatsci.2008.06.004 8. Cheng A., Humayun A., Cohen D.J., Boyan B.D., Schwartz Z.. Additively Manufactured 3D Porous Ti-6Al-4V Constructs Mimic Trabecular Bone Structure and Regulate Osteoblast Proliferation, Differentiation and Local Factor Production in a Porosity and Surface Roughness Dependent Manner. Biofabrication (2014) 6 45007. DOI: 10.1088/1758-5082/6/4/045007 9. Civantos A., Martínez-Campos E., Ramos V., Elvira C., Gallardo A., Abarrategi A.. Titanium Coatings and Surface Modifications: Toward Clinically Useful Bioactive Implants. ACS Biomater. Sci. Eng. (2017) 3 1245-1261. DOI: 10.1021/acsbiomaterials.6b00604 10. Gittens R.A., McLachlan T., Olivares-Navarrete R., Cai Y., Berner S., Tannenbaum R., Schwartz Z., Sandhage K.H., Boyan B.D.. The Effects of Combined Micron-/Submicron-Scale Surface Roughness and Nanoscale Features on Cell Proliferation and Differentiation. Biomaterials (2011) 32 3395-3403. DOI: 10.1016/j.biomaterials.2011.01.029 11. Souza J.C.M., Sordi M.B., Kanazawa M., Ravindran S., Henriques B., Silva F.S., Aparicio C., Cooper L.F.. Nano-Scale Modification of Titanium Implant Surfaces to Enhance Osseointegration. Acta Biomater. (2019) 94 112-131. DOI: 10.1016/j.actbio.2019.05.045 12. Siqueira R., Ferreira J.A., Rizzante F.A.P., Moura G.F., Mendonça D.B.S., de Magalhães D., Cimões R., Mendonça G.. Hydrophilic Titanium Surface Modulates Early Stages of Osseointegration in Osteoporosis. J. Periodontal. Res. (2021) 56 351-362. DOI: 10.1111/jre.12827 13. Stadlinger B., Korn P., Tödtmann N., Eckelt U., Range U., Bürki A., Ferguson S.J., Kramer I., Kautz A., Schnabelrauch M.. Osseointegration of Biochemically Modified Implants in an Osteoporosis Rodent Model. Eur. Cell Mater. (2012) 25 326-340. DOI: 10.22203/eCM.v025a23 14. Zhou W., Tangl S., Reich K.M., Kirchweger F., Liu Z., Zechner W., Ulm C., Rausch-Fan X.. The Influence of Type 2 Diabetes Mellitus on the Osseointegration of Titanium Implants with Different Surface Modifications-A Histomorphometric Study in High-Fat Diet/Low-Dose Streptozotocin-Treated Rats. Implant Dent. (2019) 28 11-19. DOI: 10.1097/ID.0000000000000836 15. Lotz E.M., Cohen D.J., Ellis R.A., Wayne J.S., Schwartz Z., Boyan B.D.. Ibandronate Treatment Before and After Implant Insertion Impairs Osseointegration in Aged Rats with Ovariectomy Induced Osteoporosis. JBMR Plus. (2019) 3 e10184. DOI: 10.1002/jbm4.10184 16. Boyan B.D., Berger M.B., Nelson F.R., Donahue H.J., Schwartz Z.. The Biological Basis for Surface-Dependent Regulation of Osteogenesis and Implant Osseointegration. J. Am. Acad. Orthop. Surg. (2022) 30 E894-E898. DOI: 10.5435/JAAOS-D-21-00523 17. Kieswetter K., Schwartz Z., Hummert T.W., Cochran D.L., Simpson J., Dean D.D., Boyan B.D.. Surface Roughness Modulates the Local Production of Growth Factors and Cytokines by Osteoblast-like MG-63 Cells. J. Biomed Mater. Res. A. (1996) 32 55-63. DOI: 10.1002/(SICI)1097-4636(199609)32:1<55::AID-JBM7>3.0.CO;2-O 18. Boyan B.D., Lossdörfer S., Wang L., Zhao G., Lohmann C.H., Cochran D.L., Schwartz Z.. Osteoblasts Generate an Osteogenic Microenvironment When Grown on Surfaces with Rough Microtopographies. Eur. Cell Mater. (2003) 6 22-27. DOI: 10.22203/eCM.v006a03 19. Hotchkiss K.M., Reddy G.B., Hyzy S.L., Schwartz Z., Boyan B.D., Olivares-Navarrete R.. Titanium Surface Characteristics, Including Topography and Wettability, Alter Macrophage Activation. Acta Biomater. (2016) 31 425-434. DOI: 10.1016/j.actbio.2015.12.003 20. Zhang Y., Cheng X., Jansen J.A., Yang F., van den Beucken J.J.J.P.. Titanium Surfaces Characteristics Modulate Macrophage Polarization. Mater. Sci. Eng. C Mater Biol. Appl. (2019) 95 143-151. DOI: 10.1016/j.msec.2018.10.065 21. Hotchkiss K.M., Clark N.M., Olivares-Navarrete R.. Macrophage Response to Hydrophilic Biomaterials Regulates MSC Recruitment and T-Helper Cell Populations. Biomaterials (2018) 182 202-215. DOI: 10.1016/j.biomaterials.2018.08.029 22. Raines A.L., Berger M.B., Patel N., Hyzy S.L., Boyan B.D., Schwartz Z.. VEGF-A Regulates Angiogenesis during Osseointegration of Ti Implants via Paracrine/Autocrine Regulation of Osteoblast Response to Hierarchical Microstructure of the Surface. J. Biomed Mater. Res. A. (2019) 107 423-433. DOI: 10.1002/jbm.a.36559 23. Raines A.L., Berger M.B., Schwartz Z., Boyan B.D.. Osteoblasts Grown on Microroughened Titanium Surfaces Regulate Angiogenic Growth Factor Production through Specific Integrin Receptors. Acta Biomater. (2019) 97 578-586. DOI: 10.1016/j.actbio.2019.07.036 24. Lotz E.M., Berger M.B., Schwartz Z., Boyan B.D.. Regulation of Osteoclasts by Osteoblast Lineage Cells Depends on Titanium Implant Surface Properties. Acta Biomater. (2018) 68 296-307. DOI: 10.1016/j.actbio.2017.12.039 25. Mellis D.J., Itzstein C., Helfrich M.H., Crockett J.C.. The Skeleton: A Multi-Functional Complex Organ. The Role of Key Signalling Pathways in Osteoclast Differentiation and in Bone Resorption. . J. Endocrinol. (2011) 211 131-143. DOI: 10.1530/JOE-11-0212 26. Kim J.H., Kim N.. Bone Cell Communication Factors Provide a New Therapeutic Strategy for Osteoporosis. Chonnam Med. J. (2020) 56 94. DOI: 10.4068/cmj.2020.56.2.94 27. Olivares-Navarrete R., Hyzy S.L., Haithcock D.A., Cundiff C.A., Schwartz Z., Boyan B.D.. Coordinated Regulation of Mesenchymal Stem Cell Differentiation on Microstructured Titanium Surfaces by Endogenous Bone Morphogenetic Proteins. Bone (2015) 73 208-216. DOI: 10.1016/j.bone.2014.12.057 28. Lotz E.M., Berger M.B., Boyan B.D., Schwartz Z.. Regulation of Mesenchymal Stem Cell Differentiation on Microstructured Titanium Surfaces by Semaphorin 3A. Bone (2020) 134 115260. DOI: 10.1016/j.bone.2020.115260 29. Hayashi M., Nakashima T., Taniguchi M., Kodama T., Kumanogoh A., Takayanagi H.. Osteoprotection by Semaphorin 3A. Nature (2012) 485 69-74. DOI: 10.1038/nature11000 30. Li Z., Hao J., Duan X., Wu N., Zhou Z., Yang F., Li J., Zhao Z., Huang S.. The Role of Semaphorin 3A in Bone Remodeling. Front. Cell Neurosci. (2017) 11 40. DOI: 10.3389/fncel.2017.00040 31. Kim B.J., Koh J.M.. Coupling Factors Involved in Preserving Bone Balance. Cell Mol. Life Sci. (2019) 76 1243-1253. DOI: 10.1007/s00018-018-2981-y 32. Verlinden L., Vanderschueren D., Verstuyf A.. Semaphorin Signaling in Bone. Mol. Cell Endocrinol. (2016) 432 66-74. DOI: 10.1016/j.mce.2015.09.009 33. Pinotti F.E., Aron M.A.T., Oliveira G.J.P.L., Marcantonio J.E., Marcantonio R.A.C.. Implants with Hydrophilic Surfaces Equalize the Osseointegration of Implants in Normo- and Hyperglycaemic Rats. Braz. Dent J. (2022) 33 71-77. DOI: 10.1590/0103-6440202204793 34. Mardas N., Schwarz F., Petrie A., Hakimi A.R., Donos N.. The Effect of SLActive Surface in Guided Bone Formation in Osteoporotic-like Conditions. Clin. Oral Implant. Res. (2011) 22 406-415. DOI: 10.1111/j.1600-0501.2010.02094.x 35. Li Y., He S., Hua Y., Hu J.. Effect of Osteoporosis on Fixation of Osseointegrated Implants in Rats. J. Biomed Mater. Res. B Appl. Biomater. (2017) 105 2426-2432. DOI: 10.1002/jbm.b.33787 36. Du Z., Xiao Y., Hashimi S., Hamlet S.M., Ivanovski S.. The Effects of Implant Topography on Osseointegration under Estrogen Deficiency Induced Osteoporotic Conditions: Histomorphometric, Transcriptional and Ultrastructural Analysis. Acta Biomater. (2016) 42 351-363. DOI: 10.1016/j.actbio.2016.06.035 37. Rolvien T., Amling M.. Disuse Osteoporosis: Clinical and Mechanistic Insights. Calcif. Tissue Int. (2022) 110 592-604. DOI: 10.1007/s00223-021-00836-1 38. Scott J.M., Warburton D.E.R., Williams D., Whelan S., Krassioukov A.. Challenges, Concerns and Common Problems: Physiological Consequences of Spinal Cord Injury and Microgravity. Spinal Cord. (2011) 49 4-16. DOI: 10.1038/sc.2010.53 39. Marusic U., Narici M., Simunic B., Pisot R., Ritzmann R.. Nonuniform Loss of Muscle Strength and Atrophy during Bed Rest: A Systematic Review. J. Appl. Physiol. (1985) (2021) 131 194-206. DOI: 10.1152/japplphysiol.00363.2020 40. Deng J., Cohen D.J., Redden J., McClure M.J., Boyan B.D., Schwartz Z.. Differential Effects of Neurectomy and Botox-Induced Muscle Paralysis on Bone Phenotype and Titanium Implant Osseointegration. Bone (2021) 153 116145. DOI: 10.1016/j.bone.2021.116145 41. Deng J., Cohen D.J., Sabalewski E.L., Van Duyn C., Wilson D.S., Schwartz Z., Boyan B.D.. Semaphorin 3A Delivered by a Rapidly Polymerizing Click Hydrogel Overcomes Impaired Implant Osseointegration in a Rat Type 2 Diabetes Model. Acta Biomaterialia. (2022) 157 236-251. DOI: 10.1016/j.actbio.2022.11.030 42. Lotz E.M., Cohen D.J., Schwartz Z., Boyan B.D.. Titanium Implant Surface Properties Enhance Osseointegration in Ovariectomy Induced Osteoporotic Rats without Pharmacologic Intervention. Clin. Oral. Implant. Res. (2020) 31 374-387. DOI: 10.1111/clr.13575 43. Hermann C.D., Wilson D.S., Lawrence K.A., Ning X., Olivares-Navarrete R., Williams J.K., Guldberg R.E., Murthy N., Schwartz Z., Boyan B.D.. Rapidly Polymerizing Injectable Click Hydrogel Therapy to Delay Bone Growth in a Murine Re-Synostosis Model. Biomaterials (2014) 35 9698-9708. DOI: 10.1016/j.biomaterials.2014.07.065 44. Hyzy S.L., Kajan I., Wilson D.S., Lawrence K.A., Mason D., Williams J.K., Olivares-Navarrete R., Cohen D.J., Schwartz Z., Boyan B.D.. Inhibition of Angiogenesis Impairs Bone Healing in an in Vivo Murine Rapid Resynostosis Model. J. Biomed Mater. Res. A. (2017) 105 2742-2749. DOI: 10.1002/jbm.a.36137 45. Bianchi M.L., Mazzanti A., Galbiati E., Saraifoger S., Dubini A., Cornelio F., Morandi L.. Bone Mineral Density and Bone Metabolism in Duchenne Muscular Dystrophy. Osteoporos Int. (2003) 14 761-767. DOI: 10.1007/s00198-003-1443-y 46. Lingam S., Joester J.. Lesson of the Week: Spontaneous Fractures in Children and Adolescents with Cerebral Palsy. BMJ. (1994) 309 265. DOI: 10.1136/bmj.309.6949.265 47. Lau R.Y., Guo X.. A Review on Current Osteoporosis Research: With Special Focus on Disuse Bone Loss. J. Osteoporos. (2011) 2011 293808. DOI: 10.4061/2011/293808 48. Kiratli B.J., Smith A.E., Nauenberg T., Kallfelz C.F., Perkash I.. Bone Mineral and Geometric Changes through the Femur with Immobilization Due to Spinal Cord Injury. J. Rehabil. Res. Dev. (2000) 37 225-233. PMID: 10850829 49. Vegger J.B., Brüel A., Brent M.B., Thomsen J.S.. Disuse Osteopenia Induced by Botulinum Toxin Is Similar in Skeletally Mature Young and Aged Female C57BL/6J Mice. J. Bone Miner. Metab. (2018) 36 170-179. DOI: 10.1007/s00774-017-0830-y 50. Ellman R., Grasso D.J., van Vliet M., Brooks D.J., Spatz J.M., Conlon C., Bouxsein M.L.. Combined Effects of Botulinum Toxin Injection and Hind Limb Unloading on Bone and Muscle. Calcif. Tissue Int. (2014) 94 327-337. DOI: 10.1007/s00223-013-9814-7 51. Nakamura H., Morita S., Kumei Y., Aoki K., Shimokawa H., Ohya K., Shinomiya K.I.. Effects of neurectomy and tenotomy on the bone mineral density and strength of tibiae. Acta Astronaut. (2001) 49 179-190. DOI: 10.1016/S0094-5765(01)00097-2 52. Koseki H., Osaki M., Honda Y., Sunagawa S., Imai C., Shida T., Matsumura U., Sakamoto J., Tomonaga I., Yokoo S.. Progression of microstructural deterioration in load-bearing immobilization osteopenia. PLoS ONE (2022) 17. DOI: 10.1371/journal.pone.0275439 53. Warner S.E., Sanford D.A., Becker B.A., Bain S.D., Srinivasan S., Gross T.S.. Botox Induced Muscle Paralysis Rapidly Degrades Bone. Bone (2006) 38 257-264. DOI: 10.1016/j.bone.2005.08.009 54. Warden S.J., Galley M.R., Richard J.S., George L.A., Dirks R.C., Guildenbecher E.A., Judd A.M., Robling A.G., Fuchs R.K.. Reduced Gravitational Loading Does Not Account for the Skeletal Effect of Botulinum Toxin-Induced Muscle Inhibition Suggesting a Direct Effect of Muscle on Bone. Bone (2013) 54 98-105. DOI: 10.1016/j.bone.2013.01.043 55. Poliachik S.L., Bain S.D., Threet D.W., Huber P., Gross T.S.. Transient Muscle Paralysis Disrupts Bone Homeostasis by Rapid Degradation of Bone Morphology. Bone (2010) 46 18-23. DOI: 10.1016/j.bone.2009.10.025 56. Albrektsson T., Wennerberg A.. On Osseointegration in Relation to Implant Surfaces. Clin. Implant. Dent. Relat. Res. (2019) 21 4-7. DOI: 10.1111/cid.12742 57. Zhang C., Zhang T., Geng T., Wang X., Lin K., Wang P.. Dental Implants Loaded with Bioactive Agents Promote Osseointegration in Osteoporosis: A Review. Front. Bioeng. Biotechnol. (2021) 9 591796. DOI: 10.3389/fbioe.2021.591796 58. Qiao S., Sheng Q., Li Z., Wu D., Zhu Y., Lai H., Gu Y.. 3D-Printed Ti6Al4V Scaffolds Coated with Freeze-Dried Platelet-Rich Plasma as Bioactive Interface for Enhancing Osseointegration in Osteoporosis. Mater. Des. (2020) 194 108825. DOI: 10.1016/j.matdes.2020.108825 59. Saito N., Mikami R., Mizutani K., Takeda K., Kominato H., Kido D., Ikeda Y., Buranasin P., Nakagawa K., Takemura S.. Impaired Dental Implant Osseointegration in Rat with Streptozotocin-Induced Diabetes. J. Periodontal. Res. (2022) 57 412-424. DOI: 10.1111/jre.12972 60. Wang L., Gao Z., Liu C., Li J.. Potential Biomarkers of Abnormal Osseointegration of Implants in Type II Diabetes Mellitus. BMC Oral. Health (2021) 21. DOI: 10.1186/s12903-021-01939-9 61. Song A., Jiang F., Wang Y., Wang M., Wu Y., Zheng Y., Song X., Zhang W., Zhou J.. Semaphorin3A Promotes Osseointegration of Titanium Implants in Osteoporotic Rabbits. Clin. Oral. Investig. (2022) 26 969-979. DOI: 10.1007/s00784-021-04081-6 62. García-Castellano J.M., Díaz-Herrera P., Morcuende J.A.. Is Bone a Target-Tissue for the Nervous System? New Advances on the Understanding of Their Interactions. Iowa Orthop. J. (2000) 20 49-58. PMID: 10934625 63. Wang Z., Wei S.. Local Treatment with Sema3a Could Promote the Osseointegration of Hydroxyapatite Coated Titanium Rod in Diabetic Rats. J. Biomater. Appl. (2022) 36 1775-1785. DOI: 10.1177/08853282221075707 64. Wang Q., Zhou P., Liu S., Attarilar S., Ma R.L.W., Zhong Y., Wang L.. Multi-Scale Surface Treatments of Titanium Implants for Rapid Osseointegration: A Review. Nanomaterials (2020) 10. DOI: 10.3390/nano10061244 65. Schwartz Z., Olivares-Navarrete R., Wieland M., Cochran D.L., Boyan B.D.. Mechanisms Regulating Increased Production of Osteoprotegerin by Osteoblasts Cultured on Microstructured Titanium Surfaces. Biomaterials (2009) 30 3390-3396. DOI: 10.1016/j.biomaterials.2009.03.047
--- title: Revealing Natural Intracellular Peptides in Gills of Seahorse Hippocampus reidi authors: - Claudia Neves Correa - Louise Oliveira Fiametti - Gabriel Marques de Barros - Leandro Mantovani de Castro journal: Biomolecules year: 2023 pmcid: PMC10046794 doi: 10.3390/biom13030433 license: CC BY 4.0 --- # Revealing Natural Intracellular Peptides in Gills of Seahorse Hippocampus reidi ## Abstract The seahorse is a marine teleost fish member of the Syngnathidae family that displays a complex variety of morphological and reproductive behavior innovations and has been recognized for its medicinal importance. In the Brazilian ichthyofauna, the seahorse Hippocampus reidi is among the three fish species most used by the population in traditional medicine. In this study, a protocol was performed based on fast heat inactivation of proteases plus liquid chromatography coupled to mass spectrometry to identify native peptides in gills of seahorse H. reidi. The MS/MS spectra obtained from gills allowed the identification of 1080 peptides, of which 1013 peptides were present in all samples and 67 peptide sequences were identified in an additional LC-MS/MS run from an alkylated and reduced pool of samples. The majority of peptides were fragments of the internal region of the amino acid sequence of the precursor proteins ($67\%$), and N- and C-terminal represented $18\%$ and $15\%$, respectively. Many peptide sequences presented ribosomal proteins, histones and hemoglobin as precursor proteins. In addition, peptide fragments from moronecidin-like protein, described with antimicrobial activity, were found in all gill samples of H. reidi. The identified sequences may reveal new bioactive peptides. ## 1. Introduction Fish are the major component of aquatic fauna, living in a microbe-rich environment [1]. The seahorse is a marine teleost fish member of the Syngnathidae family that displays a complex variety of morphological and reproductive behavior innovations. It has been used in Traditional Chinese Medicine. It is believed that seahorse extract plays a role in diseases such as asthma, atherosclerosis, kidney diseases, infertility, erectile dysfunction, incontinence and skin diseases such as acne and persistent nodules [2,3]. Several studies have used dried seahorse powder hydrolyzed by proteases to obtain peptide extracts. Pharmacological investigation has revealed biological applications of these extracts. Antioxidant activity of enzymatic hydrolysates from the seahorse Hippocampus abdominalis has been reported in vitro and in vivo [4,5]. Although some protocols based on protein digestion are successful for the identification of bioactive peptides, in these preparations, the molecules are generated artificially in vitro. A peptidome can be defined as the analysis of the peptides naturally generated in cells and tissues. It usually includes the following steps: sample collection, peptide extraction, fractionation, LC-MS/MS analysis, peptide identification and data mining [6]. Recently, with improvement in mass spectrometry and related proteomics analyses, it has become possible to sequence large numbers of peptides from complex mixtures. Peptides secreted from cells play roles in important physiological processes as signaling molecules (neuropeptides, hormones and growth factors), defense mechanisms (antimicrobial peptides, venoms and toxins) and modulating protein–protein interaction within cells [7]. Furthermore, fast protease inactivation techniques by heating, mainly microwave radiation, followed by molecular weight exclusion filters associated with LC-MS/MS, have revealed a constant set of peptides from cytosolic, mitochondrial and nuclear protein fragments in tissues of mice [7,8,9], zebrafish [10], human cell lines [11] and fungi [12], which have been termed intracellular peptides. Most of these peptides are intermediate products of protein metabolism and appear to be generated by the proteasome complex [13,14] that owns caspase-like (β1), trypsin-like (β2) and chymotrypsin-like (β5) proteolytic activities [6,15]. In addition, these intracellular peptides can play signaling roles after release from their parent proteins. Most bioactive peptides may be hidden in the sequences of functional proteins, such as hemoglobin and histone [16]. For example, Pep-H, a peptide derived from histone found in human brain tissue altered in schizophrenia patients, has shown protection from cell death [17]. The gills are organs that are constantly exposed to the surrounding water and are responsible for breathing, osmoregulation and excretion of nitrogenous waste products [18,19]. Furthermore, the gills are an immunocompetent organ due to the constant exposure to environmental challenges such as pathogens, pollutants and toxins. Several studies have demonstrated the presence of cells and molecules related to the innate immunity in gills, such as antimicrobial peptides (AMPs) and immunoglobulins, which play an important role in inhibiting pathogen invasion [20]. Antimicrobial peptides (AMPs) are related to the innate immune system of various organisms including humans, fish and plants [6,21]. AMPs usually consist of less than 100 amino acid residues, have a positive net charge and contain a hydrophobic region [22]. AMPs can kill bacteria, viruses and fungi by either disrupting their membrane integrity or inhibiting a cellular function. The contact of AMPs is initially due to an electrostatic and hydrophobic interaction with plasma membranes, usually involving pore-forming and non-pore-forming models [22]. Fish AMPs include cathelicidins, β-defensins, hepcidins, piscidins and histone-derived peptides [23]. Here, we performed a protocol based on fast heat inactivation of proteases to identify native peptides in gills of the seahorse Hippocampus reidi. A study of the Brazilian ichthyofauna showed that the seahorse of this species is one of the three fish species most commonly used by the population for medicinal purposes [24]. Peptide fragmentation data obtained by mass spectrometry were analyzed using a genome database of the recently characterized species Hippocampus comes [25], which allowed the identification of peptide sequences. In addition, the protein classes with the highest number of fragments found were discussed in relation to the antimicrobial potential of the peptide sequences. ## 2.1. Animals Three Hippocampus reidi adult specimens were obtained by hand while SCUBA diving in Ubatuba, São Paulo, Brazil (23°27′01.41″ S–45°02′09.00″ O), and kept in the Laboratory of Marine Proteins and Peptides of the Bioscience Institute of Sao Paulo State University. Artificial seawater was prepared using commercial sea salt and mixed according to the manufacturer’s instructions. Animals were housed in aquaria using artificial seawater (water temperature 25 °C) and fed twice a day. This study followed the guidelines of the National Council for Animal Experimentation Control (CONCEA), permission of SISBIO (Protocol Number 78669-1), and was approved by the Ethics Commission for Animal Use (CEUA) at the Bioscience Institute of Sao Paulo State University (Sao Vicente, Brazil; Protocol Number $\frac{02}{2021}$-CEUA). ## 2.2. Peptide Extraction Seahorse peptide extracts were prepared as previously described [10]. Adult male seahorses were anesthetized with a lethal dose of MS 222 (100 mg/L) and taken to the microwave for 10 s to inactivate peptidase and proteases. Next, gills were collected in centrifuge tubes. Ten one-second sonication pulses (4 Hz) were applied to homogenize each sample in 1 milliliter of deionized water, which were then maintained at 80 °C for 20 min. After cooling in ice, HCl was added to a final concentration of 10 mM. The homogenates were centrifuged at 5000× g for 40 min at 4 °C. A Millipore centrifugal filter unit with a molecular weight cut-off of 10.000 Da was used to filter the supernatants. C-18-like Oasis columns (Waters) were then used to purify and concentrate peptides contained in the samples, which were then dried in a vacuum centrifuge. The extract obtained was stored at −80 °C for a subsequent peptidomic analysis. In addition to identifying peptides with cysteine residues, 100 µg from a pool of gill samples was submitted to a reduction and alkylation reaction in order to identify cysteine residues in samples. For the reduction reaction, 5 mM of final concentration of DTT (Dithiothreitol) was added to the sample and incubated for 25 min at 56 °C. The alkylation reaction was performed with the addition of IAA (Iodoacetamide) at a final concentration of 14 mM and incubated for 30 min at room temperature and in the dark. After incubation, a quench of free IAA was performed by adding at 5 mM final concentration, being incubated for another 15 min at room temperature and in the dark. ## 2.3. Fluorescamine Fluorescamine was used to determine peptide concentration at pH 6.8. The reaction was performed at pH 6.8 to ensure that only the amino groups of the peptides react with the fluorescamine, avoiding that free amino acids react with it. In short, 2.5 µL of the sample was mixed with 25 µL of 0.2 M phosphate buffer (pH 6.8) and 12.5 µL of a 0.3 mg/mL acetone fluorescamine solution, then vortexed for 1 min. Next, 110 µL of water was added, and a SpectraMax M2e plate reader (Molecular Devices, Sunnyvale, CA, USA) was used to measure the fluorescence at an excitation wavelength of 370 nm and an emission wavelength of 480 nm. The standard reference for the determination of the peptide concentration was the peptide 5A (LTLRTKL), which has a known composition and concentration. ## 2.4. Liquid Chromatography and Mass Spectrometry The peptide mixture was suspended in $0.1\%$ formic acid and analyzed as follows. An UltiMate 3000 Basic Automated System (Thermo Fisher®) was set up and connected online with a Fusion Lumos Orbitrap mass spectrometer (Thermo Fisher®) at the mass spectrometry facility RPT02H/Carlos Chagas Institute-Fiocruz, Paraná. The peptide mixture was chromatographically separated on a column (15 cm in length with an internal diameter of 75 μm) packed in-house with ReproSil-Pur C18-AQ 3 μm resin (Dr. Maisch GmbH HPLC) with a flow rate of 250 nL/min of $5\%$ to $38\%$ ACN in $0.1\%$ formic acid on a 120 min gradient. The Fusion Lumos Orbitrap was placed in data-dependent acquisition (DDA) mode to automatically turn between full-scan MS and MS/MS acquisition with 60 s dynamic exclusion. Survey scans (300–1500 m/z) were acquired in the Orbitrap system with a resolution of 120,000 at m/z 200. The most intense ions captured in a 2 s cycle time were chosen, excluding those which were unassigned or had a 1+ charge state. The selected ions were then isolated in sequence and fragmented using HCD (higher-energy collisional dissociation) with normalized collision energy of $30\%$. The fragment ions were analyzed with a resolution of 50,000 at 200 m/z. The general mass spectrometric conditions were as follows: 2.3 kV spray voltage, no sheath or auxiliary gas flow, heated capillary temperature of 175 °C, predictive automatic gain control (AGC) enabled, and an S-lens RF level of $30\%$. Mass spectrometer scan functions and nLC solvent gradients were regulated using the Xcalibur 4.1 data system (Thermo Fisher®). ## 2.5. Peptide Identification The raw data files were submitted to search against the NCBI database filtered for taxonomy Hippocampus comes using the software PEAKS Studio (version 8.5; Bioinformatics Solution, Waterloo, ON, Canada). The decoy-fusion method was used to search a decoy database in order to calculate false discovery rate (FDR). The following search parameters were considered: (a) precursor mass tolerance to ±30 ppm; (b) fragment ion mass (tolerance of ±0.5 Da); (c) variable modifications: oxidized methionine (+15.99 Da) and acetylation (+42.01 Da); (d) no enzyme specificity. The identified peptides were then sorted by their average of local confidence to select the best spectra for annotation, and they were filtered by FDR ≤ $0.1\%$. ## 3. Results The MS/MS spectra obtained of H. reidi gills extracts allowed the identification of 1080 peptides from 396 proteins, of which 1013 peptides were present in all samples initially analyzed (Table S1). Sixty-seven peptide sequences were identified in an additional LC-MS/MS run from an alkylated and reduced pool of gill samples. This run was performed to improve the detection of peptides containing cysteine residues (Table S2). Regarding the general characteristics of these peptides, the fragments ranged from 5 to 38 amino acid residues, with $80\%$ having a length between 8 and 18 amino acid residues (Figure 1A). The precursor proteins of these intracellular peptides are mainly located in the cytosol ($58\%$), nucleus ($22\%$) and mitochondria ($5\%$) (Figure 1B). In addition, most of the peptides found in the gills originated from ribosomal proteins ($32\%$), histones ($11\%$), proteins related to cytoskeleton ($9\%$) and hemoglobin fragments ($5\%$) (Figure 1C). In order to investigate aspects related to proteolytic processing, the amino acid sequence of the precursor proteins and the peptides generated were analyzed. Most of them were fragments of the internal region of the amino acid sequence of the precursor proteins ($67\%$). N- and C-terminal regions represented $18\%$ and $15\%$, respectively (Figure 2A). About $70\%$ of the N-terminal fragments were found have an N-terminal acetyl group. The most frequent amino acid in the N-terminal region of the identified peptides was alanine, followed by serine, leucine and lysine (Figure 2B). Rare cleavage sites include tryptophan and glutamine. In addition, arginine is the most common residue in the P1 position ($60.19\%$), followed by lysine ($21.12\%$) (Figure 2C). The quantification data for each peptide sequence found in all gill samples was performed and is shown in Supplementary Table S1. The ratio of each peptide detected in one sample was calculated in relation to the average of the two other samples and the average ratios shown in Figure 3A. Of the 1013 quantified peptides, $72.8\%$ had an average ratio between 1.0 and 2.0, $16.1\%$ between 2.0 and 4.0 and the remaining ($11.1\%$) distributed in different ratio ranges above 6 (Figure 3B). ## 4. Discussion The main result shown here was the characterization of the peptide profile naturally present in the gills of the seahorse H. reidi. Most studies involving peptide extracts from seahorses have used enzymes for the preparation of hydrolysates, generating sequences artificially. Specific sample preparation methods and LC-MS/MS analysis can contribute to the predominant identification of peptides with distinct physicochemical properties [6]. Peptides in seahorse gills were extracted after microwave irradiation for rapid inactivation of proteases by heat, followed by extraction in cold acidic deionized water to avoid postmortem artifacts [7]. The features of peptides identified in this study, such as the fragment size, their subcellular distribution and protein classes, showed similarities with other peptidome studies that applied the same extraction protocol [10,12]. Most peptides within cells are generated during protein degradation by the large complex proteolytic proteasome that generates fragments within 5 to 22 amino acids [6,15]. This range of the sequences represented more than $90\%$ of peptides found in gills of H. reidi. The three major catalytic activities associated with proteasomes are caspase-like, trypsin-like and chymotrypsin-like activities that cleave after acid, basic and hydrophobic residues, respectively [15]. Despite the similarities with other peptidome studies, with respect to the general characteristics mentioned previously, differences related to proteolytic processing were observed. Our findings show strong trypsin-like activity in peptides, with a predominance of basic amino acids in the P1 position of the cleavage site, in addition to the fragments being mostly from the internal region of the precursor proteins. These differences could be related to the functional aspects of the gills, as an organ that has contact with the external environment. A meta-analysis of human fluid peptidome demonstrated that, at least for serum and tears, there is a preference for Arg and Lys, positively charged amino acids, in the P1 position [26]. The presence of basic amino acid residues in the P1 position is common in peptides generated during the classical secretory pathway, such as neuropeptides and hormones [27]. The proteasome is well known for its role in generating antigenic fragments during adaptive immunity [28,29]. However, it has been suggested that the proteasome also produces antimicrobial peptides as part of the innate immune response, as shown with the intermediate filament keratin 6a (K6a), which is constitutively processed into antimicrobial fragments in corneal epithelial cells [30]. Thus, trypsin-like protease activity seems to have a significant contribution to the shape of peptides in gills of seahorse. Quantitatively, most peptides (72,$8\%$) showed an average ratio between samples of 1 to $2\%$. In a zebrafish brain peptidome study without treatments, variations in the average ratio of peptides were also observed, with some intracellular peptides showing a greater variation than the neuropeptides [10]. Furthermore, in our analyses, peptide fragments from the same protein that presented different ratios were observed, which may indicate a distinct proteolytic processing. Mucosal barriers, such as the skin, gills and gut epithelia, provide a first line of defense against infection [31]. Antimicrobial peptides (AMPs) have been identified in gills of several species of fish [1,32,33,34]. Moronecidin is an AMP member of the piscidin family that has high salt tolerance and a broad spectrum of activity against microorganisms. The membrane disruption of moronecidin in microbes is due to the formation of a pore [31]. The alignment of moronecidin-like peptide, originally identified in H. comes, with other piscidins shows that the signal peptide is conserved, whereas the sequences of the mature peptide are variable [35]. A total of fifteen fragments from the moronecidin-like peptide were found in gill samples of H. reidi (Figure 4A). Among these sequences identified here, the fragments FFRNLWKGAK, KGAKAAFR, GAKAAFR and NLWKGAKAAFR are part of a peptide selected by in silico analysis of the H. comes genome and recently characterized as an antimicrobial peptide [35]. The sequence FFRNLWKGAK (Figure 4B) preserves amphipathic characteristics, such as α-helix formation, with hydrophobic residues on the same surface and also positively charged amino acid residues (Figure 4C), as described for piscidins [23,36]. Ribosomal proteins, histones, and hemoglobin were the intracellular proteins that had most sequences found in the gill peptidome. Antimicrobial activity was demonstrated for peptides from these intracellular proteins in many different species [16]. Peptide sequences from ribosomal proteins represented $32\%$ of the fragments in the gills of the seahorse H. reidi, with many fragments of the different ribosomal proteins that constitute the 40S small subunit and the 60S large subunit of eukaryotic ribosomes. Of the 78 described ribosomal proteins that constitute these two subunits, 61 were detected in our data, of which 36 proteins belonged to the 60S subunit and 25 proteins to the 40S subunit. In Figure 5, the number of fragments found for each ribosomal protein is shown. Ribosomal protein L13 was the one with the highest number of fragments. Hurtado-Rios and collaborators [37] have recently discussed the role of ribosomal proteins as moonlighting proteins, which are those capable of performing more than one biochemical or biophysical function within the same polypeptide chain, highlighting them as natural antimicrobials [37,38,39]. Recently, a C-terminal fragment of the 60S ribosomal protein L27 was isolated from the skin of S. asotus and identified as AMP [40]. In another study, antimicrobial activity against Bacillus megaterium, Escherichia coli and Candida albicans was detected in an extract from the epidermal mucus of Atlantic cod (Gadus morhua). Ribosomal proteins L40, L36A and L35 were identified in fractions prepared by weak cation exchange chromatography together with reversed-phase chromatography and mass spectrometry [41]. In addition, in oyster gills from Cassostrea gigas, a fragment of 60S ribosomal protein L29, with 54 amino acid residues, was identified as AMP. Table 1 lists some identified sequences of ribosomal proteins with similar properties of antimicrobial peptides. Histones are evolutionary conserved basic proteins, present in all eukaryotic cells. They are known to function in chromatin structure formation, nuclear targeting and regulation of gene expression [42]. Histone-derived AMPs have been identified in a number of fish species, with broad-spectrum activity against both human and fish pathogens, suggesting that complete sequences, N-terminal or C-terminal fragments are part of an ancient innate immune mechanism [43,44]. Histones H2A and H2B have been found in microsomes from gill epithelium of five species of primitive to advanced teleost fish, indicating that these proteins might be secreted to the extracellular environment [45]. In addition, histone H2A has been shown to be synthesized in excess, in amounts required for DNA packaging, and accumulates in cytoplasmic granules in gastric gland cells. Upon secretion into the gastric lumen, it is processed by pepsin C isozymes to yield buforin I [46,47], and proteasomal degradation of histone proteins increases after oxidative damage [48]. The peptide buforin II (TRSSRAGLQFPVGRVHRLLRK) derived from buforin I demonstrates a potent antimicrobial activity, killing the bacteria without cell lysis and having affinity with nucleic acids, apparently inhibiting the cellular function by binding to DNA and/or RNA [49]. A similar sequence TRSSRAGLQFPVGRVLR, identified as histone H2A-like found here, presents a hydrophobic ratio of $35\%$, total net charge of +4, four hydrophobic residues on the same surface, and has 80,$75\%$ of similarity with buforin II. Here, most of the peptide sequences found for H2A and H2B represented fragments of the N- and C-terminal of these proteins (Figure 6). Peptide sequences derived from hemoglobin were $5\%$ of fragments present in seahorse gills. Hemoglobin is an essential protein for maintaining cellular homeostasis due to its ability to bind and transport oxygen to the tissues. This protein has also been associated with immune response modulation, signal transduction and antimicrobial activity [50]. Hemoglobin has been known as a source of endogenous bioactive peptides that present several different functions [51,52,53]. Regarding innate immunity, the presence of antimicrobial peptide fragments of hemoglobin isolated from tissues such as skin, branchial epithelium and liver has been shown in fish. For example, an antiparasitic effect of hemoglobin-derived AMPs has been identified from the epithelium of the catfish Ictalurus punctatus, with changes in both the HbβP-1 sequence transcribed and translated in skin and gill epithelium against infection of Ichthyophthirius multifiliis, where the hemoglobin concentration expressed in vivo appeared to be similar to the antiparasitic concentrations measured in vitro [54]. Furthermore, a study in sea bass detected changes in Hb-LP gene expression by real-time RT-PCR in the gills and skin in acute crowding stress, but not in other tissues [55]. In addition, among the various fragments of the hemoglobin alpha chain found in H. reidi gill samples (Figure 7A), fifteen sequences were from a similar region corresponding to the peptide FAHWPDLGPGSPSVKKHGKVIM, derived from hemoglobin alpha in the liver of Japanese eel, Anguilla japonica, with strong antibacterial activities against Gram-positive or -negative bacteria [56]. One of these fragments showed a $56\%$ similarity with this previously described peptide (Figure 7B). Our study also verified the presence of peptides containing cysteines through an additional experiment with reduction and alkylation without trypsinization, because some classes of antimicrobial peptides, such as defensins, that are small, amphiphilic and cationic peptides, are usually rich in cysteines [57]. Of the sixty-seven peptides with cysteine residues found, ten sequences contained two residues and the others only one. However, vertebrate defensins are cationic antimicrobial peptides and have three pairs of disulfide bonds forming three intramolecular disulfide bonds [58]. Future investigations will be necessary to verify the antimicrobial activity of these sequences. Taken together, it was possible to identify the presence of fragments similar to peptides with described antimicrobial activity, and here only this aspect was explored. However, other intracellular peptide fragments identified in this study may be related to other distinct biological functions, as it has been shown in recent years [52,59,60]. ## 5. Conclusions The gill peptidome of seahorse H. reidi obtained through fast heat inactivation of proteases revealed a set of peptides consisting of many fragments from intracellular proteins residing in compartments such as the cytosol, nucleus and mitochondria. Our data bring relevant information on the levels of these molecules, showing a particular behavior for each peptide, which presents variations from animal to animal, even before being submitted to an experimental condition. Most of the sequences found showed sizes and properties similar to peptidomes from tissues of other organisms where the same protocol was used. However, the proteolytic processing analysis showed differences, such as the prevalence of internal fragments of precursor proteins and the predominance of basic residues such as arginine and lysine in the P1 position of the cleavage site. These differences may be related to some functions of gills, such as innate immune response, which may generate possible peptide sequences with antimicrobial action. Furthermore, our results confirm data from the literature that demonstrate antimicrobial activity of peptide fragments of intracellular proteins. Recognition of patterns and identifying key peptides in seahorse gills could lead to a better understanding of processes such as immune response and proteolysis dynamics. The identified sequences could yield new therapeutic peptides as an alternative to the resistance presented by many pathogenic microorganisms to antibiotics available as treatment. ## References 1. Houyvet B., Bouchon-navaro Y., Bouchon C., Corre E., Zatylny-gaudin C.. Marine Transcriptomics Analysis for the Identification of New Antimicrobial Peptides. Mar. Drugs (2021) 19. DOI: 10.3390/md19090490 2. Vincent A.C.J., Foster S.J., Koldewey H.J.. Conservation and Management of Seahorses and Other Syngnathidae. J. Fish Biol. (2011) 78 1681-1724. DOI: 10.1111/j.1095-8649.2011.03003.x 3. Kumaravel K., Ravichandran S., Balasubramanian T., Sonneschein L.. Seahorses—A Source of Traditional Medicine. Nat. Prod. Res. (2012) 26 2330-2334. DOI: 10.1080/14786419.2012.662650 4. Oh Y., Ahn C.B., Yoon N.Y., Nam K.H., Kim Y.K., Je J.Y.. Protective Effect of Enzymatic Hydrolysates from Seahorse (. Biomed. Pharmacother. (2018) 108 103-110. DOI: 10.1016/j.biopha.2018.08.143 5. Kim H.S., Kim S.Y., Fernando I.P.S., Sanjeewa K.K.A., Wang L., Lee S.H., Ko S.C., Kang M.C., Jayawardena T.U., Jeon Y.J.. Free Radical Scavenging Activity of the Peptide from the Alcalase Hydrolysate of the Edible Aquacultural Seahorse (. J. Food Biochem. (2019) 43 e12833. DOI: 10.1111/jfbc.12833 6. Lyapina I., Ivanov V., Fesenko I.. Peptidome: Chaos or Inevitability. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222313128 7. Fricker L.D.. Analysis of Mouse Brain Peptides Using Mass Spectrometry based Peptidomics: Implications for Novel Functions Ranging from Non-Classical Neuropeptides to Microproteins. Mol. Biosyst. (2010) 6 1355-1365. DOI: 10.1039/c003317k 8. Castro L.M., Cavalcanti D.M.L.P., Araujo C.B., Rioli V., Icimoto M.Y., Gozzo F.C., Juliano M., Juliano L., Oliveira V., Ferro E.S.. Peptidomic Analysis of the Neurolysin-Knockout Mouse Brain. J. Proteom. (2014) 111 238-248. DOI: 10.1016/j.jprot.2014.03.043 9. Cavalcanti D.M.L.P., Castro L.M., Neto J.C.R., Seelaender M., Neves R.X., Oliveira V., Forti F.L., Iwai L.K., Gozzo F.C., Todiras M.. Neurolysin Knockout Mice Generation and Initial Phenotype Characterization. J. Biol. Chem. (2014) 289 15426-15440. DOI: 10.1074/jbc.M113.539148 10. Teixeira C.M.M., Correa C.N., Iwai L.K., Ferro E.S., Castro L.M.. De Characterization of Intracellular Peptides from Zebrafish (. Zebrafish (2019) 16 240-251. DOI: 10.1089/zeb.2018.1718 11. Gelman J.S., Sironi J., Castro L.M., Ferro E.S., Fricker L.D.. Peptidomic Analysis of Human Cell Lines. J. Proteome Res. (2011) 10 1583-1592. DOI: 10.1021/pr100952f 12. Dasgupta S., Yang C., Castro L.M., Tashima A.K., Ferro S., Moir R.D., Willis I.M., Fricker L.D.. Analysis of the Yeast Peptidome and Comparison with the Human Peptidome. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0163312 13. Dasgupta S., Fishman M.A., Castro L.M., Tashima A.K., Ferro E.S., Fricker L.D.. Effect of Protein Denaturation and Enzyme Inhibitors on Proteasome-Mediated Production of Peptides in Human Embryonic Kidney Cells. Biomolecules (2019) 9. DOI: 10.3390/biom9060207 14. Dasgupta S., Castro L.M., Dulman R., Yang C., Schmidt M., Ferro E.S., Fricker L.D.. Proteasome Inhibitors Alter Levels of Intracellular Peptides in HEK293T and SH-SY5Y Cells. PLoS ONE (2014) 9. DOI: 10.1371/journal.pone.0103604 15. Abi Habib J., Lesenfants J., Vigneron N., Van den Eynde B.J.. Functional Differences between Proteasome Subtypes. Cells (2022) 11. DOI: 10.3390/cells11030421 16. Chen X., Yi Y., You X., Liu J., Shi Q.. High-Throughput Identification of Putative Antimicrobial Peptides from Multi-Omics Data of the lined seahorse (. Mar. Drugs (2019) 18. DOI: 10.3390/md18010030 17. Café-mendes C.C., Ferro E.S., Torrão A.S., Crun F., Rioli V., Schmitt A., Falkai P., Britto L.R., Turck C.W., Martins-de-souza D.. Peptidomic Analysis of the Anterior Temporal Lobe and Corpus Callosum from Schizophrenia Patients. J. Proteom. (2017) 151 97-105. DOI: 10.1016/j.jprot.2016.05.025 18. Evans D.H., Piermarini P.M., Choe K.P.. The Multifunctional Fish Gill: Dominant Site of Gas Exchange, Osmoregulation, Acid-Base Regulation, and Excretion of Nitrogenous Waste. Physiol. Rev. (2005) 85 97-177. DOI: 10.1152/physrev.00050.2003 19. Hwang P., Lee T., Lin L.. Ion Regulation in Fish Gills: Recent Progress in the Cellular and Molecular Mechanisms. Am. J. Physiol. Regul. Integr. Comp. Physiol. (2011) 30 28-47. DOI: 10.1152/ajpregu.00047.2011 20. Zaccone G., Capillo G., Fernandes J.M.O., Kiron V., Lauriano E.R., Alesci A., Lo Cascio P., Guerrera M.C., Kuciel M., Zuwala K.. Expression of the Antimicrobial Peptide Piscidin 1 and Neuropeptides in Fish Gill and Skin: A Potential Participation in Neuro-Immune Interaction. Mar. Drugs (2022) 20. DOI: 10.3390/md20020145 21. Huan Y., Kong Q., Mou H., Yi H.. Antimicrobial Peptides: Classification, Design, Application and Research Progress in Multiple. Front. Microbiol. (2020) 11 2559. DOI: 10.3389/fmicb.2020.582779 22. Avci F.G., Akbulut B.S., Ozkirimli E.. Membrane Active Peptides and Their Biophysical Characterization. Biomolecules (2018) 8. DOI: 10.3390/biom8030077 23. Bin Hafeez A., Jiang X., Bergen P.J., Zhu Y.. Antimicrobial Peptides: An Update on Classifications and Databases. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222111691 24. El-Deir A.C.A., Collier C.A., Almeida Neto M.S.D., Silva K.M.D.S., Policarpo I.D.S., Araujo T.A.S., Alves R.R.N., de Albuquerque U.P., Moura G.J.B.D.. Ichthyofauna Used in Traditional Medicine in Brazil. Evid.-Based Complement. Altern. Med. (2012) 2012 474716. DOI: 10.1155/2012/474716 25. Lin Q., Fan S., Zhang Y., Xu M., Zhang H., Yang Y., Lee A.P., Woltering J.M., Ravi V., Gunter H.M.. The seahorse genome and the evolution of its specialized morphology. Nature (2016) 540 395-399. DOI: 10.1038/nature20595 26. Magalhães B., Trindade F., Barros A.S., Klein J., Amado F., Ferreira R., Vitorino R.. Reviewing Mechanistic Peptidomics in Body Fluids Focusing on Proteases. Proteomics (2018) 8 1900187. DOI: 10.1002/pmic.201800187 27. Tinoco A.D., Saghatelian A.. Investigating Endogenous Peptides and Peptidases Using Peptidomics. Biochemistry (2011) 50 7447-7461. DOI: 10.1021/bi200417k 28. Kammerl X.I.E., Meiners S.. Proteasome Function Shapes Innate and Adaptive Immune Responses. Am. J. Physiol. Lung Cell. Mol. Physiol. (2016) 24 328-336. DOI: 10.1152/ajplung.00156.2016 29. Kudriaeva A.A., Belogurov A.A.. Proteasome: A Nanomachinery of Creative Destruction. Biochemistry (2019) 84 S159-S192. DOI: 10.1134/S0006297919140104 30. Chan J.K.L., Yuen D., Hiu P., Too M., Sun Y., Willard B., Man D., Tam C.. Keratin 6a Reorganization for Ubiquitin—Proteasomal Processing Is a Direct Antimicrobial Response. J. Cell Biol. (2018) 217 731-744. DOI: 10.1083/jcb.201704186 31. Raju S.V., Sarkar P., Kumar P., Arockiaraj J.. Piscidin, Fish Antimicrobial Peptide: Structure, Classification, Properties, Mechanism, Gene Regulation and Therapeutical Importance. Int. J. Pept. Res. Ther. (2021) 27 91-107. DOI: 10.1007/s10989-020-10068-w 32. Rauf A., Shariati M.A., Khalil A.A., Bawazeer S., Heydari M., Plygun S., Laishevtcev A., Hussain M.B., Alhumaydhi F.A., Aljohani A.S.M.. Hepcidin, an overview of biochemical and clinical properties. Steroids (2020) 160 108661. DOI: 10.1016/j.steroids.2020.108661 33. Salger S.A., Cassady K.R., Reading B.J., Noga E.J.. A Diverse Family of Host-Defense Peptides (Piscidins) Exhibit Specialized Anti-Bacterial and Anti-Protozoal Activities in Fishes. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0159423 34. Silphaduang U., Colorni A., Noga E.J.. Evidence for Widespread Distribution of Piscidin Antimicrobial Peptides in Teleost Fish. Dis. Aquat. Organ. (2006) 72 241-252. DOI: 10.3354/dao072241 35. Mohammadi M., Taheri B., Momenzadeh N., Salarinia R., Nabipour I., Farshadzadeh Z., Bargahi A.. Identification and Characterization of Novel Antimicrobial Peptide from Hippocampus Comes by In Silico and Experimental Studies. Mar. Biotechnol. (2018) 20 718-728. DOI: 10.1007/s10126-018-9843-3 36. Masso-silva J.A., Diamond G.. Antimicrobial Peptides from Fish. Pharmaceuticals (2014) 7 265-310. DOI: 10.3390/ph7030265 37. Hurtado-Rios J.J., Carrasco-Navarro U., Almanza-Perez J.C., Ponce-Alquicira E.. Ribosomes: The New Role of Ribosomal Proteins as Natural Antimicrobials. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23169123 38. Chen Y., Yao L., Wang Y., Ji X., Gao Z., Zhang S., Ji G.. Identification of Ribosomal Protein L30 as an Uncharacterized Antimicrobial Protein. Dev. Comp. Immunol. (2021) 120 104067. DOI: 10.1016/j.dci.2021.104067 39. Qu B., Ma Z., Yao L., Gao Z., Zhang S.. Preserved Antibacterial Activity of Ribosomal Protein S15 during Evolution. Mol. Immunol. (2020) 127 57-66. DOI: 10.1016/j.molimm.2020.08.024 40. Oh H.Y., Go H., Park N.G.. Identification and Characterization of Sa RpAMP, a 60S Ribosomal Protein L27-Derived Antimicrobial Peptide from Amur Catfish, Silurus Asotus. Fish Shellfish Immunol. (2020) 106 480-490. DOI: 10.1016/j.fsi.2020.06.038 41. Bergsson G., Agerberth B., Jo H.. Isolation and Identification of Antimicrobial Components from the Epidermal Mucus of Atlantic Cod (. FEBS J. (2005) 272 4960-4969. DOI: 10.1111/j.1742-4658.2005.04906.x 42. Parseghian M.H., Luhrs K.A.. Beyond the Walls of the Nucleus: The Role of Histones in Cellular Signaling and Innate Immunity 1. Biochem. Cell Biol. (2006) 84 589-604. DOI: 10.1139/o06-082 43. Smith V.J., Desbois A.P., Dyrynda E.A.. Conventional and Unconventional Antimicrobials from Fish, Marine Invertebrates and Micro-Algae. Mar. Drugs (2010) 8 1213-1262. DOI: 10.3390/md8041213 44. Fernandes J.M.O., Molle G., Kemp G.D., Smith V.J.. Isolation and Characterisation of Oncorhyncin II, a Histone H1-Derived Antimicrobial Peptide from Skin Secretions of Rainbow Trout,. Dev. Comp. Immunol. (2004) 28 127-138. DOI: 10.1016/S0145-305X(03)00120-4 45. Stekhoven F.M.A.H.S., Bonga S.E.W., Flik G.. Extranuclear Histones in Teleost Gills: An Evolutionary Study. Fish Physiol. Biochem. (2005) 30 201-211. DOI: 10.1007/s10695-005-7442-5 46. Kim H.S., Yoon H., Minn I., Park C.B., Taek W., Zasloff M., Kim S.C.. Pepsin-Mediated Processing of the Cytoplasmic Histone H2A to Strong Antimicrobial Peptide Buforin I. J. Immunol. (2000) 165 3268-3274. DOI: 10.4049/jimmunol.165.6.3268 47. Roshanak S., Shahidi F., Yazdi F.T., Javadmanesh A., Movaffagh J.. Buforin I an alternative to conventional antibiotics: Evaluation of the antimicrobial properties, stability, and safety. Microb. Pathog. (2021) 161 105301. DOI: 10.1016/j.micpath.2021.105301 48. Wolf-Levy H., Javitt A., Eisenberg-Lerner A., Kacen A., Ulman A., Sheban D., Dassa B., Fishbain-yoskovitz V., Carmona-rivera C., Kramer M.P.. Revealing the Cellular Degradome by Mass Spectrometry Analysis of Proteasome-Cleaved Peptides. Nat. Biotechnol. (2022) 36 1110-1116. DOI: 10.1038/nbt.4279 49. Kobayashi S., Takeshima K., Park C.B., Kim S.C., Matsuzaki K.. Interactions of the Novel Antimicrobial Peptide Buforin 2 with Lipid Bilayers: Proline as a Translocation Promoting Factor. Biochemistry (2000) 39 8648-8654. DOI: 10.1021/bi0004549 50. Drvenica I.T., Stancic A.Z., Maslovaric I.S., Trivanovi D.I., Ilic V.L.. Extracellular Hemoglobin: Modulation of Cellular Functions and Pathophysiological Effects. Biomolecules (2022) 12. DOI: 10.3390/biom12111708 51. Groß R., Bauer R., Krüger F., Rücker-braun E.. A Placenta Derived C-Terminal Fragment of β -Hemoglobin with Combined Antibacterial and Antiviral Activity. Front. Microbiol. (2020) 11 508. DOI: 10.3389/fmicb.2020.00508 52. Silva R.N.O., Llanos R.P., Eichler R.A.S., Oliveira T.B., Gozzo F.C., Festuccia W.T., Ferro E.S.. New Intracellular Peptide Derived from Hemoglobin Alpha Chain Induces Glucose Uptake and Reduces Blood Glycemia. Pharmaceutics (2021) 13. DOI: 10.3390/pharmaceutics13122175 53. Gomes I., Dale C.S., Casten K., Geigner M.A., Gozzo F.C., Ferro E.S., Heimann A.S., Devi L.A.. Hemoglobin-Derived Peptides as Novel Type of Bioactive Signaling Molecules. AAPS J. (2010) 12 658-669. DOI: 10.1208/s12248-010-9217-x 54. Ullal A.J., Litaker R.W., Noga E.J.. Antimicrobial Peptides Derived from Hemoglobin Are Expressed in Epithelium of Channel Catfish. Dev. Comp. Immunol. (2008) 32 1301-1312. DOI: 10.1016/j.dci.2008.04.005 55. Terova G., Cattaneo A.G., Preziosa E., Bernardini G., Saroglia M.. Impact of Acute Stress on Antimicrobial Polypeptides MRNA Copy Number in Several Tissues of Marine Sea Bass (. BMC Immunol. (2011) 12. DOI: 10.1186/1471-2172-12-69 56. Zhang D.L., Guan R.Z., Huang W.S., Xiong J.. Isolation and Characterization of a Novel Antibacterial Peptide Derived from Hemoglobin Alpha in the Liver of Japanese Eel, Anguilla Japonica. Fish Shellfish Immunol. (2013) 35 625-631. DOI: 10.1016/j.fsi.2012.08.022 57. Silva P.M., Gonçalves S., Santos N.C.. Defensins: Antifungal Lessons from Eukaryotes. Front. Microbiol. (2014) 5 97. DOI: 10.3389/fmicb.2014.00097 58. Casadei E., Wang T., Zou J., Vecino J.L.G., Wadsworth S., Secombes C.. Characterization of Three Novel β-Defensin Antimicrobial Peptides in Rainbow Trout (. Mol. Immunol. (2009) 46 3358-3366. DOI: 10.1016/j.molimm.2009.07.018 59. Parada C.A., Oliveira I.P., De Gewehr M.C.F., Machado-Neto J.A., Lima K., Eichler R.A.S., Lopes L.R., Bechara L.R.G., Ferreira J.C.B., Festuccia W.T.. Effect of FKBP12-Derived Intracellular Peptides on Rapamycin-Induced FKBP-FRB Interaction and Autophagy. Cells (2022) 11. DOI: 10.3390/cells11030385 60. Gewehr M.C.F., Silverio R., Rosa-Neto J.C., Lira F.S., Ferro E.S.. Peptides from Natural or Rationally Designed Sources Can Be Used in Overweight, Obesity, and Type 2 Diabetes Therapies. Molecules (2020) 25. DOI: 10.3390/molecules25051093
--- title: 'Research on the influence of patient cost-sharing on medical expenses and health outcomes: Taking patients with heart failure as an example' authors: - Huyang Zhang - Ke Ning - Jinxi Wang - Hai Fang journal: Frontiers in Public Health year: 2023 pmcid: PMC10046806 doi: 10.3389/fpubh.2023.1121772 license: CC BY 4.0 --- # Research on the influence of patient cost-sharing on medical expenses and health outcomes: Taking patients with heart failure as an example ## Abstract ### Objective The objective of this study is to assess the impact of the changes in patient cost-sharing on the medical expenses and health outcomes of patients with heart failure in China. ### Methods The claim data of patients diagnosed with heart failure enrolled in the Urban Employees' Basic Medical Insurance (UEBMI) in the Zhejiang province, China, was used, covering the period from 1 January 2013 to 31 December 2017. The impact of the policy change was estimated through the use of the difference-in-differences method and the event study method. ### Results A total of 6,766 patients and their electronic health insurance claim data were included in the baseline year of 2013. Following the change in the UEBMI reimbursement policies (policy change), a notable decrease was observed in the patient cost-sharing ratios, particularly in the copayment ratio within the policy. However, it did not result in a reduction of the out-of-pocket ratio, which remains a primary concern among patients. An increase was observed in annual outpatient medical expenses, while annual inpatient medical expenses decreased, leading to higher annual medical expenses in the treatment group in comparison to the control group. The effect of the UEBMI reimbursement policy change on health outcomes showed a reduction in the rehospitalization rate within 90 days; however, no significant impact was seen on the rehospitalization rate within 30 days. ### Conclusion The impact of the policy change on medical expenses and health outcomes was found to be modest. To effectively address the financial burden on patients, it is crucial for policymakers to adopt a comprehensive approach that considers all aspects of medical insurance policies, including reimbursement policies. ## 1. Introduction Research into the impact of patient cost-sharing on patient medical expenses and health outcomes has been a subject of significant interest within the field of health economics. One of the most seminal studies in this area is the RAND Health Insurance Experiment (RAND HIE), which was initiated in the United States in 1971 [1]. The RAND HIE demonstrated that medical care utilization increased with a reduction in patient cost-sharing, as evidenced by an elasticity of −0.2 (2–4). This conclusion has been supported by numerous subsequent studies—with different elasticity sizes—including the Oregon Health Insurance Experiment (Oregon HIE), which was initiated in the United States in 2008 (5–15). However, the evidence is not consistent regarding the impact of patient cost-sharing on health outcomes. Some studies, such as the RAND HIE and the Oregon HIE, have shown no significant impact on participants' health [5, 8, 10], while others have reported lower mortality rates or improved health among those with lower cost-sharing [12, 16]. In China, Huang and Gan found a significant decrease in both outpatient service utilization and medical expenditure but no significant effect on self-assessed health, following the implementation of a policy change in 1998, which involved an increase in patient cost-sharing through the transition from the previous labor insurance medical system to the new urban employees' medical insurance system [13]. These inconsistencies may be attributed to heterogeneity in the demand response to medical services across different population subgroups, such as those with various diseases or health statuses. The current body of literature primarily focuses on evaluating the impact of patient cost-sharing on medical utilization or expenses, while there is a limited number of studies examining its effect on health outcomes and focusing on a specific disease [11, 15]. The American Economic Review paper in 2022 discovered that various diseases displayed varying patterns, where some, like heart failure, had a positive productivity growth, whereas others, such as musculoskeletal conditions, did not [17]. This highlights the need for a more tailored approach to guide policymaking, as a one-size-fits-all approach may not be effective in addressing the needs of all patient populations. Policymakers need to consider the unique characteristics of different diseases to achieve the maximum effect of policy decisions. The prevalence of heart failure, an Ambulatory Care Sensitive Condition (ACSC), has been increasing globally, with 4.49 per 1,000 persons in 2019 according to the Institute for Health Metrics and Evaluation (iHME) data. This reveals a progressively larger population with heart failure, especially among older adults [18]. With the population of China surpassing 1.4 billion and urging an aging problem, medical care utilization among heart failure patients is unignorable. Congestive heart failure, as the main type of heart failure, imposes the greatest direct and indirect financial burden among 30 major diseases, accounting for $9.96\%$ of China's national health costs in 2008 and also ranked among the top 10 in 2013 [19, 20]. Chernew et al. [ 21] conducted a study using data from the United States and found that higher cost-sharing for heart failure patients resulted in decreased medication use and subsequently lower medical expenses, particularly in low-income areas. In a separate study conducted in the United States, Snider et al. [ 22] found that the increase in cost-sharing was associated with greater per-patient cost increases for individuals with both diabetes and heart failure compared to those with diabetes alone. However, research on how patient cost-sharing affects medical expenses or health outcomes in heart failure populations is limited. Therefore, this study aims to contribute to the existing evidence concerning the impact of changes in patient cost-sharing on the medical expenses and health outcomes of patients with heart failure using health insurance claims data in China. To our knowledge, this is one of few studies on the impact of patient cost-sharing in a specific disease population in a developing country using large provincial representative datasets at the patient level. From a public health perspective, this study provides critical knowledge for all stakeholders to better understand the impact of insurance reimbursement policies among patients with heart failure in China. Specifically, for policymakers, this knowledge is critical to informing potential policy shifts to reduce the disease burden. ## 2.1. Policy introduction This study uses a natural experiment of policy change in some cities of the Zhejiang province, China to study the impact of patient cost-sharing on medical expenses and health outcomes among patients with heart failure. The health insurance reimbursement policies of five cities in the Zhejiang province underwent changes between the years 2013 and 2017, with the aim of alleviating the financial burden on patients. These changes included increasing the reimbursement rate, raising the cap line, and decreasing the deductible line of the Urban Employees' Basic Medical Insurance (UEBMI). The summary of these modifications is illustrated in Table 1. As the policy change in Shaoxing city took place in 2017, which was late compared to the other four cities, Shaoxing city was treated as a control group after excluding their data in 2017. In addition, Lishui city was omitted due to a lack of data. Table 1 \| Prefecture \| Ever change \| Date of new policy \| What changed \| Government document \| \| --- \| --- \| --- \| --- \| --- \| \| Hangzhou \| YES \| Jan 1, 2014 \| Change A and Change B \| No.68 (2013) (23) No.8 (2013) (24) \| \| Ningbo \| NO \| – \| \| – \| \| Wenzhou \| NO \| – \| \| – \| \| Jiaxing \| YES \| Jan 1, 2015 \| Change A and Change B \| No.87 (2014) (25) \| \| Huzhou \| NO \| – \| \| – \| \| Shaoxing \| YES \| Jan 1, 2017 \| Change A \| \| \| Jinhua \| YES \| Jul 1, 2014 \| Change A and Change B \| No.40 (2014) (26) \| \| Quzhou \| NO \| – \| \| – \| \| Zhoushan \| NO \| – \| \| – \| \| Taizhou \| YES \| Aug 1, 2015 \| Change A, Change B, and Change C \| No.17 (2015) (27) \| \| Lishui \| NO \| – \| \| – \| ## 2.2. Data source and study population The data used in this study was obtained from the UEBMI Database, which had undergone the deidentification of individual information. The database was established and managed by the Chinese Ministry of Human Resources and Social Security, and localities regularly reported to the ministry on a monthly basis. After the national institutional reform in 2018, the ministry-related responsibility was transferred to the newly established National Medical Security Bureau [28]. For the purpose of this study, patients diagnosed with heart failure through either inpatient or outpatient visits were included, using the ICD-10 code I50 and its corresponding Chinese name in medical insurance records as criteria. Individuals younger than 20 years old were excluded. A stratified sampling method was applied to a sample of all heart failure patients' data. The primary stratum was prefecture-level cities in the Zhejiang province. Within each prefecture-level city, the population was stratified by gender and age. Simple random sampling of $1\%$ was conducted within each stratum. The data were aggregated into annual panel data to form the sample for this study. ## 2.3. Outcome measures In this study, three types of dependent variables were used: [1] Medical expenses, [2] Cost-sharing ratios, and [3] Health outcomes (rehospitalization rate). The medical expenses comprised the total expenses, which were further divided into self-expenses within policy, insurance expenses within policy, and self-expenses not covered by policy. The out-of-pocket expenses were calculated as the sum of self-expenses within policy and self-expenses not covered by policy. The cost-sharing ratios included the ratio of self-expenses within policy on total expenses, the out-of-pocket ratio, and the copayment ratio within policy. The health outcomes were measured as the rehospitalization rate within 30 and 90 days, which are widely used indicators (29–32). The definition of each index is presented in Table 2. Patients' sex, age, and comorbidities were included as control variables. Table 2 \| Variable \| Number \| Calculation \| \| --- \| --- \| --- \| \| Type1: medical expenses \| Type1: medical expenses \| Type1: medical expenses \| \| Total expenses \| a \| = b + c + d \| \| Self-expenses within policy \| b \| \| \| Insurance expenses within policy \| c \| \| \| Self-expenses not covered by policy \| d \| \| \| Out-of-pocket expenses \| e \| = a – c = b + d \| \| Type2: cost-sharing ratios \| Type2: cost-sharing ratios \| Type2: cost-sharing ratios \| \| Ratio of self-expenses within policy on total expenses \| f \| = b/a \| \| Out-of-pocket ratio \| g \| = e/a \| \| Copayment ratio within policy \| h \| = b/(b + c) \| \| Type3: health outcomes \| Type3: health outcomes \| Type3: health outcomes \| \| Hospitalization rate \| \| \| \| Rehospitalization rate within 30 days \| \| \| \| Rehospitalization rate within 90 days \| \| \| ## 2.4. Study design and statistical models The difference-in-differences (DID) method is a widely used method of analysis for evaluating the impact of exogenous shocks such as policy changes. This method was first introduced by Ashenfelter [33] as a way to assess the impact of education and training programs on income. It divides the sample into two groups: a treatment group that is subject to the shock (after the shock occurs) and a control group that is not. The DID method assumes that if the treatment group were not subject to the shock, it would have had similar trends in variables as the control group. In this study, the DID method was applied to two types of data. The first specific regression was the main model, which used the annual panel data. The model was as follows: In this model, i in Yiy denoted patient ID and y denoted year, forming a patient-year panel data. Yiy was the dependent variable, which represented various medical expenses, cost-sharing ratios, and health outcomes, as shown in Table 2. To account for the skewed distribution of medical expenses, all indices of medical expenses were transformed by the natural logarithm [34]. Treatediy was a dummy variable that took a value of 1 if the patient was in the treatment group (i.e., they were from one of the four cities where the health insurance reimbursement policies changed) and a value of 0 if they were in the control group (i.e., they were from one of the other six cities where the health insurance reimbursement policies did not change). Postpolicyiy represented the change in health insurance reimbursement policies and took a value of 0 if the change had not occurred yet, 1 if the whole year was after the policy change, or (12-M + 1)/12 if the change started in the Mth month of the current year. The interaction term PostpolicyiyTreatediywas of interest as it indicated whether the change in health insurance reimbursement policies in the treatment group had an impact on Yiy after considering the time effect of the control group and, if so, the direction of the impact. Controliy indicated a set of control variables such as age, gender, and 31 dummy variables for various comorbidities, which were described by Quan et al. [ 35] and can be found in Table 3. εiy was the error term. Table 3* \| Number \| Disease name \| \| --- \| --- \| \| 1 \| AIDS/HIV \| \| 2 \| Alcohol abuse \| \| 3 \| Blood loss anemia \| \| 4 \| Cardiac arrhythmias \| \| 5 \| Chronic pulmonary disease \| \| 6 \| Coagulopathy \| \| 7 \| Congestive heart failure \| \| 8 \| Deficiency anemia \| \| 9 \| Depression \| \| 10 \| Diabetes, complicated \| \| 11 \| Diabetes, uncomplicated \| \| 12 \| Drug abuse \| \| 13 \| Fluid and electrolyte disorders \| \| 14 \| Hypertension, complicated \| \| 15 \| Hypertension, uncomplicated \| \| 16 \| Weight loss \| \| 17 \| Hypothyroidism \| \| 18 \| Liver disease \| \| 19 \| Lymphoma \| \| 20 \| Metastatic \| \| 21 \| Obesity \| \| 22 \| Other neurological \| \| 23 \| Paralysis \| \| 24 \| Peptic ulcer disease excluding bleeding \| \| 25 \| Peripheral \| \| 26 \| Psychoses \| \| 27 \| Pulmonary circulation disorders \| \| 28 \| Renal failure \| \| 29 \| Rheumatoid arthritis/collagen vascular disease \| \| 30 \| Solid tumor without metastasis \| \| 31 \| Valvular disease \| The second specific regression model was not annual but aggregated per visit to the hospital. The model was as follows: In this model, i in iclt denoted the patient ID, c denoted the prefecture-level city, l denoted the quarter of the visit or discharge time, and t denoted the type of visit (outpatient or inpatient). Yiclt was the dependent variable with medical expenses and cost-sharing ratios. Treatediclt denoted the dummy variables for the treatment and control groups, a value of 1 denoted the group with the change of health insurance reimbursement policies, 0 denoted the group without the change of health insurance reimbursement policies. Postpolicyiclt indicated whether it was before or after the change of health insurance reimbursement policies, a value of 1 meant the time was after the policy change, while a value of 0 meant the time was before the policy change. PostpolicyicltTreatediclt was the interaction term that was of interest; the significance of its coefficient indicated whether the change of health insurance reimbursement policies in the treatment group had an impact on Yiclt after considering the time effect of the control group, and if so, the direction of the impact. Controliclt was a series of control variables, including patients' age and gender and 31 dummy variables for various comorbidities, which were described by Quan et al. [ 35] and can be found in Table 3. εiclt was the error term. After illustrating the impact of policy change by DID models quantitatively. The event study method provided a qualitative trend visually. The event study was performed as follows: In this* model, i in iclt denoted the patient ID, c denoted the prefecture-level city, l denoted the quarter of the visit or discharge time, and t denoted the type of visit (outpatient or inpatient). Yiclt were the dependent variables including medical expenses and cost-sharing ratios, as in model [2]. Tc was the quarter of the policy change, 1(l − Tc = τ) denoted the dummy variable, a value of 1 meant that the time difference between the current quarter l and Tc was τ, otherwise a value of 0 was assigned; the model denoted the dummy variables from −m to 8 (−1 was treated as the control period and was, thus, not included) when l − Tc was less than –m, the dummy variable was uniformly set to ≤ -m – 1 and when l − Tc was >8, the dummy variable was uniformly set to ≥9. For example, the policy change in the city of Hangzhou was implemented on 1 January 2014, corresponding to the first quarter of 2014 (2014q1), therefore, if the hospital visit was 2014q2, its time distance from the policy change was 1. Controliclt indicated control variables, such as age, gender, and 31 dummy variables for various comorbidities, which were described by Quan et al. [ 35] and can be found in Table 3; and time variables were added to control for time effects. εiclt was the error term. As part of the robustness check, we analyzed data from 2014 to 2017 to assess the impact of the zero-markup policy implementation in the Zhejiang province, China on 1 April 2014 [36]. The results remained robust even after excluding the months with potential data fluctuations due to the Spring festival. All data cleaning and analysis was performed using the statistical software STATA 17, and all the results reported in this study could be fully reproducible. ## 3.1. Characteristics of study population Six thousand seven hundred and sixty six patients and their corresponding electronic health insurance claim data were included in the baseline year 2013. Of these, 2,899 patients were from the four cities that underwent the policy change and were designated as the treatment group. Table 4 presented a comparison of the basic descriptive results between the treatment and control groups, both before and after the policy change. Table 4 \| Variables \| Description \| Before policy change \| Before policy change.1 \| Before policy change.2 \| After policy change \| After policy change.1 \| After policy change.2 \| \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| \| \| \| Treatment (n = 2,899) \| Control (n = 3,867) \| P \| Treatment (n = 2,833) \| Control (n = 4,324) \| P \| \| Age, years \| ≥ 60 \| 68.8% \| 66.7% \| 0.066 \| 69.0% \| 69.6% \| 0.567 \| \| Gender \| Female \| 44.7% \| 42.0% \| 0.027 \| 45.0% \| 41.3% \| 0.002 \| \| Charlson index \| Mean (SD) \| 4.0 (2.3) \| 4.0 (2.4) \| 0.606 \| 4.7 (2.4) \| 5.2 (2.4) \| 0.000 \| \| Charlson index \| 0 \| 6.1% \| 6.3% \| 0.656 \| 3.0% \| 1.9% \| 0.002 \| \| Charlson index \| 1–3 \| 36.3% \| 37.1% \| 0.507 \| 29.8% \| 24.0% \| 0.000 \| \| Charlson index \| ≥ 4 \| 57.6% \| 56.6% \| 0.388 \| 67.3% \| 74.1% \| 0.000 \| \| Medical expenses \| Medical expenses \| Medical expenses \| Medical expenses \| Medical expenses \| Medical expenses \| Medical expenses \| Medical expenses \| \| Annual total expenses \| Median (IQR) \| 4,470.3 (11,422.1) \| 4,476.1 (7,965.1) \| 0.980 \| 10,080.1 (21,052.4) \| 7,981.3 (13,883.1) \| 0.000 \| \| Annual self-expenses within policy \| Median (IQR) \| 1,096.6 (2,560.1) \| 1,229.5 (2,027.4) \| 0.037 \| 2,072.9 (3,745.2) \| 2,114.6 (2,977.5) \| 0.618 \| \| Annual outpatient total expenses \| Median (IQR) \| 3,116.0 (5,569.8) \| 3,732.7 (5,606.1) \| 0.000 \| 6,605.5 (10,670.8) \| 5,809.3 (70,66.1) \| 0.000 \| \| Annual inpatient total expenses \| Median (IQR) \| 14,048.2 (27,935.1) \| 10,991.3 (19,411.9) \| 0.001 \| 19,117.3 (40,133.7) \| 14,514.0 (26,691.7) \| 0.000 \| \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| Cost-sharing ratios \| \| Annual ratio of self-expenses within policy on total expenses \| Mean (SD) \| 0.30 (0.30) \| 0.30 (0.24) \| 0.199 \| 0.27 (0.29) \| 0.28 (0.21) \| 0.028 \| \| Annual outpatient ratio of self-expenses within policy on total expenses \| Mean (SD) \| 0.31 (0.32) \| 0.32 (0.25) \| 0.364 \| 0.28 (0.31) \| 0.31 (0.23) \| 0.000 \| \| Annual inpatient ratio of self-expenses within policy on total expenses \| Mean (SD) \| 0.19 (0.10) \| 0.17 (0.10) \| 0.000 \| 0.17 (0.11) \| 0.16 (0.12) \| 0.859 \| \| Annual inpatient out-of-pocket ratio \| Mean (SD) \| 0.31 (0.14) \| 0.32 (0.11) \| 0.328 \| 0.29 (0.13) \| 0.32 (0.12) \| 0.000 \| \| Annual inpatient copayment ratio within policy \| Mean (SD) \| 0.22 (0.14) \| 0.19 (0.12) \| 0.000 \| 0.19 (0.12) \| 0.19 (0.13) \| 0.515 \| \| Health outcomes \| Health outcomes \| Health outcomes \| Health outcomes \| Health outcomes \| Health outcomes \| Health outcomes \| Health outcomes \| \| Annual hospitalization rate \| Rate \| 24.8% \| 18.2% \| 0.000 \| 31.2% \| 32.1% \| 0.408 \| \| Rehospitalization rate within 30 days \| Rate \| 14.3% \| 15.5% \| 0.525 \| 25.6% \| 22.6% \| 0.097 \| \| Rehospitalization rate within 90 days \| Rate \| 27.2% \| 23.3% \| 0.091 \| 40.8% \| 38.4% \| 0.260 \| ## 3.2. How the policy change affected medical expenses, cost-sharing ratios, and health outcomes To examine the overall impact of the policy change on medical expenses, cost-sharing ratios, and health outcomes, we used the difference-in-differences (DID) model mentioned previously [model [1]]. The impact on medical expenses is presented in Table 5, which includes coefficients and a $95\%$ confidence interval. Columns [1] and [2] report the estimate of annual medical expenses, including the annual total expenses, annual self-expenses within policy, and annual out-of-pocket expenses. Columns [3]–[5] report the estimate of annual inpatient medical expenses, and columns [6] and [7] report the estimate of annual outpatient medical expenses, respectively. The coefficients of Post policyTreated are of particular interest. The results indicate that the policy change increased the annual outpatient total expenses by $29\%$ ($95\%$ CI 24–$34\%$) while decreasing the annual inpatient total expenses by $47\%$ ($95\%$ CI 28–$66\%$). After taking into account the offset between inpatient and outpatient expenses, we estimated that the policy change resulted in a $14\%$ ($95\%$ CI 9–$20\%$) increase in the annual total expenses. Table 5* \| Unnamed: 0 \| (1) \| (2) \| (3) \| (4) \| (5) \| (6) \| (7) \| \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| \| \| Annual total expenses \| Annual self-expenses within policy \| Annual inpatient total expenses \| Annual inpatient self-expenses within policy \| Annual inpatient out-of-pocket expenses \| Annual outpatient total expenses \| Annual outpatient self-expenses within policy \| \| Post policy \| 0.19* \| 0.15* \| 0.16 \| 0.12 \| 0.15 \| 0.23* \| 0.22* \| \| Post policy \| (0.13, 0.24) \| (0.07, 0.24) \| (−0.05, 0.36) \| (−0.04, 0.28) \| (−0.03, 0.33) \| (0.18, 0.28) \| (0.14, 0.30) \| \| Post policy* treated \| 0.14* \| 0.034 \| −0.47* \| −0.37* \| −0.44* \| 0.29* \| 0.17* \| \| Post policy* treated \| (0.09, 0.20) \| (−0.05, 0.11) \| (−0.66, −0.28) \| (−0.53, −0.22) \| (−0.61, −0.27) \| (0.24, 0.34) \| (0.09, 0.24) \| \| Constant \| 8.03* \| 6.41* \| 5.57 \| 4.29 \| 4.92 \| 6.18* \| 4.80* \| \| Constant \| (7.00, 9.06) \| (4.83, 7.98) \| (1.76, 9.38) \| (1.26, 7.32) \| (1.58, 8.26) \| (5.24, 7.11) \| (3.35, 6.24) \| \| Observations \| 36460 \| 36460 \| 36460 \| 36460 \| 36460 \| 36460 \| 36460 \| Table 6 shows the DID results for cost-sharing ratios. It indicates that after the policy change, the annual outpatient ratio of self-expenses within policy on total expenses, the annual inpatient ratio of self-expenses within policy on total expenses, and the copayment ratio within policy all decreased. Table 6** \| Unnamed: 0 \| (1) \| (2) \| (3) \| (4) \| (5) \| \| --- \| --- \| --- \| --- \| --- \| --- \| \| \| Annual total \| \| Inpatient \| \| Outpatient \| \| \| Ratio of self-expenses within policy on total expenses \| Ratio of self-expenses within policy on total expenses \| Out-of-pocket ratio \| Copayment ratio within policy \| Ratio of self-expenses within policy on total expenses \| \| Post policy \| −0.0076* \| 0.010* \| −0.0015 \| 0.0085 \| −0.0010 \| \| Post policy \| (−0.01, −0.00) \| (0.00, 0.02) \| (−0.01, 0.01) \| (−0.00, 0.02) \| (−0.01, 0.00) \| \| Post policy Treated \| −0.0059 \| −0.021 \| −0.012 \| −0.019* \| −0.010*** \| \| Post policy Treated \| (−0.01, 0.00) \| (−0.03, −0.01) \| (−0.02, 0.00) \| (−0.03, −0.01) \| (−0.02, −0.00) \| \| Constant \| 0.40 \| 0.23* \| 0.28* \| 0.24* \| 0.45* \| \| Constant \| (0.28, 0.53) \| (0.17, 0.28) \| (0.21, 0.35) \| (0.18, 0.30) \| (0.34, 0.56) \| \| Observations \| 36460 \| 10203 \| 10203 \| 10196 \| 36286 \| Regarding health outcomes, as represented by the rehospitalization rate within 30 days and rehospitalization rate within 90 days, Table 7 shows that the policy change did not have a significant effect on the rehospitalization rate within 30 days but did lower the rehospitalization rate within 90 days. Table 7** \| Unnamed: 0 \| (1) \| (2) \| (3) \| \| --- \| --- \| --- \| --- \| \| \| Hospitalization rate \| Rehospitalized in 30 days \| Rehospitalized in 90 days \| \| Post policy \| 0.17* \| −0.0047 \| −0.057 \| \| Post policy \| (0.01, 0.32) \| (−0.37, 0.36) \| (−0.38, 0.26) \| \| Post policy treated \| −0.42** \| −0.0060 \| −0.39* \| \| Post policy treated \| (−0.57, −0.27) \| (−0.40, 0.39) \| (−0.72, −0.05) \| \| Observations \| 20909 \| 3888 \| 4880 \| To further examine how medical expenses per visit and cost-sharing ratios per visit responded to the policy change, we analyzed the results shown in Tables 8, 9. These results indicate that both inpatient and outpatient medical expenses per visit increased after the policy change. The results of the event study are presented in Figures 1, 2. The coefficients of inpatient medical expenses and cost-sharing ratios were almost all insignificant compared to the quarter before the policy change. Figure 2 shows that for outpatient expenses and cost-sharing ratios, the coefficients of outpatient total expenses were also insignificant in all periods, while the outpatient cost-sharing ratios continued to decrease after the policy change. Figure 1:* Changes in inpatient medical expenses and cost-sharing ratios over time. The figure illustrates the coefficient and 90% confidence intervals generated by event study. The horizontal axis represents the number of months relative to the month of policy change implementation. Figure 2: Changes in outpatient medical expenses and cost-sharing ratios over time. The figure illustrates the coefficient and 90% confidence intervals generated by event study. The horizontal axis represents the number of months relative to the month of policy change implementation. ## 4. Discussion This study evaluates the impact of changes to the UEBMI reimbursement policies aimed at reducing patient cost-sharing on medical expenses and health outcomes among patients with heart failure. The findings indicate limited impact. While the total medical expenses per visit increased for both outpatient and inpatient care in response to lower patient cost-sharing (Tables 8, 9), as previously reported in the literature [7, 14], our results show that the annual total expenses in the treatment group did not decrease compared to the control group (Table 5). This suggests that the policy change did not result in a reduction in the economic burden faced by patients with heart failure from a societal perspective. With regard to health outcomes, this study has found that the policy change was associated with a reduced likelihood of rehospitalization within 90 days, but not within 30 days (Table 7), which aligned with the findings of a study in the acute coronary syndrome population [37]. One possible mechanism for reducing medical expenses through lower patient cost-sharing is that patients can choose higher-quality therapies to delay disease progression and reduce hospitalization rates, thereby reducing annual medical expenses. However, this study shows that the reduction in patient cost-sharing did not reduce annual medical expenses, which may be explained by the fact that the reduction in patient cost-sharing did not effectively increase the availability of higher-quality therapies. Additionally, despite the reduced patient cost-sharing, many cost-effective drugs or therapies were not included in the medical insurance catalog. Patient cost-sharing plays an important role in influencing physician and patient behavior, directly and subsequently affecting the reallocation of medical services. This study shows that the impact of the patient cost-sharing on medical expenses differed between outpatients and inpatients. On a micro level, lower patient cost-sharing for outpatients resulted in higher annual medical expenses and expenses per visit (Tables 5, 8). This increase in annual outpatient total expenses led to an overall increase in the annual total expenses, even though annual inpatient total expenses decreased (Table 5). Therefore, from a societal perspective, the policy change did not reduce the medical burden on patients with heart failure. This finding indicates that the reimbursement policies may focus on how to reduce the patient cost-sharing of outpatients in the future. On a macro level, patient cost-sharing can directly impact the economic burden of patients, which can then influence the proportion of the disease burden among all illnesses. Policymakers may use figures on the proportion of the disease burden to inform policy-making decisions, which can, in turn, indirectly impact the economic burden for patients with heart failure. In our study, even though the policy change decreased the inpatient copayment ratio within policy among inpatients, both per visit and annually, it had no effect on the inpatient out-of-pocket ratio or annual total ratio of self-expenses within policy on total expenses (Table 6), which were more concerning to patients. As a result, the findings of this study have significant implications for practical applications. Policymakers may need to consider incorporating reimbursement policies into other medical insurance policies, for example by including more cost-efficient drugs or services in the medical insurance reimbursement catalog. These policies should take into account the unique characteristics of the disease, such as treating heart failure as a special outpatient disease and increasing the reimbursement ratio for outpatient care. These measures would effectively reduce the economic burden on patients with heart failure and improve their overall health outcomes. The limitations of this study should be considered when interpreting the results. First, the study only focused on a limited number of health outcomes and did not consider the impact of patient cost-sharing on other important health outcomes, such as mortality, due to the unavailability of data. Nevertheless, the hospitalization and rehospitalization rates are widely used and are representative indicators. Second, the study was unable to separate out-of-pocket expenses for outpatients due to a lack of information on outpatient insurance expenses within policy. Third, the study did not control for important patient demographics, such as income and education level, due to a lack of data from other sources. To better understand the impact of patient cost-sharing on medical expenses and health outcomes, future studies should employ multiple sources of data, consider a broader range of health outcomes, and incorporate relevant patient demographics. ## 5. Conclusion Our study found that the impact of the policy change on medical expenses and health outcomes was modest. While the patient cost-sharing ratio (specifically the copayment ratio within policy) decreased, the annual total expenses for heart failure patients did not. This indicates that further efforts, such as expanding the health insurance reimbursement catalog to cover more drugs and items, may be needed to relieve patients of medical expenses. These results emphasize the need for ongoing efforts to lower cost-sharing ratios, particularly the out-of-pocket ratio for patients. ## Data availability statement The data analyzed in this study is subject to the following licenses/restrictions: the data used in this study are nonpublic electronic Urban Employees' Basic Medical Insurance claim records belonging to Shanghai Songsheng Business Consulting Co. Ltd. Requests to access these datasets should be directed to JW, bonejizi@126.com. ## Ethics statement Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. ## Author contributions HZ designed the study under the supervision of HF. JW contributed to data collection and discussion. HZ carried out the data analysis and wrote the manuscript. HZ, KN, and HF contributed to the interpretation of the results and revised the manuscript. All authors have read and agreed to the published version of the manuscript. ## Conflict of interest JW was employed by company Shanghai Songsheng Business Consulting Co. Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. ## Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. ## References 1. 1.RAND Corporation,. Rand's Health Insurance Experiment (Hie). Santa Monica: RAND Corporation (2017). Available online at: https://www.rand.org/health-care/projects/hie.html (accessed May 1, 2017).. (2017) 2. Newhouse JP, Group RCIE, Staff IEG. Free for All? Lessons from the Rand Health Insurance Experiment. (1993) 3. Manning WG, Newhouse JP, Duan N, Keeler EB, Leibowitz A. Health insurance and the demand for medical care: evidence from a randomized experiment. Am Econ Rev. (1987) 77 251-77. PMID: 10284091 4. Aron-Dine A, Einav L, Finkelstein A. The rand health insurance experiment, three decades later. J Econ Perspect. (2013) 27 197-222. DOI: 10.1257/jep.27.1.197 5. Finkelstein A, Taubman S, Wright B, Bernstein M, Gruber J, Newhouse JP. The oregon health insurance experiment: evidence from the first year. Quart J Econ. (2012) 127 1057-106. DOI: 10.1093/qje/qjs020 6. Chandra A, Gruber J, McKnight R. The impact of patient cost-sharing on low-income populations: evidence from Massachusetts. J Health Econ. (2014) 33 57-66. DOI: 10.1016/j.jhealeco.2013.10.008 7. Chandra A, Gruber J, McKnight R. Patient cost-sharing and hospitalization offsets in the elderly. Am Econ Rev. (2010) 100 193-213. DOI: 10.1257/aer.100.1.193 8. Shigeoka H. The effect of patient cost sharing on utilization, health, and risk protection. Am Econ Rev. (2014) 104 2152-84. DOI: 10.1257/aer.104.7.2152 9. Brot-Goldberg ZC, Chandra A, Handel BR, Kolstad JT. What does a deductible do? The impact of cost-sharing on health care prices, quantities, and spending dynamics. Quart J Econ. (2017) 132 1261-318. DOI: 10.1093/qje/qjx013 10. Fukushima K, Mizuoka S, Yamamoto S, Iizuka T. Patient cost sharing and medical expenditures for the elderly. J Health Econ. (2016) 45 115-30. DOI: 10.1016/j.jhealeco.2015.10.005 11. Zhao S, Zang W, Yin Q. The welfare effect of medical insurance. Econ Res J. (2015) 50 130-45 12. Huang F, Gan L. Excess demand or appropriate demand? Health insurance, medical care and mortality of the elderly in urban China. Econ Res J. (2010) 45 105-19 13. Huang F, Gan L. The impacts of China's urban employee basic medical insurance on healthcare expenditures and health outcomes. Health Econ. (2017) 26 149-63. DOI: 10.1002/hec.3281 14. Ta Y, Fu H, Li L. The impact of patient cost-sharing on medical expenditure and health outcome—evidence from hospital discharge data. China Econ Quart. (2020) 19 380. DOI: 10.13821/j.cnki.ceq.2020.03.14 15. Wang Z, Feng J, Song H. Health insurance benefits and the utilization of health care among the elderly in China. Finan Trade Econ. (2019) 40 42-59 16. Robinson JC, Brown TT. Increases in consumer cost sharing redirect patient volumes and reduce hospital prices for orthopedic surgery. Health Aff. (2013) 32 1392-7. DOI: 10.1377/hlthaff.2013.0188 17. Cutler DM, Ghosh K, Messer KL, Raghunathan T, Rosen AB, Stewart ST. satellite account for health in the United States. Am Econ Rev. (2022) 112 494-533. DOI: 10.1257/aer.20201480 18. Curtis LH, Whellan DJ, Hammill BG, Hernandez AF, Anstrom KJ, Shea AM. Incidence and prevalence of heart failure in elderly persons, 1994–2003. Arch Intern Med. (2008) 168 418-24. DOI: 10.1001/archinternmed.2007.80 19. Lan L, Li X, Zhou T, Wu R, Meng Q. Measurement of economic burden of diseases in China sampled with data in 2008. Chin Health Serv Manag. (2017) 34 123-7 20. Song X, Lan L, Zhou T, Yin J, Meng Q. Economic burden of major diseases in China in 2013. Front Public Health. (2021) 9 649624. DOI: 10.3389/fpubh.2021.649624 21. Chernew M, Gibson TB, Yu-Isenberg K, Sokol MC, Rosen AB, Fendrick AM. Effects of increased patient cost sharing on socioeconomic disparities in health care. J Gen Intern Med. (2008) 23 1131-6. DOI: 10.1007/s11606-008-0614-0 22. Snider JT, Seabury S, Lopez J, McKenzie S, Wu YY, Goldman DP. Impact of type 2 diabetes medication cost sharing on patient outcomes and health plan costs. Am J Manag Care. (2016) 22 433. PMID: 27355811 23. 23.Hangzhou Municipal Government,. Notice on Printing Distributing the Measures for Basic Medical Security in Hangzhou. Hangzhou: Hangzhou Municipal Government (2013). Available online at: http://hrss.hangzhou.gov.cn/art/2013/7/9/art_1229196644_1159234.html (accessed July 9, 2013).. (2013) 24. 24.Hangzhou Municipal Government,. Notice on Printing Distributing the Detailed Rules for the Implementation of Hangzhou Basic Medical Security Measures in Main Urban Areas. Hangzhou: Hangzhou Municipal Government (2014). Available online at: http://hzbj.zjzwfw.gov.cn/art/2014/6/13/art_4876_6129.html (accessed June 13, 2014).. (2014) 25. 25.Jiaxing Municipal Government,. Notice on Printing Distributing the Interim Measures for the Basic Medical Insurance of Employees in Jiaxing. Jiaxing: Jiaxing Municipal Government (2018). Available online at: http://www.jiaxing.gov.cn/art/2018/11/15/art_1556142_26848811.html (accessed November 15, 2018).. (2018) 26. 26.Jinhua Municipal Government,. Notice on Adjusting Relevant Policies of Basic Medical Insurance in Urban Areas. Jinhua: Jinhua Municipal Government (2014). Available online at: http://jhjd.zjzwfw.gov.cn/art/2014/6/20/art_27048_30085.html (accessed June 20, 2014).. (2014) 27. 27.Taizhou Municipal Government,. Notice on Printing Distributing the Measures of Taizhou City on Basic Medical Insurance for Employees. Taizhou: Taizhou Municipal Government (2014). Available online at: http://tazhy.zjzwfw.gov.cn/art/2014/6/10/art_30005_5820.html (accessed June 10, 2014).. (2014) 28. 28.Central Committee of the Communist Party of China. Deepening Reform of Party and State Institutions. Beijing: Central Committee of the Communist Party of China (2018). Available online at: http://www.gov.cn/zhengce/2018-03/21/content_5276191.htm#allContent (accessed March 21, 2018).. (2018) 29. Aghababian RV. Acutely decompensated heart failure: opportunities to improve care and outcomes in the emergency department. Rev Cardiovasc Med. (2002) 3 S3-9. PMID: 12439425 30. Han Y. Reducing Rehospitalization Mortality in Heart Failure Patients, Hope's Total Management Journey Begins Again.. (2019) 31. Saito H, Kagiyama N, Nagano N, Matsumoto K, Yoshioka K, Endo Y. Social isolation is associated with 90-day rehospitalization due to heart failure. Eur J Cardiovasc Nurs J Work Group Cardiovasc Nurs Eur Soc Cardiol. (2019) 18 16-20. DOI: 10.1177/1474515118800113 32. Wang M. Clinical effect of cardiac exercise rehabilitation on patients with chronic heart failure. Chin J Microcircul. (2020) 30 37-44 33. Ashenfelter O. Estimating the effect of training programs on earnings. Rev Econ Stat. (1978) 60 47-57. DOI: 10.2307/1924332 34. Manning WG, Mullahy J. Estimating log models: to transform or not to transform?. J Health Econ. (2001) 20 461-94. DOI: 10.1016/S0167-6296(01)00086-8 35. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC. Coding algorithms for defining comorbidities in Icd-9-Cm and Icd-10 administrative data. Med Care. (2005) 43 1130-9. DOI: 10.1097/01.mlr.0000182534.19832.83 36. 36.Lv J,. All Public Hospitals in Zhejiang Province to Fully Implement Zero Mark-up for Drugs. Beijing: The People's Daily (2014). Available online at: http://www.gov.cn/xinwen/2014-04/02/content_2651525.htm (accessed April 2, 2014).. (2014) 37. Philipson TJ, Mozaffari E, Maclean JR. Pharmacy cost sharing, antiplatelet therapy utilization, and health outcomes for patients with acute coronary syndrome. Am J Manag Care. (2010) 16 290-7. PMID: 20394466
--- title: Characterization of Triacylglycerol Estolide Isomers Using High-Resolution Tandem Mass Spectrometry with Nanoelectrospray Ionization authors: - Lukáš Cudlman - Aleš Machara - Vladimír Vrkoslav - Miroslav Polášek - Zuzana Bosáková - Stephen J. Blanksby - Josef Cvačka journal: Biomolecules year: 2023 pmcid: PMC10046810 doi: 10.3390/biom13030475 license: CC BY 4.0 --- # Characterization of Triacylglycerol Estolide Isomers Using High-Resolution Tandem Mass Spectrometry with Nanoelectrospray Ionization ## Abstract Triacylglycerol estolides (TG-EST) are biologically active lipids extensively studied for their anti-inflammatory and anti-diabetic properties. In this work, eight standards of TG-EST were synthesized and systematically investigated by nanoelectrospray tandem mass spectrometry. Mass spectra of synthetic TG-EST were studied with the purpose of enabling the unambiguous identification of these lipids in biological samples. TG-EST glycerol sn-regioisomers and isomers with the fatty acid ester of hydroxy fatty acid (FAHFA) subunit branched in the ω-, α-, or 10-position were used. Ammonium, lithium, and sodium adducts of TG-EST formed by nanoelectrospray ionization were subjected to collision-induced dissociation (CID) and higher-energy collisional dissociation (HCD). Product ion spectra allowed for identification of fatty acid (FA) and FAHFA subunits originally linked to the glycerol backbone and distinguished the α-branching site of the FAHFA from other estolide-branching isomers. The ω- and 10-branching sites were determined by combining CID with ozone-induced dissociation (OzID). Lithium adducts provided the most informative product ions, enabling characterization of FA, hydroxy fatty acid (HFA), and FAHFA subunits. Glycerol sn-regioisomers were distinguished based on the relative abundance of product ions and unambiguously identified using CID/OzID of lithium and sodium adducts. ## 1. Introduction Triacylglycerol estolides (TG-ESTs) are a minor class of lipids consisting of a glycerol backbone to which a combination of fatty acids (FA) and estolides is esterified (also reported as fatty acid esters of hydroxy fatty acid, FAHFA, and (O-acyl)-hydroxy fatty acid [1,2,3]) as a structural subunit [4]. TG-ESTs are biologically active molecules known from mammals [5,6,7,8,9,10,11], plants [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38], and fungi [29,33,39,40,41]. In mammalian TG-ESTs, a hydroxy fatty acid (HFA) is usually esterified with a non-hydroxy FA (“capped” estolide), while in plants and fungi, another HFA may be attached (“uncapped” estolide). TG-ESTs serve as a reservoir of biologically active FAHFAs that exhibit various biological activities. FAHFAs are released from TG-ESTs by the action of lipases which may exhibit hydrolytic specificity regarding the estolide-branching site. A subgroup of FAHFAs with specific biological functions can thus be mobilized [7]. Understanding TG-EST and FAHFA metabolism can help to establish new strategies for treating obesity, diabetes, and other diseases [6,42]. The structural variability of TG-EST stems from numerous FA and HFA that can be combined. There are many TG-EST positional isomers, including glycerol sn-regioisomers that differ by the arrangement of acyl chains on the glycerol backbone and estolide-branching regioisomers with different positions of the estolide ester bond within the FAHFA subunit [6,7]. The structural variability of TG-EST is further extended by stereoisomers existing due to the chirality of the estolide-branching carbon [43]. The estolide regioisomers branched in the ω-position (terminal carbon atom of the HFA chain) are found in some plants [12,14,19,20,21,24,31]. The FAHFA subunit is, however, more often branched on an “inner” carbon of the HFA chain; such TG-ESTs exist in plants, fungi, and mammals [6,7,8,9,11,15,16,17,18,22,25,26,27,28,29,30,32,33,34,35,36,39,40,41]. To our knowledge, TG-ESTs with the FAHFA branched in the α-position (carbon number 2 of HFA) have not been reported yet. Their structural unit, α-FAHFA, was recently detected in vernix caseosa, a biofilm that covers the skin of newborn babies [3]. Additionally, the α-FAHFA is a major motif in estolides attached to sugar backbones as in lipid A within Escherichia coli [44]. Due to the metabolic connections between FAHFAs and TG-ESTs [6,7,42], the existence of TG-ESTs with α-branched FAHFA subunits is likely. After hydrolysis and derivatization, TG-EST can be analyzed by gas chromatography [12,13,14,15,16,18,19,24,39,40]. High-performance liquid chromatography (HPLC) is preferred nowadays because it separates intact TG-EST molecules, preserving key structural information on ester bond linkages. Reversed-phase HPLC makes it possible to separate TG-EST isomers according to the number of carbon atoms, carbon–carbon double bonds, and the estolide-branching site in the FAHFA subunit [7]. HPLC coupled with mass spectrometry (HPLC/MS) offers excellent sensitivity and selectivity for TG-EST in biological samples [6,7,8,9,10,11,25]. Electrospray ionization (ESI) of this lipid class typically generates ammonium [M + NH4]+, sodium [M + Na]+, or lithium [M + Li]+ molecular adducts depending on conditions, with subsequent ion activation by collision-induced dissociation (CID) or higher-energy collisional dissociation (HCD) generating structurally informative product ions [6,7,8,9,10,25,26,27,28,35,37]. These spectra usually show product ions identifying the FA(s) linked to the glycerol backbone and the composition of the FAHFA subunit(s) [6,7,25,26,27]. Lithium adducts provide somewhat more informative spectra, with product ion peaks useful for deducing the sub-structure of FAHFA subunits and, in some cases, differentiating between glycerol sn-regioisomers [26,28]. The structure of lipid ions can also be probed by alternative ion activation methods, such as ozone-induced dissociation (OzID) [45,46]. OzID makes it possible to determine the carbon–carbon double bond position within the acyl chain [47,48], the position of estolide-branching [49], and the glycerol sn-position of the acyl chain [48,50,51]. Although some information on TG-EST fragmentation can be found in the literature [6,7,25,26,27], a systematic study of the fragmentation behavior of these lipids is not available. In the absence of these data, identifying TG-EST isomers in biological samples is challenging, and leads to a greater reliance on structural inferences from HPLC retention data [6,7,8,9,11]. In this study, we synthesized eight TG-EST standards to study the mass spectra of estolide-branching regioisomers and glycerol sn-regioisomers. Ammonium, lithium, and sodium adducts generated by nanoelectrospray ionization were fragmented using CID and HCD. Structurally significant ions were identified, and some unimolecular dissociation mechanisms are suggested and further investigated by multistage mass spectrometry, including with OzID. ## 2.1. Solvents and Additives Acetonitrile (LC/MS grade) and propan-2-ol (LC/MS grade) were purchased from Biosolve BV (Valkenswaard, The Netherlands). Methyl t-butyl ether (>$99.8\%$), methanol ($99.9\%$), ammonium formate (>$99.0\%$), lithium formate monohydrate ($98\%$), and sodium formate (>$99.0\%$,) were obtained from Sigma-Aldrich (St. Luis, MO, USA). Chloroform ($99.8\%$) stabilized by ~$1\%$ ethanol from Penta (Chrudim, Czech Republic) was purified by distillation. ## 2.2. TG-EST Standards The structures of eight TG-EST standards investigated in this work are shown in Figure 1. The standards each had one FAHFA subunit in either the sn-$\frac{1}{3}$ (Standards TG-EST 1–7) or the sn-2 position (Standard TG-EST 8). The FAHFA subunits contained the estolide ester bond in positions-2 (α; Standards TG-EST 2, and 5), -10 (Standard TG-EST 3), or at the HFA chain terminus (ω; Standards TG-EST 1, 4, 6, 7, and 8). The standards were synthesized by Steglich esterification mediated by N,N′-dicyclohexylcarbodiimide (DCC) [53]. 10-Hydroxyhexadecanoate, the key compound for preparation of TG-EST 3, was prepared according to our previous work [54]. Details regarding the synthesis, together with spectral data of intermediates and final products, are provided in Supplementary Materials (Schemes S1–S8 and Figures S1–S75). For CID and HCD mass spectrometry experiments, the TG-EST standards were dissolved in a mixture of chloroform: propan-2-ol: acetonitrile: aqueous salt solution (0.5:0.05:0.80:0.10, by vol.) at a concentration of 10 µg mL−1. The aqueous solutions of the salts (ammonium, lithium, or sodium formate) were prepared at a concentration of 1.0 mmol L−1. For OzID experiments, standards of TG-EST were dissolved in a mixture of chloroform: methyl t-butyl ether: methanol (0.05:0.5:0.5, by vol.) at a concentration of 100 µg mL−1. The TG-EST sample was further diluted 1:1 (v/v) in methanolic salt solution (Li+ or Na+; 2.5 mmol L−1). ## 2.3. Mass Spectrometry For CID and HCD workflows, nanoESI-MSn experiments were performed on the Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) equipped with a robotic chip-based nanoESI apparatus TriVersa NanoMate (Advion, Inc., Ithaca, NY, USA). Data were recorded and interpreted manually using Xcalibur 4.1.50 software (Thermo). The TriVersa NanoMate, controlled by Chipsoft 8.3.1 software, was operated with an ionization spray voltage of 1800 V and nitrogen delivery gas at 0.5 psi. TG-EST solutions were loaded into 96-well plates, and the sample aliquots (5–15 μL) were aspirated and infused into the mass spectrometer. The Orbitrap mass analyzer was operated at a resolution of 120,000 FWHM, and m/z values were acquired with an accuracy of less than 3.0 ppm. [M + NH4]+, [M + Li]+, and [M + Na]+ ions were fragmented in MS2 and MS3 using CID and HCD. Helium and nitrogen served as collision gases for CID and HCD, respectively. Values of normalized collision energies (NCEs) are shown in spectra and are a unitless parameter (as NCE automatically compensates for the mass dependence on collision energy). The precursor isolation window was 1.0–1.5 mass units. The mass spectra were recorded in the m/z range of 200−1200 for full MS and MS2, and a m/z range of 150–900 for MS3 was used. The energy-resolved dissociation curves (Supplementary Materials, Figures S76–S87) were calculated from spectra acquired by introducing incremental changes to NCE values. The relative ion abundance values were averaged from 20–40 scans. For MS3 CID/OzID and MS4 CID/CID/OzID sequential workflows, nanoESI-MSn experiments were performed on the LTQ Orbitrap Elite mass spectrometer (Thermo) modified for OzID experiments [55]. An ozone generator (Titan30 UHC; Absolute Ozone, Edmonton, AB, Canada) was used for the external production of ozone (ca. $17\%$ in oxygen), which was introduced into the ion-trap region of the instrument. The TriVersa NanoMate was used in the same way as in the case of CID and HCD with identical parameters (ionization spray voltage, nitrogen delivery, resolution, precursor window, and accuracy). Lithiated estolide-branching isomers and lithiated/sodiated glycerol sn-regioisomers were activated by CID (NCE $40\%$). Depending on the sequential workflow, the re-isolated MS2 or MS3 CID product ions reacted with ozone for 1.0 s at NCE of 0–$1\%$. ## 3.1. Mass Spectra of ω-, 10-, and α-Estolide-Branching Regioisomers The FAHFA subunit of TG-EST consists of an HFA esterified (via its hydroxy group) to a second FA. The position of the hydroxy group can range from second (α-)carbon (adjacent to the carboxylic acid moiety) to the final (ω-)carbon on the chain terminus. We investigated ω-, 10-, and α-estolide-branching regioisomers to establish whether their mass spectra provided information characteristic of the estolide ester bond position. When subjected to positive ion electrospray ionization from solutions containing ammonium, lithium, and sodium cations, TG-EST produced abundant [M + NH4]+, [M + Li]+, and [M + Na]+ adducts, respectively. ## 3.1.1. Fragmentation of Ammonium Adducts Ammonium formate and acetate are relatively volatile salts that are widely used in ESI and HPLC/MS. Ammonium adducts have been broadly adopted for the analysis neutral lipids, e.g., TGs [56,57], and they have also been utilized in the structural analysis of TG-ESTs [6,7,8,9,10,11,25,37]. Activation of these adducts typically leads to the neutral loss of ammonia, yielding product ions arising from proton transfer to the neutral lipid. MS2 HCD spectra of TG-EST ammonium adducts yielded fragments consistent with the neutral loss of ammonia combined with FA and FAHFA. As illustrated in Figure 2, upon activation, ammonium adducts TG-EST 1–3 (m/z 1100.98; C69H130O8N+) were determined to eliminate ammonia and palmitoleic acid (m/z 829.73; C53H97O6+) and ammonia and the FAHFA subunit, i.e., ester of oleic and hydroxy palmitic acid (m/z 547.47; C35H63O4+). The elimination of oleic acid bound in the FAHFA subunit proceeded only for TG-EST 3 (m/z 801.70; C51H93O6+). In this isomer [10-], eliminations of FA linked to glycerol and bound in the FAHFA moiety were similarly efficient. Neutral loss of FAs from FAHFA moiety was documented previously for TG-EST with estolide-branching site on “inner” carbons of HFA chain [6,7]. In addition to neutral loss peaks, FAHFA acylium ion (m/z 519.48; C34H63O3+) was present at low abundance in TG EST 1 and TG-EST 2. In the α-isomer (TG-EST 2), m/z 491.48 (C33H63O2+) was probably formed from [M + H]+ (m/z 1083.95; C69H127O8+) by a cleavage of C1-C2 bond in HFA weakened by the esterified α-OH group. Oleic acid acylium ion (m/z 265.25; C18H33O+) likely originated from this ion after proton rearrangement, analogously to forming FA acylium ions from TGs [58] (p. 109). The abundant acylium ion made it possible to identify FA in the FAHFA subunit. As the acylium ions were formed only from the TG-EST 2 (Figure 2B) and TG-EST 5 (Figure S88B), they can be considered diagnostic ions for the α-estolide-branching regioisomers. Mass spectra MS2 HCD of ammonium adducts of TG-EST 4–8 are presented in Supplementary Materials (Figure S88). The ion trap CID spectra contained fragments identical to HCD spectra, but the low mass cut-off in the ion trap prevented detection of product ions of m/z < 350 (Figure S89). Therefore, HCD was more useful for TG-EST structure elucidation than CID. Energy-resolved dissociation curves of TG-EST ammonium adducts helped us to determine the optimal value of NCE for MS2 experiments. As shown in Figures S76–S79, the optimum NCEs in CID were almost the same for TG-EST 1–3 (NCE 30–35). In HCD, the optimum NCE values ensuring the detection of all diagnostic product ions were dependent on the TG-EST structure. While TG-EST 1 required NCE values above 20, TG-EST 2 fragmented best at NCE around 15, and TG-EST 3 needed an NCE value set to approximately 10. This observation is consistent with differences in the energetics of competing dissociation pathways across the isomeric structures. MS3 offered additional insight into the TG-EST structure. CID spectra of the first generator product ions at m/z 829.73 (Figure S90A–C) differed significantly for the investigated isomers. Under these conditions, all isomers eliminated the second palmitoleic acid (m/z 575.50; C37H67O4+) from the glycerol backbone and the oleic acid from the FAHFA subunit (m/z 547.47; C35H63O4+). Unlike the other isomers, the ω-isomer (TG-EST 1) provided a strong FAHFA acylium ion signal (m/z 519.48; C34H63O3+) along with its dehydration product ions (m/z 501.47; m/z 483.46). The spectrum of α-isomer (TG-EST 2) was characterized by the oleic acid acylium ion (m/z 265.25; C18H33O+) and neutral loss of palmitic acid ketene (m/z 593.51; C37H69O5+). MS3 of m/z 829.73 thus made it possible to distinguish the individual isomers. CID spectra of m/z 547.47 (Figure S90D–F) showed less significant differences in peak intensities, making these fragmentation patterns less likely to be useful for discriminating between isomers in a complex biological mixture. In theory, further information about the estolide structure could also be obtained by re-isolation and further dissociation of FAHFA acylium ion at m/z 519.48, but in practice, its abundance was too low to get measurable diagnostic product ion signals. The MS2 HCD spectra of TG-EST ammonium adducts made it possible to characterize FAHFA subunit and FAs linked to glycerol. FA bound in the FAHFA subunit was detectable as an acylium ion in α-isomer and as a neutral loss in the 10-isomer. Therefore, the spectra allowed us to distinguish estolide-branching isomers and characterize all four FA chains in TG-EST, except for ω-isomers, for which the spectra did not provide information about individual chains in the FAHFA moiety (see spectra of TG-EST 4 and TG-EST 6 in Figure S88A,C, respectively). Determining the estolide-branching position in MS2 was thus possible for α- and 10-isomers. MS3 allowed us to obtain additional structural information; characteristic fragments indicating α- and ω-isomers were identified. ## 3.1.2. Fragmentation of Lithium Adducts Under CID and HCD conditions, lithium adducts of neutral lipids most often provide more structural information than the corresponding sodium or ammonium adducts due to the high oxygen affinity of the lithium [59,60,61]. Figure 3 shows MS2 HCD spectra of lithiated TG-EST 1–3 (m/z 1089.96; C69H126O8Li+). Some of the fragmentation pathways observed in these spectra resemble those observed for the ammonium adducts. Lithiated TG-EST eliminated palmitoleic acids linked to the glycerol backbone (m/z 835.74; C53H96O6Li+) and the FAHFA moiety both as the neutral acid (m/z 553.48; C35H62O4Li+) and its lithium salt (m/z 547.47; C35H63O4+). Eliminating oleic acid from the FAHFA subunit (m/z 807.70; C51H92O6Li+) occurred only from the 10-isomer (TG-EST 3), with an efficiency comparable to the neutral loss of glycerol-linked FA. HCD of lithium adducts (Figure 3) also opened dissociation channels that were not observed for ammonium adducts. The FAHFA moiety provided an abundant lithiated fragment m/z 543.50 (C34H64O4Li+), which offered the possibility to characterize FAHFA subunit in the subsequent MS3 step (see below). Lithium adducts of TG-EST 1 (ω-) and 2 (α-) eliminated both glycerol-linked FAs (m/z 583.53; C37H68O4Li+) while this pathway was almost absent in the TG-EST 3 (10-isomer). These product ions likely arise following the elimination of the first palmitoleic acid by charge-remote fragmentation, as previously suggested for TGs [59] (see in Supplementary Materials, Figure S91). The acylium ion of palmitoleic acid originally bound to glycerol (m/z 237.22; C16H29O+) was detected as a small peak in all three TG-EST spectra. For α- and 10-isomers, lithiated palmitoleic acid (m/z 261.24; C16H30O2Li+) and lithiated oleic acid (m/z 289.27; C18H34O2Li+) ions were formed. The ion m/z 279.25 (C16H32O3Li+), corresponding to the lithium adduct of the HFA, was observed in the spectrum of α-isomer TG-EST 2 (Figure 3B) and TG-EST 5 (Figure S92B). This product ion was rationalized as resulting from the neutral loss of the ketene of oleic acid from m/z 543.50 that is the lithium adduct of FAHFA subunit. This proposal was confirmed by the explicit MS3 of m/z 543.50 shown in Figure 4B. This product ion was determined to be diagnostic of α-estolide-branching regioisomers as evidenced by the wider test-set of HCD spectra of TG-EST 4–8 lithium adducts that are presented in Figure S92. The ion-trap CID mass spectra of lithium adduct ions yielded fragments identical to those of HCD, with the exception of the absence of low mass ions (Figure S93). The energy-resolved dissociation curves of lithiated TG-EST (Figures S80–S83) displayed an optimal NCE for MS2 HCD experiments of 33. The MS3 experiments with lithiated FAHFA (m/z 543.50) were performed with the preceding ion trap-CID step providing a more abundant precursor ion than HCD. The MS3 (CID/CID) spectra of ω-, α-, and 10-isomers (TG-EST 1–3) are shown in Figure 4. All the spectra provided lithiated oleic acid (m/z 289.27; C18H34O2Li+) and lithiated FAHFA after the elimination of neutral oleic acid (m/z 261.24; C16H30O2Li+) with the latter ion corresponding to the lithium adduct of dehydrated HFA. The α-isomer MS2 spectrum (Figure 4B) showed abundant m/z 279.25 (C16H32O3Li+), the same ion already observed in MS2. Since the ion was negligibly small in ω- and absent in 10-isomers, it could be considered a diagnostic ion for α-estolide-branching regioisomers that carries information about HFA component of the FAHFA unit. This peak was accompanied by low abundant product ion at m/z 233.25 (C15H30OLi+), differing from it by the mass of CO2. Compared to ammonium adducts, the CID and HCD mass spectra of the lithium adducts allowed a deeper insight into the structure of TG-EST. Abundant lithiated FAHFA species made it possible to determine FA and HFA in the FAHFA moiety for all estolide-branching regioisomers. As in the case of ammonium adducts, clear differentiation of the α-isomers from the others was possible. In summary, the MS2 HCD and MS3 CID/CID spectra of lithiated FAHFA made it possible to characterize the FAHFA subunit, i.e., to identify FA, HFA, and the α-estolide-branching site. OzID has previously been proven to be a useful ion activation method for the structural analysis of lipids and in particular for the resolution of lipid regioisomers [47,48,49,50,51,62]. This technique harnesses ozonolysis reactions of mass-selected ions within the mass spectrometer to identify the location of carbon–carbon double bonds that are either present in the native lipid structure or are formed during preceding ion activation events. As ozonolysis is selective for carbon–carbon double bonds, only highly-specific fragmentations are induced that can provide for unambiguous structure elucidation [45,46]. We employed ozonolysis in a sequential CID/CID/OzID workflow to differentiate estolide-branching site of isomers. The MS4 spectra of lithiated TG-EST 1–3 are shown in Figure S94. Elimination of FA from FAHFA moiety yields dehydration products [49,63]. In the case of lithiated TG-EST 3 (10-isomer), the CID product ion m/z 261.24 is a mixture of lithiated 9- and 10-Hexadecenoic acids. Their ozonolysis in the final OzID step provided aldehyde ions m/z 179.12 (C9H16O3Li+) and m/z 193.14 (C10H18O3Li+) which unambiguously identified n-7 and n-6 carbon–carbon double bond positions and thus also the estolide-branching site (Figure S94C). The product ion m/z 263.22 (C15H28O3Li+) diagnostic for ω-estolide-branching site was observed in the spectrum of TG-EST 1 (Figure S94A). OzID did not provide any diagnostic fragment for the α-estolide-branching site in TG-EST 2 (Figure S94B). ## 3.1.3. Fragmentation of Sodium Adducts Sodium adducts are readily formed from neutral lipids [64], including TG-EST [25,36,41], and provide informative fragmentation spectra. The MS2 HCD spectra of sodiated TG-EST 1–3 (m/z 1105.94; C69H126O8Na+) closely resembled spectra of lithium adducts (Figure 5). All estolide-branching regioisomers easily eliminated glycerol-linked palmitoleic acid (m/z 851.71; C53H96O6Na+) and readily formed sodiated FAHFA (m/z 559.47; C34H64O4Na+). Elimination of palmitoleic acid sodium salt (m/z 829.73; C53H97O6+) and combined loss of both glycerol-linked palmitoleic acids (m/z 599.50; C37H68O4Na+) yielded minor signals. While the neutral loss of FAHFA moiety as neutral acid (m/z 569.45; C35H62O4Na+) or sodium salt (m/z 547.47; C35H63O4+) proceeded inefficiently in ω- (TG-EST 1), it was a favored process in α- (TG-EST 2) and 10- (TG-EST 3) isomers. Neutral loss of oleic acid from the FAHFA subunit occurred only from the 10-isomer (TG-EST 3), yielding abundant fragment m/z 823.68 (C51H92O6Na+). In the low mass range region, acylium ions of glycerol-linked palmitoleic acid (m/z 237.22; C16H29O+) and HFA-linked oleic acid (m/z 265.25; C18H33O+) were detected, alongside with sodium adducts of these FAs (m/z 277.21; C16H30O2Na+ and m/z 305.25; C18H34O2Na+, respectively). The fragment m/z 295.22 (C16H32O3Na+) in the spectrum of the α-isomer (TG-EST 2) corresponded to neutral loss of ketene of oleic acid from m/z 559.47 (FAHFA subunit sodium adduct). The HCD spectra of TG-EST 4–8 sodium adducts are presented in Supplementary Materials (Figure S95). The HCD spectra of sodium adducts provided approximately the same structural information as lithium adducts. The ion trap CID of TG-EST sodium adducts showed fragments identical to HCD, with some ions missing due to the low mass cut-off in the ion trap (Figure S96). Further fragmentation of sodiated ions by collisional activation turned out to be more challenging than for lithium with great competition for dissociation to the bare cation upon CID and thus leading to a lower abundance of lipid-related product ions. ## 3.2. Mass Spectra of Glycerol sn-Regioisomers Since the neutral loss of FAs from the sn-1 and sn-3 positions of TGs are equally favored and more competitive than losses from sn-2, the relative abundance of the resulting diacylglycerol-like product ions can be used to infer the regiochemical assignment [59,65]. Here, we compared MS2 CID and HCD spectra of TG-EST 7 and TG-EST 8 differing by the relative position of FAHFA moiety on the glycerol backbone. As follows from the energy-resolved dissociation curves for [M + NH4]+, [M + Li]+, and [M + Na]+ (Figures S79, S83 and S87, respectively), CID provided a relatively stable ratio of fragment peak intensities across a wide range of NCE values (typically NCE 30–35). This was not the case for HCD, where a small change in NCE caused significant changes in product ion abundance. Therefore, CID was used to investigate the relative intensity ratios of peaks corresponding to the neutral loss of palmitic acid and FAHFA units from their positions on glycerol (Figure 6). All the spectra showed a greater peak intensity corresponding to the elimination of palmitic acid (m/z 859.77, C55H103O6+ for [M + NH4]+; m/z 865.78, C55H102O6Li+ for [M + Li]+; and m/z 881.76, C55H102O6Na+ for [M + Na]+), consistent with the presence of two palmitic acid radyls in each isomer. In contrast, the relative abundance of the product ions corresponding to the neutral loss of FAHFA (m/z 551.50, C35H67O4+, for [M + NH4]+) or lithiated/sodiated FAHFA (m/z 571.53, C36H68O4Li+ for [M + Li]+; and m/z 587.50, C36H68O4Na+ for [M + Na]+), while variable across the adducts, was consistently higher where the FAHFA was esterified at sn-$\frac{1}{3}$ (compared to FAHFA linked to sn-2). The effect was most pronounced in the case of ammonium adducts (Figure 6A,D). Therefore, the relative abundance of these diagnostic product ions consistently reflected the FAHFA position on the glycerol backbone. Relying on product ion abundances can lead to ambiguous regiochemical assignments particularly where standard reference materials are not available or in the case of complex samples that may comprise isomeric mixtures. Therefore, we searched for product ions specific to the backbone regiochemistry of TG-EST subunit on glycerol. Previous investigations have shown that CID of lithium and sodium adduct ions of TGs [50] lead to a five-membered 1,3-dioxolane ring and neutral the loss of one FA. The adjacent FA remains tethered to the glycerol by a newly formed carbon–carbon double bond that can react with ozone and reveal the adjacency of the two FA on the backbone. Analogous MS3 CID/OzID workflow was used for TG-EST 7 and TG-EST 8. Activation of lithium and sodium adducts in CID resulted in a neutral loss of the palmitic acid originally linked to the glycerol backbone, yielding product ions at m/z 865.78 (C55H102O6Li+) and m/z 881.76 (C55H102O6Na+), respectively. When exposed to ozone, the lithiated product ion provided MS3 (CID/OzID) spectra shown in Figure 7. For both sn-isomers, an ozonide at m/z 913.77 (C55H102O9Li+) and its oxygen elimination product (m/z 881.77; C55H102O7Li+) were formed. Oxidative cleavage across the nascent carbon–carbon double bond led to Criegee, aldehyde and carbonate ester product ions indicative of the neutral loss of the FA and/or FAHFA moiety adjacent to the palmitic acid and thus a fragmentation pattern suitable for distinguishing the sn-regioisomers (Figure 7). The TG-EST 7, with the FAHFA moiety in the sn-$\frac{1}{3}$ position, provided a pair of Criegee and carbonate ester fragments at m/z 687.54 (C40H72O8Li+) and m/z 671.54 (C40H72O7Li+), respectively. On the contrary, lithiated TG-EST 8 with the FAHFA moiety in the sn-2 decomposed to two pairs of diagnostic ions, (i) Criegee fragment m/z 557.51 (C35H66O4Li+) and aldehyde ion m/z 541.51 (C35H66O3Li+), and (ii) Criegee fragment m/z 379.27 (C20H36O6Li+) accompanied by carbonate ester m/z 363.27 (C20H36O5Li+). An analogous fragmentation was observed for sodiated TG-EST isomers (Figure 8). CID/OzID mass spectra of both sodiated and lithiated adduct ions showed the presence of low abundant product ions that were inconsistent with either the putative lipid regiochemistry or the exclusivity of the dissociation mechanisms indicated in Figure 7 and Figure 8, for example, the presence of a small amount of m/z 671.54 in Figure 7B and m/z 687.52 in Figure 8B. These ions point to either a small amount of the alternate regioisomer, perhaps arising from transacylation during synthesis, or the participation of an alternative, but poorly competitive, dissociation pathway. Future work to combine multi-stage ion activation with chromatographic resolution of isomers would be a suitable means to address this ambiguity. ## 4. Discussion High-resolution tandem mass spectrometry of ammonium, lithium, and sodium adducts provided important pieces of information on the structure of TG-EST. By selecting the ion type and activation method, it was possible to distinguish TG-EST isomeric species and characterize their structural features. Lithium was determined to be the most useful cationization agent. Lithiated species provided more informative CID and HCD spectra than sodium and ammonium adducts. In the MS2, ions permitting the identification of FA, HFA, and FAHFA subunits were present, and some of the lithiated fragments allowed us to characterize branching of the FAHFA moiety. In the MS3, all isomers showed loss of FA from lithiated FAHFA ions; abundant ions corresponding to a loss of ketene of FA were additionally present in the α-isomers. The MS4 with OzID as the last fragmentation step allowed an even more detailed characterization of the estolide-branching site in FAHFA. In this experiment, ozonolysis was used to determine the position of carbon–carbon double bonds formed after elimination of FA from lithiated FAHFA. A similar level of structural information was obtained with sodiated TG-EST. Sodiated species were, however, difficult to fragment in subsequent MS steps because sodium tended to dissociate as bare cation during collisions. Despite their easy formation, ammonium adducts proved less useful as they did not provide FAHFA fragments that could be activated further. All the adducts allowed us to distinguish between sn-$\frac{1}{3}$ and sn-2 TG-EST isomers. The relative abundance of CID fragments corresponding to neutral loss of FAHFA and FA (ammonium adducts) or lithiated/sodiated FAHFA and neutral loss of FA was significantly higher for sn-$\frac{1}{3}$ isomers. The effect was pronounced the most in the case of ammonium adducts. The glycerol sn-regioisomers were also distinguished by CID/OzID of lithium or sodium adducts. Specific peaks in the spectra made it possible to identify the regioisomers unambiguously. The fragmentation pathways described in this work are the basis for developing new methods for analyzing TG-EST in biological samples using HPLC/MS. A high-resolution mass spectrometer was used to understand the fragmentation pathways. High-resolution instruments are also recommended for analyzing biological samples; however, reliable identification of TG-EST can also be achieved using devices with lower resolving power. CID in QqQ or QTOF-type instruments can be expected to provide similar spectra to HCD discussed in this work. We worked with a concentration of standards of 10 µmol L−1. Spectra of similar quality can, however, be obtained at significantly lower concentrations (Figure S97) that are biologically relevant (concentration of TG-EST in the HPLC/MS samples are typically in µmol L−1 range). To comprehensively analyze TG-ESTs in biological matrices, it will be necessary to combine mass spectrometry with efficient chromatography. The HPLC separation of isomeric lipids is a great challenge, especially when it comes to very complex lipid matrices, such as vernix caseosa. ## References 1. Brejchova K., Balas L., Paluchova V., Brezinova M., Durand T., Kuda O.. Understanding FAHFAs: From structure to metabolic regulation. Prog. Lipid Res. (2020) 79. DOI: 10.1016/j.plipres.2020.101053 2. Riecan M., Paluchova V., Lopes M., Brejchova K., Kuda O.. Branched and linear fatty acid esters of hydroxy fatty acids (FAHFA) relevant to human health. Pharmacol. Ther. (2021) 231. DOI: 10.1016/j.pharmthera.2021.107972 3. Vavrušová A., Vrkoslav V., Plavka R., Bosáková Z., Cvačka J.. Analysis of (O-acyl) alpha-and omega-hydroxy fatty acids in vernix caseosa by high-performance liquid chromatography-Orbitrap mass spectrometry. Anal. Bioanal. Chem. (2020) 412 2291-2302. DOI: 10.1007/s00216-019-02348-2 4. Chen Y., Biresaw G., Cermak S.C., Isbell T.A., Ngo H.L., Chen L., Durham A.L.. Fatty Acid Estolides: A Review. J. Am. Oil Chem. Soc. (2020) 97 231-241. DOI: 10.1002/aocs.12323 5. McLean S., Davies N.W., Nichols D.S., McLeod B.J.. Triacylglycerol estolides, a new class of mammalian lipids, in the paracloacal gland of the brushtail possum (. Lipids (2015) 50 591-604. DOI: 10.1007/s11745-015-4025-9 6. Tan D., Ertunc M.E., Konduri S., Zhang J., Pinto A.M., Chu Q., Kahn B.B., Siegel D., Saghatelian A.. Discovery of FAHFA-Containing Triacylglycerols and Their Metabolic Regulation. J. Am. Chem. Soc. (2019) 141 8798-8806. DOI: 10.1021/jacs.9b00045 7. Brejchova K., Radner F.P.W., Balas L., Paluchova V., Cajka T., Chodounska H., Kudova E., Schratter M., Schreiber R., Durand T.. Distinct roles of adipose triglyceride lipase and hormone-sensitive lipase in the catabolism of triacylglycerol estolides. Proc. Natl. Acad. Sci. USA (2021) 118. DOI: 10.1073/pnas.2020999118 8. Paluchova V., Oseeva M., Brezinova M., Cajka T., Bardova K., Adamcova K., Zacek P., Brejchova K., Balas L., Chodounska H.. Lipokine 5-PAHSA Is Regulated by Adipose Triglyceride Lipase and Primes Adipocytes for De Novo Lipogenesis in Mice. Diabetes (2020) 69 300-312. DOI: 10.2337/db19-0494 9. Brezinova M., Cajka T., Oseeva M., Stepan M., Dadova K., Rossmeislova L., Matous M., Siklova M., Rossmeisl M., Kuda O.. Exercise training induces insulin-sensitizing PAHSAs in adipose tissue of elderly women. Biochim. Biophys. Acta (BBA)-Mol. Cell Biol. Lipids (2020) 1865. DOI: 10.1016/j.bbalip.2019.158576 10. Saponara E., Penno C., Orsini V., Wang Z.-Y., Fischer A., Aebi A., Matadamas-Guzman M.L., Brun V., Fischer B., Brousseau M.. Loss of Hepatic Leucine-Rich Repeat-Containing G-protein Coupled Receptors 4 and 5 Promotes Nonalcoholic Fatty Liver Disease. Am. J. Pathol. (2023) 193 161-181. DOI: 10.1016/j.ajpath.2022.10.008 11. Brejchova K., Paluchova V., Brezinova M., Cajka T., Balas L., Durand T., Krizova M., Stranak Z., Kuda O.. Triacylglycerols containing branched palmitic acid ester of hydroxystearic acid (PAHSA) are present in the breast milk and hydrolyzed by carboxyl ester lipase. Food Chem. (2022) 388. DOI: 10.1016/j.foodchem.2022.132983 12. Sprecher H.W., Maier R., Barber M., Holman R.T.. Structure of an optically active allene-containing tetraester triglyceride isolated from the seed oil of. Biochemistry (1965) 4 1856-1863. DOI: 10.1021/bi00885a024 13. Mikolajczak K.L., Smith C.R., Wolff I.A.. Glyceride structure of. Lipids (1968) 3 215-220. DOI: 10.1007/BF02531189 14. Christie W.W.. The glyceride structure of sapium sebiferum seed oil. Biochim. Biophys. Acta (BBA)-Lipids Lipid Metab. (1969) 187 1-5. DOI: 10.1016/0005-2760(69)90125-8 15. Phillips B.E., Smith C.R.. Glycerides of. Biochim. Biophys. Acta (BBA)-Lipids Lipid Metab. (1970) 218 71-82. DOI: 10.1016/0005-2760(70)90093-7 16. Kleiman R., Spencer G.F., Earle F.R., Nieschlag H.J., Barclay A.S.. Tetra-acid triglycerides containing a new hydroxy eicosadienoyl moiety in. Lipids (1972) 7 660-665. DOI: 10.1007/BF02533073 17. Payne-Wahl K., Plattner R.D., Spencer G.F., Kleiman R.. Separation of tetra-, penta-, and hexaacyl triglycerides by high performance liquid chromatography. Lipids (1979) 14 601-605. DOI: 10.1007/BF02533443 18. Plattner R.D., Payne-Wahl K., Tjarks L.W., Kleiman R.. Hydroxy acids and estolide triglycerides of. Lipids (1979) 14 576-579. DOI: 10.1007/BF02533535 19. Madrigal R.V., Smith C.R.. Estolide triglycerides of. Lipids (1982) 17 650-655. DOI: 10.1007/BF02535373 20. Payne-Wahl K.. Quantitation of estolide triglycerides in sapium seeds by high performance liquid chromatography with infrared detection. J. Am. Oil Chem. Soc. (1983) 60 1011-1012. DOI: 10.1007/BF02660218 21. Aitzetmüller K., Xin Y., Werner G., Grönheim M.. High-performance liquid chromatographic investigations of stillingia oil. J. Chromatogr. A (1992) 603 165-173. DOI: 10.1016/0021-9673(92)85358-Z 22. Hayes D.G., Kleiman R., Phillips B.S.. The triglyceride composition, structure, and presence of estolides in the oils of Lesquerella and related species. J. Am. Oil Chem. Soc. (1995) 72 559-569. DOI: 10.1007/BF02638857 23. Hayes D.G., Kleiman R., Weisleder D., Adlof R.O., Cuperus F.P., Derksen J.T.P.. Occurrence of estolides in processed. Ind. Crops Prod. (1995) 4 295-301. DOI: 10.1016/0926-6690(95)00045-3 24. Spitzer V., Tomberg W., Pohlentz G.. Structure analysis of an allene-containing estolide tetraester triglyceride in the seed oil of. Lipids (1997) 32 549-557. DOI: 10.1007/s11745-997-0070-8 25. Lin J.-T., Arcinas A., Harden L.R., Fagerquist C.K.. Identification of (12-Ricinoleoylricinoleoyl)diricinoleoylglycerol, an Acylglycerol Containing Four Acyl Chains, in Castor (. J. Agric. Food Chem. (2006) 54 3498-3504. DOI: 10.1021/jf060150e 26. Lin J.-T., Arcinas A.. Regiospecific Identification of 2-(12-Ricinoleoylricinoleoyl)-1,3-diricinoleoyl-sn-glycerol in Castor (. J. Agric. Food Chem. (2008) 56 3616-3622. DOI: 10.1021/jf072750z 27. Lin J.-T., Chen G.Q.. Identification of Minor Acylglycerols Less Polar than Triricinolein in Castor Oil by Mass Spectrometry. J. Am. Oil Chem. Soc. (2012) 89 1773-1784. DOI: 10.1007/s11746-012-2082-2 28. Lin J.-T., Chen G.Q.. Ratios of Regioisomers of Minor Acylglycerols Less Polar than Triricinolein in Castor Oil Estimated by Mass Spectrometry. J. Am. Oil Chem. Soc. (2012) 89 1785-1792. DOI: 10.1007/s11746-012-2083-1 29. Zhang H., Olson D.J.H., Van D., Purves R.W., Smith M.A.. Rapid identification of triacylglycerol-estolides in plant and fungal oils. Ind. Crops Prod. (2012) 37 186-194. DOI: 10.1016/j.indcrop.2011.10.026 30. Lin J.-T., Chen G.Q.. Identification of Tetraacylglycerols in Lesquerella Oil by Electrospray Ionization Mass Spectrometry of the Lithium Adducts. J. Am. Oil Chem. Soc. (2013) 90 1831-1836. DOI: 10.1007/s11746-013-2332-y 31. Smith M.A., Zhang H., Forseille L., Purves R.W.. Characterization of novel triacylglycerol estolides from the seed oil of. Lipids (2013) 48 75-85. DOI: 10.1007/s11745-012-3721-y 32. Lin J.-T., Chen G.. Quantification of the molecular species of tetraacylglycerols in lesquerella (. Enliven Bio Anal. Tech. (2014) 1 1-5 33. Zhang H., Smith M.A., Purves R.W.. Optimization of triacylglycerol-estolide analysis by matrix-assisted laser desorption/ionization-mass spectrometry. J. Am. Oil Chem. Soc. (2014) 91 905-915. DOI: 10.1007/s11746-014-2437-y 34. Smith M.A., Zhang H.. Apocynaceae seed lipids: Characterization and occurrence of isoricinoleic acid and triacylglycerol estolides. J. Am. Oil Chem. Soc. (2016) 93 105-114. DOI: 10.1007/s11746-015-2741-1 35. Lin J.-T., Chen G.Q.. Structural characteristics of the molecular species of tetraacylglycerols in lesquerella (. Biocatal. Agric. Biotechnol. (2017) 10 167-173. DOI: 10.1016/j.bcab.2017.03.005 36. Romsdahl T., Shirani A., Minto R.E., Zhang C., Cahoon E.B., Chapman K.D., Berman D.. Nature-Guided Synthesis of Advanced Bio-Lubricants. Sci. Rep. (2019) 9. DOI: 10.1038/s41598-019-48165-6 37. Matsuzawa Y., Higashi Y., Takano K., Takahashi M., Yamada Y., Okazaki Y., Nakabayashi R., Saito K., Tsugawa H.. Food Lipidomics for 155 Agricultural Plant Products. J. Agric. Food Chem. (2021) 69 8981-8990. DOI: 10.1021/acs.jafc.0c07356 38. Liu Y., Tu X., Lin L., Du L., Feng X.. Analysis of Lipids in Pitaya Seed Oil by Ultra-Performance Liquid Chromatography-Time-of-Flight Tandem Mass Spectrometry. Foods (2022) 11. DOI: 10.3390/foods11192988 39. Morris L.J., Hall S.W.. The structure of the glycerides of ergot oils. Lipids (1966) 1 188-196. DOI: 10.1007/BF02531871 40. Batrakov S.G., Tolkachev O.N.. The structures of triacylglycerols from sclerotia of the rye ergot. Chem. Phys. Lipids (1997) 86 1-12. DOI: 10.1016/S0009-3084(96)02653-9 41. Lucejko J.J., La Nasa J., Porta F., Vanzetti A., Tanda G., Mangiaracina C.F., Corretti A., Colombini M.P., Ribechini E.. Long-lasting ergot lipids as new biomarkers for assessing the presence of cereals and cereal products in archaeological vessels. Sci. Rep. (2018) 8. DOI: 10.1038/s41598-018-22140-z 42. Patel R., Santoro A., Hofer P., Tan D., Oberer M., Nelson A.T., Konduri S., Siegel D., Zechner R., Saghatelian A.. ATGL is a biosynthetic enzyme for fatty acid esters of hydroxy fatty acids. Nature (2022) 606 968-975. DOI: 10.1038/s41586-022-04787-x 43. Nelson A.T., Kolar M.J., Chu Q., Syed I., Kahn B.B., Saghatelian A., Siegel D.. Stereochemistry of Endogenous Palmitic Acid Ester of 9-Hydroxystearic Acid and Relevance of Absolute Configuration to Regulation. J. Am. Chem. Soc. (2017) 139 4943-4947. DOI: 10.1021/jacs.7b01269 44. Yasuda S., Okahashi N., Tsugawa H., Ogata Y., Ikeda K., Suda W., Arita M.. Elucidation of Gut Microbiota-Associated Lipids Using LC-MS/MS and 16S rRNA Sequence Analyses. iScience (2020) 23. DOI: 10.1016/j.isci.2020.101841 45. Thomas M.C., Mitchell T.W., Harman D.G., Deeley J.M., Nealon J.R., Blanksby S.J.. Ozone-Induced Dissociation: Elucidation of Double Bond Position within Mass-Selected Lipid Ions. Anal. Chem. (2008) 80 303-311. DOI: 10.1021/ac7017684 46. Brown S.H.J., Mitchell T.W., Blanksby S.J.. Analysis of unsaturated lipids by ozone-induced dissociation. Biochim. Biophys. Acta (BBA)-Mol. Cell Biol. Lipids (2011) 1811 807-817. DOI: 10.1016/j.bbalip.2011.04.015 47. Poad B.L.J., Marshall D.L., Harazim E., Gupta R., Narreddula V.R., Young R.S.E., Duchoslav E., Campbell J.L., Broadbent J.A., Cvačka J.. Combining Charge-Switch Derivatization with Ozone-Induced Dissociation for Fatty Acid Analysis. J. Am. Soc. Mass Spectrom. (2019) 30 2135-2143. DOI: 10.1007/s13361-019-02285-5 48. Pham H.T., Maccarone A.T., Thomas M.C., Campbell J.L., Mitchell T.W., Blanksby S.J.. Structural characterization of glycerophospholipids by combinations of ozone-and collision-induced dissociation mass spectrometry: The next step towards “top-down” lipidomics. Analyst (2014) 139 204-214. DOI: 10.1039/C3AN01712E 49. Marshall D.L., Saville J.T., Maccarone A.T., Ailuri R., Kelso M.J., Mitchell T.W., Blanksby S.J.. Determination of ester position in isomeric (O-acyl)-hydroxy fatty acids by ion trap mass spectrometry. Rapid Commun. Mass Spectrom. (2016) 30 2351-2359. DOI: 10.1002/rcm.7715 50. Marshall D.L., Pham H.T., Bhujel M., Chin J.S.R., Yew J.Y., Mori K., Mitchell T.W., Blanksby S.J.. Sequential Collision- and Ozone-Induced Dissociation Enables Assignment of Relative Acyl Chain Position in Triacylglycerols. Anal. Chem. (2016) 88 2685-2692. DOI: 10.1021/acs.analchem.5b04001 51. Kozlowski R.L., Mitchell T.W., Blanksby S.J.. Separation and Identification of Phosphatidylcholine Regioisomers by Combining Liquid Chromatography with a Fusion of Collision-and Ozone-Induced Dissociation. Eur. J. Mass Spectrom. (2015) 21 191-200. DOI: 10.1255/ejms.1300 52. Liebisch G., Fahy E., Aoki J.. Update on LIPID MAPS classification, nomenclature, and shorthand notation for MS-derived lipid structures. J. Lipid Res. (2020) 61 1539-1555. DOI: 10.1194/jlr.S120001025 53. Machara A., Křivánek J., Dolejšová K., Havlíčková J., Bednárová L., Hanus R., Kyjaková P.. Identification and Enantiodivergent Synthesis of (5Z,9S)-Tetradec-5-en-9-olide, a Queen-Specific Volatile of the Termite Silvestritermes minutus. J. Nat. Prod. (2018) 81 2266-2274. DOI: 10.1021/acs.jnatprod.8b00632 54. Tupec M., Culka M., Machara A., Macháček S., Bím D., Svatoš A., Rulíšek L., Pichová I.. Understanding desaturation/hydroxylation activity of castor stearoyl Δ9-Desaturase through rational mutagenesis. Comput. Struct. Biotechnol. J. (2022) 20 1378-1388. DOI: 10.1016/j.csbj.2022.03.010 55. Poad B.L., Young R.S., Marshall D.L., Trevitt A.J., Blanksby S.J.. Accelerating Ozonolysis Reactions Using Supplemental RF-Activation of Ions in a Linear Ion Trap Mass Spectrometer. Anal. Chem. (2022) 94 3897-3903. DOI: 10.1021/acs.analchem.1c04915 56. McAnoy A.M., Wu C.C., Murphy R.C.. Direct qualitative analysis of triacylglycerols by electrospray mass spectrometry using a linear ion trap. J. Am. Soc. Mass Spectrom. (2005) 16 1498-1509. DOI: 10.1016/j.jasms.2005.04.017 57. Dorschel C.A.. Characterization of the TAG of peanut oil by electrospray LC-MS-MS. J. Am. Oil Chem. Soc. (2002) 79 749-753. DOI: 10.1007/s11746-002-0553-z 58. Murphy R.C.. Tandem Mass Spectrometry of Lipids: Molecular Analysis of Complex Lipids (2015) 59. Hsu F.F., Turk J.. Structural characterization of triacylglycerols as lithiated adducts by electrospray ionization mass spectrometry using low-energy collisionally activated dissociation on a triple stage quadrupole instrument. J. Am. Soc. Mass Spectrom. (1999) 10 587-599. DOI: 10.1016/S1044-0305(99)00035-5 60. Tran A., Wan L., Xu Z., Haro J.M., Li B., Jones J.W.. Lithium Hydroxide Hydrolysis Combined with MALDI TOF Mass Spectrometry for Rapid Sphingolipid Detection. J. Am. Soc. Mass Spectrom. (2021) 32 289-300. DOI: 10.1021/jasms.0c00322 61. Hsu F.-F., Bohrer A., Turk J.. Formation of lithiated adducts of glycerophosphocholine lipids facilitates their identification by electrospray ionization tandem mass spectrometry. J. Am. Soc. Mass Spectrom. (1998) 9 516-526. DOI: 10.1016/S1044-0305(98)00012-9 62. Young R.S.E., Flakelar C.L., Narreddula V.R., Jekimovs L.J., Menzel J.P., Poad B.L.J., Blanksby S.J.. Identification of Carbon-Carbon Double Bond Stereochemistry in Unsaturated Fatty Acids by Charge-Remote Fragmentation of Fixed-Charge Derivatives. Anal. Chem. (2022) 94 16180-16188. DOI: 10.1021/acs.analchem.2c03625 63. Buenger E.W., Reid G.E.. Shedding light on isomeric FAHFA lipid structures using 213 nm ultraviolet photodissociation mass spectrometry. Eur. J. Mass Spectrom. (2020) 26 311-323. DOI: 10.1177/1469066720960341 64. Herrera L.C., Potvin M.A., Melanson J.E.. Quantitative analysis of positional isomers of triacylglycerols via electrospray ionization tandem mass spectrometry of sodiated adducts. Rapid Commun. Mass Spectrom. (2010) 24 2745-2752. DOI: 10.1002/rcm.4700 65. Laakso P.. Mass spectrometry of triacylglycerols. Eur. J. Lipid Sci. Technol. (2002) 104 43-49. DOI: 10.1002/1438-9312(200201)104:1<43::AID-EJLT43>3.0.CO;2-J
--- title: Quantitative Biomarkers Derived from a Novel Contrast-Free Ultrasound High-Definition Microvessel Imaging for Distinguishing Thyroid Nodules authors: - Melisa Kurti - Soroosh Sabeti - Kathryn A. Robinson - Lorenzo Scalise - Nicholas B. Larson - Mostafa Fatemi - Azra Alizad journal: Cancers year: 2023 pmcid: PMC10046818 doi: 10.3390/cancers15061888 license: CC BY 4.0 --- # Quantitative Biomarkers Derived from a Novel Contrast-Free Ultrasound High-Definition Microvessel Imaging for Distinguishing Thyroid Nodules ## Abstract ### Simple Summary Low specificity of ultrasound in detecting thyroid cancer warrants the development of new noninvasive modalities for the optimal characterization of thyroid nodules. Here, we present a new ultrasound-based technique, high-definition microvasculature imaging (HDMI) that provides quantitative measures of tumor microvasculature morphological features as new imaging biomarkers. This technique utilizes vessel enhancement filtering, morphological filtering, and vessel segmentation, which enable extraction of vessel morphological features including tortuosity, vessel density, diameter, Murray’s deviation, microvessel fractal dimension, bifurcation angle, number of branch points, and vessel segments. Without the help of contrast agents, through the utilization of HDMI on patients with suspicious thyroid nodules, we were able to resolve tumor microvessels at size scales of a few hundred microns. We further showed that analysis of tumor vessel morphological parameters could detect thyroid malignancy with high sensitivity and specificity. These findings provide a translational rationale for the clinical implementation of quantitative HDMI for thyroid cancer detection. ### Abstract Low specificity in current ultrasound modalities for thyroid cancer detection necessitates the development of new imaging modalities for optimal characterization of thyroid nodules. Herein, the quantitative biomarkers of a new high-definition microvessel imaging (HDMI) were evaluated for discrimination of benign from malignant thyroid nodules. Without the help of contrast agents, this new ultrasound-based quantitative technique utilizes processing methods including clutter filtering, denoising, vessel enhancement filtering, morphological filtering, and vessel segmentation to resolve tumor microvessels at size scales of a few hundred microns and enables the extraction of vessel morphological features as new tumor biomarkers. We evaluated quantitative HDMI on 92 patients with 92 thyroid nodules identified in ultrasound. A total of 12 biomarkers derived from vessel morphological parameters were associated with pathology results. Using the Wilcoxon rank-sum test, six of the twelve biomarkers were significantly different in distribution between the malignant and benign nodules (all $p \leq 0.01$). A support vector machine (SVM)-based classification model was trained on these six biomarkers, and the receiver operating characteristic curve (ROC) showed an area under the curve (AUC) of 0.9005 ($95\%$ CI: [0.8279,0.9732]) with sensitivity, specificity, and accuracy of 0.7778, 0.9474, and 0.8929, respectively. When additional clinical data, namely TI-RADS, age, and nodule size were added to the features, model performance reached an AUC of 0.9044 ($95\%$ CI: [0.8331,0.9757]) with sensitivity, specificity, and accuracy of 0.8750, 0.8235, and 0.8400, respectively. Our findings suggest that tumor vessel morphological features may improve the characterization of thyroid nodules. ## 1. Introduction The incidence of thyroid cancer, the most prevalent endocrine cancer worldwide, has increased in the last few decades [1,2]. Furthermore, thyroid nodules are commonly found in routine physical examinations or as incidental findings on diagnostic imaging performed for other non-thyroidal indications [3]. Clinical examination via palpation is subjective for the detection of thyroid nodules and depends on the experience of the examining clinician, as well as the size and location of the nodule [4]. Although ultrasonography is the first-line imaging tool in evaluating thyroid nodules, the high sensitivity for identifying nodules but unsatisfactory specificity for cancer detection results in overwhelming benign fine needle aspiration biopsy (FNAB), about 60–$80\%$ [5]. While FNAB is a widely used and safe procedure, complications such as discomfort or local pain and self-limited small hematomas may occur [6]. The low specificity of ultrasound features for classifying thyroid nodules leads to unnecessary FNABs; therefore, new imaging modalities with special attention to anarchical angiogenesis observed in malignancy are of paramount importance in the characterization of nodules. The addition of strain elastography [7,8], shear wave elastography [9,10,11,12,13], and multimodality ultrasound combining B-mode ultrasound, elastography, and contrast-enhanced ultrasound (CEUS) [14] relatively increased the sensitivity and specificity of ultrasound. Thyroid nodules exhibit different patterns of blood flow and vascular morphology that are useful in separating malignant nodules from benign ones [15]. Furthermore, angiogenic activity and sprouting angiogenesis are crucial to thyroid cancer progression [16]. As a consequence of neovascularization, the hallmark of cancer, the microvessel structures in malignant nodules look quite different from those of benign [17]. Imaging modalities that could image and quantify the morphology of tumor microvessels can facilitate cancer diagnosis. Patterns of vascularity, intra-nodular with absent or insignificant perinodular blood flow shown in color Doppler ultrasound and power Doppler ultrasound (PDUS) suggest malignancy [18,19]. With recent advances in slow blood flow imaging, attempts have begun to non-invasively image a tumor’s microvessel structures. With the help of contrast agents, acoustic angiography enables high-resolution imaging of microvasculature [20]. Recently, superb microvascular imaging (SMI) has reported the added value of its microvessel imaging with [21] and without [22,23] TI-RADS for distinguishing benign and malignant thyroid nodules; however, this method is based on visual inspection and quantification is limited to pixel counting. A newly developed contrast-free ultrasound-based modality has been introduced to visualize microvessels at a submillimeter level, about 300 μm in diameter [24], labeling it as high-definition microvessel imaging (HDMI) [25]. A series of morphological filtering and vessel enhancement has complemented the HDMI approach to quantify tumor vessel morphological parameters as quantitative vessel biomarkers [24,26,27]. Quantitative HDMI has been tested for distinguishing malignant breast masses from benign ones with remarkable results [25,28]. The objective of the present study is to evaluate the performance of our proposed two-dimensional (2D) quantitative microvessel imaging in classifying malignant and benign thyroid nodules. In this study, microvessel morphological features of thyroid nodules (vessel diameter, tortuosity, vascular density, number of branch points, number of vessel segments, microvessel fractal dimension, bifurcation angle, and Murray’s deviation) are used as HDMI biomarkers. We hypothesized that the aforementioned quantitative biomarkers obtained by HDMI could enhance the differentiation of malignant and benign thyroid nodules with higher specificity. Additionally, significant vessel biomarkers are used to create a model capable of classifying the thyroid nodules as benign or malignant. ## 2.1. Patient Study This study was performed in compliance with the Health Insurance Portability and Accountability Act (HIPAA) and under the guidelines and regulations of an approved institutional review board (IRB) protocol (IRB#: 08-008778). A written IRB-approved informed consent with permission for publication was obtained from each patient participant prior to the imaging study. Study participants were prospectively enrolled in this study from March 2015 to May 2017. Our study population comprised patients 18 years of age or older with suspicious thyroid nodule(s), who were referred for FNAB. The study participants were previously assigned Thyroid Imaging Reporting and Data System (TI-RADS) assessments by their radiologists based on clinical ultrasound features. TI-RADS scores above 3 were referred for FNAB as part of their clinical care. In addition, nodules with TI-RADS score 3 that were larger than 2.5 cm and those with TI-RADS score 2 but associated with cervical lymphadenopathy and/or a history of thyroid cancer were referred for FNAB. The HDMI study was conducted prior to FNAB. All nodules with positive or indeterminate results of FNABs underwent surgery. The histological results of FNAB and/or surgical pathology reports were utilized as a reference gold standard for malignancy status. The final diagnosis of all malignant cases, regardless of the size, was based on the surgical pathology results—not FNAB. The diagnosis of all benign cases, regardless of the size, was based on the standard clinical findings and FNAB, as there is no need for the benign cases to undergo surgery. ## 2.2. Clinical Ultrasound Features All enrolled patients had clinical thyroid ultrasound imaging examinations. TI-RADS points were given based on various ultrasound features in a thyroid nodule. Among the ultrasound features of thyroid; echogenicity (hyperechoic, hypoechoic, or isoechoic); composition (solid or mixed or spongiform); shape (taller than wide); margin (smooth, ill-defined or irregular); calcifications (macrocalcifications, peripheral or rim microcalcification); and vascularity were considered in awarding TI-RADS points to each nodule. The total points in each nodule determined the TI-RADS score. TI-RADS scores were assigned before the HDMI study. ## 2.3. High-Definition Microvasculature Imaging For all participants, HDMI was conducted before FNAB. Ultrasound examination was conducted by one of two sonographers with more than 30 and 18 years of experience. Thyroid nodules were identified using an ultrasound platform equipped with plane wave imaging, Alpinion E-Cube 12R ultrasound scanner (Alpinion Medical System Co., Seoul, Republic of Korea), equipped with an L3-12H linear probe operating at 8.5 MHz. Patients were scanned in the supine position with their neck inclined back and turned to the left or right, depending on the position of the nodule. To reduce the compression effect on altering tissue microvessels, our sonographers were instructed to reduce the preload during the ultrasound examination. To diminish motion artifacts, patients were requested to remain at a standstill and pause their breathing for approximately 3 s during data acquisition. After finding the nodule in B-mode ultrasound, a sequence of high frame rate data was processed, detailed in [24,25]. The acquisitions were acquired in both longitudinal and transverse cross-sections of the thyroid gland. Since out-of-plane motion occurs less in longitudinal cross-sections due to the distal location of the trachea and carotid artery with respect to the thyroid gland, the longitudinal view was selected as a more reliable cross-section for microvascular blood flow images. To ensure repeatability, two acquisitions in each orientation were acquired. After HDMI acquisitions, image processing, and denoising were performed [24,25]. The nodules were manually segmented using B-mode images obtained from the IQ data reconstruction and images were prepared for quantification of morphological parameters of tumor microvessels [26]. All the methods for HDMI image formation, vessel extractions, denoising, morphological filtering, steps for vessel segmentation, and quantification have been detailed in [24,25,26,27,29,30,31,32]. ## 2.4. Microvessel Morphological Parameters Subsequently, after image reconstruction, the morphological parameters of thyroid nodules were extracted from the HDMI images and were used in this study as imaging biomarkers. Vessel density is one of the best-known vessel parameters that defined as the ratio of the geometric area of vessel segments to the geometric area of the associated region of interest of nodule [26,33]. In addition to vessel density, the number of vessel segments (NV), and the number of branch points (NB) that is defined as a common point connected to three or more vessel segments are also calculated [26,34]. Another important parameter, the diameter of the vessel, defined as two times the minimum distance between the vessel centerline and the vessel border has also been calculated [26]. Moreover, we have found that Murray’s deviation (MD), defined as the deviations from Murray’s law, is an important biomarker that presents a diameter mismatch, the definition and calculation of which have been detailed in [27]. Another important vessel morphological parameter, vessel tortuosity determined by distance metric (DM), which is defined as the ratio between the actual path length of a twisting vessel and the linear distance between the two endpoints of that vessel, was measured in this study [26]. Fractal dimension (mvFD), a unitless, geometrical feature that presents the structural complexity of a vascular network has been included in the analysis to provide additional diagnostic and prognostic information. The definition and calculation of mvFD are detailed in [27]. Furthermore, the vessel density ratio (VDR), defined as the ratio of vessel density of the tumor center to the periphery [27], and spatial vascularity pattern (SVP), calculated by VDR, can present the tumor vascular distribution pattern as being either intratumoral or being peritumoral [27,35]. The proposed morphological operations and quantification steps have been well detailed in our previous papers [26,27]. ## 2.5. Fine Needle Aspiration Biopsy All study patients underwent FNAB within a day after the HDMI test. Under ultrasound guidance, our board-certified endocrinologists or radiologists performed FNAB using a standard sterile technique and a 25-gauge needle to obtain six fine needle aspirates for each nodule. Immediately after FNAB, slides were prepared and sent for cytology. Pathological diagnosis was made by pathologists with more than 15 to 20 years of experience. The histopathological results of FNAB and surgical excision for all thyroid nodules were included for data analysis as a reference gold standard. ## 2.6. Statistical Analysis Methods All image processing and data analyses were performed by the members of our investigative team who were blinded to the pathology results of thyroid FNAB and or surgical pathology. Quantitative variables were summarized as mean ± standard deviation (SD), while nominal variables were summarized as counts and percentages. Differences in distributions of HDMI biomarkers by thyroid nodule malignancy status were assessed using the nonparametric Wilcoxon rank-sum test. A two-sided p-value < 0.05 was considered to be statistically significant. To develop a classification model and to investigate the specificity, sensitivity, and area under the curve (AUC) of the receiver operating characteristics (ROC) curve, a multivariable analysis and classifier training were performed using Classification Learner, MATLAB toolbox. Biomarkers with p-values less than 0.05 were used as features to train and evaluate classifiers. Subsequent to generating the ROC curves, optimal cut-off thresholds were determined as points with the minimum distance from the maximum sensitivity and specificity point (top left corner) on the curve. Data were randomly partitioned into independent subsets to train the algorithm ($70\%$) and the remaining data ($30\%$) for testing. A support vector machine (SVM) classifier with a Gaussian kernel trained in the space of HDMI biomarkers was found to be the best-performing method for our analysis. Two feature sets were considered for model building: the first was restricted to HDMI biomarkers outlined above, while the second considered the addition of the clinical factors of age, nodule size, and TI-RADS. The corresponding models were, respectively, designated as the HDMI model and HDMI-C model. To prevent overfitting during the tuning process, a five-fold cross-validation procedure was used. All of the processes were implemented in MATLAB R2022a (The Mathworks Inc., Natick, MA, USA). ## 3.1. Characteristics of the Study Population and Thyroid Nodules A total of 92 patients were successfully enrolled in this study and examined by HDMI. The histopathological results of FNAB confirmed $\frac{55}{92}$ ($60\%$) of thyroid nodules as benign, and $\frac{2}{92}$ ($2\%$) of the nodules were confirmed to be benign by surgical pathology, constituting a total of $\frac{57}{92}$ ($62\%$) benign nodules. All malignant nodules, $\frac{35}{92}$ ($38\%$), were confirmed by surgical pathology. From the entire cohort, $\frac{74}{92}$ ($80\%$) patients were female and $\frac{18}{92}$ ($20\%$) were male, with an age range of 25 to 86 years (mean age ± standard deviation: 53.5 ± 14.8 years). The nodule size in the largest dimension ranged from 6 to 61 mm with a mean ± standard deviation of 21.06 ± 12.70 mm. The participant demographic information, lesion characteristics, and the distribution of malignant lesion type by the pathology are shown in Table 1. The most common malignant histologic type was papillary thyroid carcinoma, corresponding to $\frac{31}{35}$ ($88\%$) nodules. ## 3.2. Visualization and Quantification of Microvessel Biomarkers of Thyroid Nodules Representative images of benign and malignant thyroid nodules in two groups of patients, based on nodule size and spatial vascularity pattern along with quantified biomarkers are displayed in Figure 1 and Figure 2. These include conventional B-mode ultrasound and HMDI images of larger nodules, with a diameter in the largest dimension, of 24 mm in benign nodules (Figure 1A,B) and 23 mm in malignant nodules (Figure 1E,F), respectively. Figure 2 shows, B-mode ultrasound, and HDMI images of smaller nodules with a diameter in the largest dimension, of 12 mm in benign nodules, Figure 2A,B and 11 mm in malignant nodules, Figure 2E,F. The SVP diagram displays the vascular distribution pattern as perinodular in Figure 1D and intra-nodular in nodules displayed in Figure 1H and Figure 2D,H. The visual inspection shows microvessels with high density along with irregularity in malignant nodules while microvessels in benign are much less dense and more regular. HDMI biomarkers, shown on the right side of Figure 1 and Figure 2 (top) for benign nodules and (bottom) for malignant, demonstrate differentiating values between benign and malignant. ## 3.3. Analysis of HDMI Biomarkers for Differentiation of Thyroid Nodules Malignant thyroid nodules had significantly higher values of NV, NB, VD, mvFD, and MDmax when compared to benign nodules (all $p \leq 0.01$, Table 2). The value of BAmean was also a parameter with significantly different distributions between the two groups, showing a decrease in bifurcation angle in malignant nodules when compared to benign ones ($$p \leq 0.0029$$). The distributions of HDMI biomarkers with significant differences between malignant and benign nodules are shown in Figure 3 (box plots, a–f). ## 3.4. Differentiating Malignant Nodules from Benign with HDMI Biomarkers, and Combined with Clinical Factors The corresponding ROC curves for the performance of the HDMI and HDMI-C models on the test set are shown in Figure 4. Six significant HDMI biomarkers were included in the HDMI model. For the HDMI model, the AUC was 0.9005 ($95\%$ CI: [0.8279,0.9732]), with a sensitivity of $77.78\%$, a specificity of $94.74\%$, and an accuracy of $89.29\%$. The AUC slightly increased when clinical factors of age, nodule size, and TI-RADS were added to the HDMI model. The corresponding AUC estimate was 0.9044 ($95\%$ CI: [0.8331,0.9757]), with a sensitivity of 87.50 %, a specificity of $82.35\%$, and an accuracy of $84.00\%$. ## 4. Discussion The present study evaluated a set of microvessel morphological parameters, extracted from the newly developed US-based quantitative HDMI, as quantitative tumor biomarkers for the differentiation of malignant and benign thyroid nodules. No contrast agent was applied to extract submillimeter microvessels. We identified six HDMI biomarkers, namely vessel density, number of vessel segments, number of branch points, microvessel fractal dimension, bifurcation angle (mean), and Murray’s deviation (max) that demonstrated significantly different distributions between malignant and benign nodules. It is known that the color Doppler flow pattern has very limited value in differentiating benign from malignant thyroid nodules [36]. Efforts have been made to investigate the value of microvessel imaging without the help of contrast agents for thyroid nodule differentiation, either using AngioPLUS Microvascular Imaging [37] or superb microvessel imaging (SMI) [22,38,39]; however, the evaluation is mostly based on visual inspection and the quantification is limited to vessel index and pixel counting. Furthermore, studies on super-resolution imaging combined SMI and CEUS to achieve better results in differentiating benign from malignant nodules [40]; however, the limited quantification along with the inconvenience of injecting contrast agents exists. In support of our observation, Caresio et al. confirmed the correlation between the morphology and distribution of blood vessels and the malignancy, assessing reconstructed vascular architecture from 3D PDUS and CEUS images of thyroid nodules [35]. A study using SMI reported that the addition of TI-RADS scores to other imaging parameters improved the diagnostic accuracy in distinguishing benign from malignant thyroid nodules [21]. In contrast, our study shows that adding TI-RADS scores to the HDMI parameters increased the sensitivity but reduced the specificity while keeping the overall diagnostic performance in terms of AUC essentially the same. Such changes in the sensitivity and specificity of combined HDMI and TI-RADS scores may be attributed to the high sensitivity and low specificity of the sole TI-RADS in thyroid cancer detection. The present study demonstrates that benign nodules had lower numbers of vessel segments and branch points compared to malignant thyroid nodules. It is known that the rapid growth of the neoplasm is associated with a greater level of vessel sprouting, resulting in increased NV and NB in malignant tumors [41]. Our findings are also supported by our previous studies in breast cancer detection [25,28] and studies by others [35,42]. In our study vessel density showed statistically significant differences between benign and malignant nodules. Similar findings have been reported by other studies on thyroid nodule differentiation [35] and in renal cell carcinoma [43]. Additionally, we observed higher values of Murray’s deviation in malignant thyroid nodules, with MDmax showing a statistically significant difference between the two groups. The diagnostic value of MD has been demonstrated for different diseases [27,44], indicating that the vascular network of diseased and malignant tissue may show a deviation from Murray’s law [45,46,47]. Moreover, our study found lower levels in the mean of the bifurcation angle in malignant thyroid nodules compared with the benign ones. Similar findings have been reported in previous studies on invasive colon carcinomas [48] and breast cancer detection [27,28], showing smaller angles in the vessel network of malignant tissues and with fewer branches in benign, the bifurcation angle among them is wider. Consistent with previous research on oral cancer carcinoma [49], renal cell carcinoma [50], glioblastoma [51], and breast lesions [25], we found that malignant nodules have significantly higher values of mvFD than benign and the difference in distribution among the two groups was statistically significant suggesting that the hypervascularity in malignant nodules is associated with complexity, irregularly branched, and distorted microvessel networks. Similar results were obtained from other studies [25]. To avoid selection bias, the patient selection in the current study was not based on age or gender; rather, adult patients with suspicious thyroid nodules who were referred for FNAB were recruited regardless of age and sex. Malignant cases in the present study comprised $77\%$ women and $23\%$ men. This distribution is in agreement with national trends reported in 2022 cancer statistics [52]. The current study also shows a mean age of 42 years old in the group of patients with malignant thyroid nodules, where this number is close to the national average age of 47 years reported for patients with thyroid cancer [53]. The focus of the present study is to validate the performance of the quantitative 2D HDMI for the differentiation of benign and malignant thyroid nodules. In a previous study, using a motorized 3D technique and linear array transducer, the performance of our new three-dimensional (3D) HDMI for differentiation of breast masses was shown in [54]. The future direction may include developing a new 3D system using a 2D matrix transducer suitable for thyroid imaging and performing quantitative 3D HDMI and morphometric analysis of nodule microvessels in three dimensions to improve the diagnostic performance of HDMI. In the future, conducting a large-scale multi-center clinical trial will allow a more thorough investigation of the role of the new microvessel imaging in differentiating thyroid nodules. It should be noted that there is a potential for data degradation due to the motion caused by the thyroid’s proximity to pulsating carotid artery, thus affecting the visualization of microvessels [55]. In future studies, one may employ motion correction algorithms [32] to improve the performance of HDMI in distinguishing benign from cancerous thyroid nodules. ## 5. Conclusions The low specificity of traditional ultrasound leads to a great number of unnecessary (i.e., benign) biopsies that causes a significant financial and physical burden to the patients. To overcome the present challenging dilemma, we developed and investigated a new contrast-free ultrasound-based quantitative microvasculature imaging technique for better characterization of thyroid nodules. By assessing the morphological features of tumor microvasculature derived from HDMI images, as new biomarkers, we were able to differentiate between malignant and benign thyroid nodules with high specificity. The results of this study indicate the potential of this technique for the optimum characterization of thyroid nodules. In conclusion, quantitative HDMI could offer a way to accurately characterize thyroid nodules and reduce the number of unnecessary thyroid nodules benign biopsies. ## 6. Patents US Patent 11,213,278 and EP 3 634 238 B1, JP 7139357 B2. ## References 1. Seib C.D., Sosa J.A.. Evolving understanding of the epidemiology of thyroid cancer. Endocrinol. Metab. Clin. (2019) 48 23-35. DOI: 10.1016/j.ecl.2018.10.002 2. Prete A., Borges de Souza P., Censi S., Muzza M., Nucci N., Sponziello M.. Update on fundamental mechanisms of thyroid cancer. Front. Endocrinol. (2020) 11 102. DOI: 10.3389/fendo.2020.00102 3. Haugen B.R., Alexander E.K., Bible K.C., Doherty G.M., Mandel S.J., Nikiforov Y.E., Pacini F., Randolph G.W., Sawka A.M., Schlumberger M.. 2015 American Thyroid Association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer: The American Thyroid Association guidelines task force on thyroid nodules and differentiated thyroid cancer. Thyroid (2016) 26 1-133. DOI: 10.1089/thy.2015.0020 4. Wiest P.W., Hartshorne M.F., Inskip P.D., Crooks L.A., Vela B.S., Telepak R.J., Williamson M.R., Blumhardt R., Bauman J.M., Tekkel M.. Thyroid palpation versus high-resolution thyroid ultrasonography in the detection of nodules. J. Ultrasound Med. (1998) 17 487-496. DOI: 10.7863/jum.1998.17.8.487 5. Bessey L.J., Lai N.B.K., Coorough N.E., Chen H., Sippel R.S.. The incidence of thyroid cancer by f ine needle aspiration varies by age and gender. J. Surg. Res. (2013) 184 761-765. DOI: 10.1016/j.jss.2013.03.086 6. Valderrabano P., McIver B.. Evaluation and management of indeterminate thyroid nodules: The revolution of risk stratification beyond cytological diagnosis. Cancer Control (2017) 24 1073274817729231. DOI: 10.1177/1073274817729231 7. Zhao C.-K., Xu H.-X.. Ultrasound elastography of the thyroid: Principles and current status. Ultrasonography (2019) 38 106. DOI: 10.14366/usg.18037 8. Cantisani V., Maceroni P., D’Andrea V., Patrizi G., Di Segni M., De Vito C., Grazhdani H., Isidori A.M., Giannetta E., Redler A.. Strain ratio ultrasound elastography increases the accuracy of colour-Doppler ultrasound in the evaluation of Thy-3 nodules. A bi-centre university experience. Eur. Radiol. (2016) 26 1441-1449. DOI: 10.1007/s00330-015-3956-0 9. Gregory A., Bayat M., Kumar V., Denis M., Kim B.H., Webb J., Meixner D.D., Ryder M., Knudsen J.M., Chen S.. Differentiation of benign and malignant thyroid nodules by using comb-push ultrasound shear elastography: A preliminary two-plane view study. Acad. Radiol. (2018) 25 1388-1397. DOI: 10.1016/j.acra.2018.02.025 10. Park A.Y., Son E.J., Han K., Youk J.H., Kim J.-A., Park C.S.. Shear wave elastography of thyroid nodules for the prediction of malignancy in a large scale study. Eur. J. Radiol. (2015) 84 407-412. DOI: 10.1016/j.ejrad.2014.11.019 11. Kumar V., Webb J., Gregory A., Meixner D.D., Knudsen J.M., Callstrom M., Fatemi M., Alizad A.. Automated segmentation of thyroid nodule, gland, and cystic components from ultrasound images using deep learning. IEEE Access (2020) 8 63482-63496. DOI: 10.1109/ACCESS.2020.2982390 12. Kohlenberg J., Gu J., Parvinian A., Webb J., El Kawkgi O., Larson N.B., Ryder M., Fatemi M., Alizad A.. Added value of mass characteristic frequency to 2-D shear wave elastography for differentiation of benign and malignant thyroid nodules. Ultrasound Med. Biol. (2022) 48 1663-1671. DOI: 10.1016/j.ultrasmedbio.2022.04.218 13. Chambara N., Lo X., Chow T.C.M., Lai C.M.S., Liu S.Y.W., Ying M.. Combined Shear Wave Elastography and EU TIRADS in Differentiating Malignant and Benign Thyroid Nodules. Cancers (2022) 14. DOI: 10.3390/cancers14225521 14. Brandenstein M., Wiesinger I., Künzel J., Hornung M., Stroszczynski C., Jung E.-M.. Multiparametric Sonographic Imaging of Thyroid Lesions: Chances of B-Mode, Elastography and CEUS in Relation to Preoperative Histopathology. Cancers (2022) 14. DOI: 10.3390/cancers14194745 15. Reginelli A., Urraro F., di Grezia G., Napolitano G., Maggialetti N., Cappabianca S., Brunese L., Squillaci E.. Conventional ultrasound integrated with elastosonography and B-flow imaging in the diagnosis of thyroid nodular lesions. Int. J. Surg. (2014) 12 S117-S122. DOI: 10.1016/j.ijsu.2014.05.033 16. Ramsden J.. Angiogenesis in the thyroid gland. J. Endocrinol. (2000) 166 475-480. DOI: 10.1677/joe.0.1660475 17. Rajabi S., Dehghan M.H., Dastmalchi R., Mashayekhi F.J., Salami S., Hedayati M.. The roles and role-players in thyroid cancer angiogenesis. Endocr. J. (2019) 66 277-293. DOI: 10.1507/endocrj.EJ18-0537 18. Ebeed A.E., Romeih M.A.E.-h., Refat M.M., Salah N.M.. Role of ultrasound, color doppler, elastography and micropure imaging in differentiation between benign and malignant thyroid nodules. Egypt. J. Radiol. Nucl. Med. (2017) 48 603-610. DOI: 10.1016/j.ejrnm.2017.03.012 19. Chammas M.C., Gerhard R., Oliveira I.R.S.D., Widman A., Barros N.D., Durazzo M., Ferraz A., Cerri G.G.. Thyroid nodules: Evaluation with power Doppler and duplex Doppler ultrasound. Otolaryngol. Head Neck Surg. (2005) 132 874-882. DOI: 10.1016/j.otohns.2005.02.003 20. Newsome I.G., Dayton P.A.. Visualization of microvascular angiogenesis using dual-frequency contrast-enhanced acoustic angiography: A review. Ultrasound Med. Biol. (2020) 46 2625-2635. DOI: 10.1016/j.ultrasmedbio.2020.06.009 21. Zhu C., Zhong L., Lin M., Tian C., Wang C.. The value of TI-RADS combined with superb micro-vascular imagine in distinguishing benign and malignant thyroid nodules: A meta-analysis. PLoS ONE (2022) 17. DOI: 10.1371/journal.pone.0261521 22. Hong M.J., Ahn H.S., Ha S.M., Park H.J., Oh J.. Quantitative analysis of vascularity for thyroid nodules on ultrasound using superb microvascular imaging: Can nodular vascularity differentiate between malignant and benign thyroid nodules?. Medicine (2022) 101 e28725. DOI: 10.1097/MD.0000000000028725 23. Lee S., Lee J.Y., Yoon R.G., Kim J.-h., Hong H.S.. The value of microvascular imaging for triaging indeterminate cervical lymph nodes in patients with papillary thyroid carcinoma. Cancers (2020) 12. DOI: 10.3390/cancers12102839 24. Bayat M., Fatemi M., Alizad A.. Background removal and vessel filtering of noncontrast ultrasound images of microvasculature. IEEE Trans. Biomed. Eng. (2018) 66 831-842. DOI: 10.1109/TBME.2018.2858205 25. Ternifi R., Wang Y., Gu J., Polley E.C., Carter J.M., Pruthi S., Boughey J.C., Fazzio R.T., Fatemi M., Alizad A.. Ultrasound high-definition microvasculature imaging with novel quantitative biomarkers improves breast cancer detection accuracy. Eur. Radiol. (2022) 32 7448-7462. DOI: 10.1007/s00330-022-08815-2 26. Ghavami S., Bayat M., Fatemi M., Alizad A.. Quantification of morphological features in non-contrast-enhanced ultrasound microvasculature imaging. IEEE Access (2020) 8 18925-18937. DOI: 10.1109/ACCESS.2020.2968292 27. Ternifi R., Wang Y., Polley E.C., Fazzio R.T., Fatemi M., Alizad A.. Quantitative biomarkers for cancer detection using contrast-free ultrasound high-definition microvessel imaging: Fractal dimension, murray’s deviation, bifurcation angle & spatial vascularity pattern. IEEE Trans. Med. Imaging (2021) 40 3891-3900. PMID: 34329160 28. Gu J., Ternifi R., Larson N.B., Carter J.M., Boughey J.C., Stan D.L., Fazzio R.T., Fatemi M., Alizad A.. Hybrid high-definition microvessel imaging/shear wave elastography improves breast lesion characterization. Breast Cancer Res. (2022) 24 16. DOI: 10.1186/s13058-022-01511-5 29. Adabi S., Ghavami S., Fatemi M., Alizad A.. Non-local based denoising framework for in vivo contrast-free ultrasound microvessel imaging. Sensors (2019) 19. DOI: 10.3390/s19020245 30. Nayak R., Kumar V., Webb J., Gregory A., Fatemi M., Alizad A.. Non-contrast agent based small vessel imaging of human thyroid using motion corrected power Doppler imaging. Sci. Rep. (2018) 8 15318. DOI: 10.1038/s41598-018-33602-9 31. Nayak R., Kumar V., Webb J., Fatemi M., Alizad A.. Non-invasive small vessel imaging of human thyroid using motion-corrected spatiotemporal clutter filtering. Ultrasound Med. Biol. (2019) 45 1010-1018. DOI: 10.1016/j.ultrasmedbio.2018.10.028 32. Nayak R., MacNeill J., Flores C., Webb J., Fatemi M., Alizad A.. Quantitative assessment of ensemble coherency in contrast-free ultrasound microvasculature imaging. Med. Phys. (2021) 48 3540-3558. DOI: 10.1002/mp.14918 33. You Q.S., Chan J.C., Ng A.L., Choy B.K., Shih K.C., Cheung J.J., Wong J.K., Shum J.W., Ni M.Y., Lai J.S.. Macular vessel density measured with optical coherence tomography angiography and its associations in a large population-based study. Investig. Ophthalmol. Vis. Sci. (2019) 60 4830-4837. DOI: 10.1167/iovs.19-28137 34. Edgar L.T., Hoying J.B., Utzinger U., Underwood C.J., Krishnan L., Baggett B.K., Maas S.A., Guilkey J.E., Weiss J.A.. Mechanical interaction of angiogenic microvessels with the extracellular matrix. J. Biomech. Eng. (2014) 136 021001. DOI: 10.1115/1.4026471 35. Caresio C., Caballo M., Deandrea M., Garberoglio R., Mormile A., Rossetto R., Limone P., Molinari F.. Quantitative analysis of thyroid tumors vascularity: A comparison between 3-D contrast-enhanced ultrasound and 3-D Power Doppler on benign and malignant thyroid nodules. Med. Phys. (2018) 45 3173-3184. DOI: 10.1002/mp.12971 36. Sultan L.R., Xiong H., Zafar H.M., Schultz S.M., Langer J.E., Sehgal C.M.. Vascularity assessment of thyroid nodules by quantitative color Doppler ultrasound. Ultrasound Med. Biol. (2015) 41 1287-1293. DOI: 10.1016/j.ultrasmedbio.2015.01.001 37. Chambara N., Liu S.Y.W., Lo X., Ying M.. Diagnostic Value of AngioPLUS Microvascular Imaging in Thyroid Nodule Diagnosis Using Quantitative and Qualitative Vascularity Grading. Biomedicines (2022) 10. DOI: 10.3390/biomedicines10071554 38. Lu R., Meng Y., Zhang Y., Zhao W., Wang X., Jin M., Guo R.. Superb microvascular imaging (SMI) compared with conventional ultrasound for evaluating thyroid nodules. BMC Med. Imaging (2017) 17. DOI: 10.1186/s12880-017-0241-5 39. Chen L., Zhan J., Diao X.-H., Liu Y.-C., Shi Y.-X., Chen Y., Zhan W.-W.. Additional value of superb microvascular imaging for thyroid nodule classification with the thyroid imaging reporting and data system. Ultrasound Med. Biol. (2019) 45 2040-2048. DOI: 10.1016/j.ultrasmedbio.2019.05.001 40. Zhang G., Yu J., Lei Y.-M., Hu J.-R., Hu H.-M., Harput S., Guo Z., Cui X.-W., Ye H.-R.. Ultrasound super-resolution imaging for the differential diagnosis of thyroid nodules: A pilot study. Front. Oncol. (2022) 12 978164. DOI: 10.3389/fonc.2022.978164 41. Chappell J.C., Wiley D.M., Bautch V.L.. How blood vessel networks are made and measured. Cells Tissues Organs (2012) 195 94-107. DOI: 10.1159/000331398 42. Lindsey B.D., Rojas J.D., Martin K.H., Shelton S.E., Dayton P.A.. Acoustic characterization of contrast-to-tissue ratio and axial resolution for dual-frequency contrast-specific acoustic angiography imaging. IEEE Trans. Ultrason. Ferroelectr. Freq. Control (2014) 61 1668-1687. DOI: 10.1109/TUFFC.2014.006466 43. Rojas J.D., Papadopoulou V., Czernuszewicz T.J., Rajamahendiran R.M., Chytil A., Chiang Y.-C., Chong D.C., Bautch V.L., Rathmell W.K., Aylward S.. Ultrasound measurement of vascular density to evaluate response to anti-angiogenic therapy in renal cell carcinoma. IEEE Trans. Biomed. Eng. (2018) 66 873-880. DOI: 10.1109/TBME.2018.2860932 44. Sun M., Lv W., Zhao X., Qin L., Zhao Y., Xin X., Jian J., Chen X., Hu C.. Vascular branching geometry relating to portal hypertension: A study of liver microvasculature in cirrhotic rats by X-ray phase-contrast computed tomography. Quant. Imaging Med. Surg. (2020) 10 116. DOI: 10.21037/qims.2019.11.04 45. Schoenenberger A.W., Urbanek N., Toggweiler S., Seelos R., Jamshidi P., Resink T.J., Erne P.. Deviation from Murray's law is associated with a higher degree of calcification in coronary bifurcations. Atherosclerosis (2012) 221 124-130. DOI: 10.1016/j.atherosclerosis.2011.12.040 46. Murray C.D.. The physiological principle of minimum work applied to the angle of branching of arteries. J. Gen. Physiol. (1926) 9 835. DOI: 10.1085/jgp.9.6.835 47. McAllister A., Abramoff M., Xu X.. Deviation from the optimal branching relationship of retinal vessels in diabetes mellitus. Investig. Ophthalmol. Vis. Sci. (2013) 54 2421 48. Konerding M., Fait E., Gaumann A.. 3D microvascular architecture of pre-cancerous lesions and invasive carcinomas of the colon. Br. J. Cancer (2001) 84 1354-1362. DOI: 10.1054/bjoc.2001.1809 49. Goutzanis L.P., Papadogeorgakis N., Pavlopoulos P.M., Petsinis V., Plochoras I., Eleftheriadis E., Pantelidaki A., Patsouris E., Alexandridis C.. Vascular fractal dimension and total vascular area in the study of oral cancer. Head Neck (2009) 31 298-307. DOI: 10.1002/hed.20959 50. Sabo E., Boltenko A., Sova Y., Stein A., Kleinhaus S., Resnick M.B.. Microscopic analysis and significance of vascular architectural complexity in renal cell carcinoma. Clin. Cancer Res. (2001) 7 533-537. PMID: 11297244 51. Chen C., He Z.-C., Shi Y., Zhou W., Zhang X., Xiao H.-L., Wu H.-B., Yao X.-H., Luo W.-C., Cui Y.-H.. Microvascular fractal dimension predicts prognosis and response to chemotherapy in glioblastoma: An automatic image analysis study. Lab. Investig. (2018) 98 924-934. DOI: 10.1038/s41374-018-0055-2 52. Grimm D.. Recent Advances in Thyroid Cancer Research. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23094631 53. Wang J., Jiang J., Zhang D., Zhang Y.-Z., Guo L., Jiang Y., Du S., Zhou Q.. An integrated AI model to improve diagnostic accuracy of ultrasound and output known risk features in suspicious thyroid nodules. Eur. Radiol. (2022) 32 2120-2129. DOI: 10.1007/s00330-021-08298-7 54. Gu J., Ternifi R., Sabeti S., Larson N.B., Carter J.M., Fazzio R.T., Fatemi M., Alizad A.. Volumetric imaging and morphometric analysis of breast tumor angiogenesis using a new contrast-free ultrasound technique: A feasibility study. Breast Cancer Res. (2022) 24 85. DOI: 10.1186/s13058-022-01583-3 55. Nayak R., Nawar N., Webb J., Fatemi M., Alizad A.. Impact of imaging cross-section on visualization of thyroid microvessels using ultrasound: Pilot study. Sci. Rep. (2020) 10 415. DOI: 10.1038/s41598-019-57330-w
--- title: Caloric Restriction Can Ameliorate Postoperative Cognitive Dysfunction by Upregulating the Expression of Sirt1, MeCP2 and BDNF in the Hippocampal CA1 Region of Aged C57BL/6 Mice authors: - Lan Wei - Qiang Tao - Minmin Yao - Zhimeng Zhao - Shengjin Ge journal: Brain Sciences year: 2023 pmcid: PMC10046819 doi: 10.3390/brainsci13030462 license: CC BY 4.0 --- # Caloric Restriction Can Ameliorate Postoperative Cognitive Dysfunction by Upregulating the Expression of Sirt1, MeCP2 and BDNF in the Hippocampal CA1 Region of Aged C57BL/6 Mice ## Abstract This study aimed to investigate the impact of caloric restriction (CR) on cognitive function in aged C57BL/6 mice after surgery, as well as the underlying mechanisms. Forty 14-month-old male C57BL/6 mice were randomly assigned to the ad libitum (AL, $$n = 20$$) group and the CR ($$n = 20$$) group. After feeding for 12 weeks, they were subdivided into four groups: AL control (ALC, $$n = 10$$), AL with surgery (ALS, $$n = 10$$), CR control (CRC, $$n = 10$$), and CR with surgery (CRS, $$n = 10$$). The Morris Water Maze (MWM) test was used to assess learning and memory capacity. By using western blot and immunofluorescence, the expression of Sirt1, MeCP2, and BDNF in the hippocampus and hippocampal CA1 region was quantified. According to the behavioral test, the CRC and CRS groups had significantly better learning and memory abilities than the ALC and ALS groups, respectively. Sirt1, MeCP2, and BDNF expression in the hippocampus and CA1 region in the hippocampus of the ALC and CRC groups of mice were correlated with cognitive improvement. In conclusion, CR could enhance the postoperative cognitive function in aged mice, most likely by increasing the expression of Sirt1, MeCP2, and BDNF in the CA1 region of the hippocampus. ## 1. Introduction Postoperative cognitive dysfunction (POCD) is a common neurocognitive complication after surgery in the elderly, and is characterized by functional impairments, such as memory, executive function, direction, emotion, and visual–spatial structural ability [1,2]. According to recent research, elderly people over 65 years were highly likely to have POCD; moreover, the incidence can reach up to $47.6\%$ of patients at one week and $34.2\%$ of patients at three months after surgery [3]. There are numerous risk factors for POCD, including aging, educational level, inappropriate diet, obesity, anesthesia, surgery, etc. [ 4,5,6,7,8]. The main risk factor for POCD among them is aging, which is a sign of the gradual decline of physiological functions [4]. Unfortunately, despite extensive research on POCD for more than a century [1], due to the complex pathophysiology and mechanisms, there is still no effective preventive or therapeutic approach. Early in 1935, McCay et al. discovered that dietary changes could prolong the survival time of mammals and decrease the incidence of aging-related diseases, like Alzheimer’s disease (AD) [9,10]. Research has shown that CR has positive effects on reducing neurodegenerative diseases and brain aging, in addition to extending lifespan [11,12]. CR in older adults has also been verified to improve recognition memory, which is closely related to hippocampal function [13]. As is common knowledge, the hippocampus plays a critical role in learning and memory [14,15]. Age-related cognitive deficits may be improved by dietary factors that enhance hippocampal neurogenesis and brain plasticity [16]. The hippocampus CA1 region is important for memory and learning, and cognitive dysfunction is closely linked to the activity of the neurons in this region [17,18,19]. CR has received increasing attention for its role in the cognitive abilities of the elderly. CR can be achieved by reducing calorie intake over a specific time period while maintaining adequate levels of macro- and micronutrients, and the typical levels of CR in mice and rats range from 10 to $50\%$ [20,21,22]. Sirt1, an NAD+-dependent deacetylase that is expressed in neurons of the hippocampus, is essential for a number of brain processes, including normal learning, memory, and synaptic plasticity in mice [23]. Mice’s spatial learning and memory were weakened by hippocampal Sirt1 knockdown, which also caused hippocampal atrophy [24]. By altering Sirt1 expression in the mouse hippocampus, CR can have neuroprotective effects [25]. MeCP2 is a transcriptional regulator that is highly abundant in the brain and can bind to methylated genomic DNA to regulate a range of physiological processes that are associated with adult synaptic plasticity and neuronal development [26]. Hippocampal MeCP2 knockdown has the opposite effects of overexpression, which could improve synaptic plasticity and cognitive function [27]. However, less is known about MeCP2’s function in neurodegenerative diseases, such as AD and POCD. BDNF is highly expressed in the brain, and it stands out for its broad roles in brain homeostasis, health, and disease via its complex downstream signals [28,29]. Numerous studies have firmly found that BDNF has a critical effect on hippocampal long-term potentiation (LTP), a process that prolongs synaptic efficacy and is thought to be the foundation of learning and memory [29]. BDNF deficiencies have been linked to neurological diseases like Huntington’s disease, AD, and POCD. But according to research, intermittent fasting (IF) and targeting cognitive performance to increase BDNF are two potential methods for enhancing brain health [30]. As a result, altering BDNF expression might be a workable way to reduce POCD. CR has not only been shown in multiple studies to improve learning and memory in mice, but it can also activate Sir2/Sirt1, involving a number of molecular links, including nicotinamide adenine dinucleotide, nicotinamide, biotin, and related metabolites [25,31]. Previous research has indicated that BDNF in humans may be modulated by diet composition, such as CR [32]. Furthermore, SIRT1-mediated deacetylation of MeCP2 contributes to BDNF expression [33]. There is currently no evidence that CR can regulate the expression of MeCP2 or improve POCD by regulating the expression of Sirt1, MeCP2, and BDNF in the hippocampus. Therefore, this study aims to settle the following three issues: [1] whether CR can improve the POCD in aged C57BL/6 mice after surgery; [2] whether the expression of hippocampal Sirt1, MeCP2, and BDNF proteins changes after CR intervention; and [3] whether the responsible hippocampal region is CA1. ## 2.1. Animals and CR Model Forty 14-month-old C57BL/6 male mice, weighing 36–47 g, were ordered from Shanghai SLAC (Shanghai, China, License number: 2008001622124). All experimental procedures were approved by the Ethics Committee of Zhongshan Hospital, Fudan University (Ethics number: 2017-0001). All mice were housed in separate cages and exposed to a clean environment. After acclimating for 1 week, they were randomly assigned to receive a regular diet ad libitum (AL group, $$n = 20$$) or a CR diet (CR group, $$n = 20$$) using a random number table. Mice in the AL group were fed 20 g of food regularly at 8:00 a.m. every day, and the residual food was weighed after 24 h. According to the literature, the average daily calorie intake was calculated and used as a guide to create a low-calorie diet with a $40\%$ calorie reduction [34]. After 12 weeks of feeding, the mice were subdivided based on surgery performance: ad libitum control (ALC, $$n = 10$$), ad libitum with surgery (ALS, $$n = 10$$), CR control (CRC, $$n = 10$$) and CR with surgery (CRS, $$n = 10$$). Body weight was measured every Monday. Shanghai Pu Lu Teng Company (Q/VGBD1-2014, 25 kg) provided the regular and low-calorie food. The compositions of normal chow and low-calorie chow were displayed in Table 1. Protein, fat, carbohydrate, and other content in normal vs. low-calorie foods were $22.1\%$ vs. $36.8\%$, $5.28\%$ vs. $8.8\%$, $52\%$ vs. $20\%$, and $20.62\%$ vs. $34.4\%$, respectively. In order to prevent adverse events, the blood volume sampled is typically less than $10\%$ of the total weight because the circulating blood volume makes up $6\%$ of the total weight in normal adult mice. Given that a mouse’s total blood volume decreases with age, the blood-sampling period in this case was prolonged while the volume was maintained. To prevent unneeded stress reactions, blood samples were taken every four weeks, and the blood glucose level was measured. ## 2.2. Tail Vein Blood Glucose Test The mice were fixed to expose their tails. After disinfection, the tail was submerged for 3–5 min in warm (45 °C) water to ensure adequate filling and vasodilation of local vessels. Blood was taken from the dorsal caudal vein by making a transverse incision at the distal third of the tail. An amount of 0.1 mL of blood was collected and the incision compressed with gauze to stop the bleeding. Blood samples were centrifuged at 12,000 rpm and 4 °C in a refrigerated centrifuge. After 5 min, the supernatants were collected and preserved at −80 °C. According to the assay kit’s instructions (F006; Nanjing Jiancheng Bioengineering Institute, Nanjing, China), a serum glucose test was performed. The serum glucose concentration was calculated using the following formula: Glucose (mmol/l) = (Experimental OD − Blank OD) × calibrant concentration (5.55 mmol/l)/[(Calibrant OD − Blank OD) × Experimental protein concentration (mgprot/mL)]. ## 2.3. POCD Model The temperature of the mice in the surgery groups was maintained at 37 °C on an electric blanket (yuyan am-92). According to the literature [35], for 4 h, mice were continuously given $2.8\%$ isoflurane and $33\%$ oxygen with the assistance of a small animal anesthesia unit to induce anesthesia. The end-tidal carbon dioxide partial pressure (PETCO2), minimal alveolar concentration (MAC) of the inhaled anesthetics, and oxygen saturation (SpO2) were closely monitored. Following anesthesia and the loss of the righting reflex, open reduction internal fixation for tibial fractures was carried out using the modeling technique described in Terrando N’s study [36]. In short, the left lower limb’s knee region was shaved. The skin was opened 0.5 cm below the knee joint, exposing the tibia. To alleviate early postoperative pain, 0.3 mL of $2\%$ lidocaine was used to induce topical anesthesia. Following that, a 0.38-mm steel needle was inserted and fixed into the bone marrow cavity along the long axis of the tibia. The mid-tibia was immediately clamped with surgical forceps, and the incision was sutured. Local application of anesthetic mixed with $2.5\%$ Lidocaine and $2.5\%$ Prilocaine, smeared at 8-h intervals within 48 h after surgery, provided analgesia. To avoid the spatial crowding effect interfering with the subsequent behavioral tests, mice in the control groups were anesthetized by breathing $33\%$ oxygen for 4 h in the housing cage. Furthermore, PETCO2 and SpO2 were not monitored in these mice to avoid additional stress. ## 2.4. Morris Water Maze Test Daily behaviors and activities were fully recovered in mice 24 h after surgery. The typical MWM experiment was used to test mice’s capacity for spatial learning and memory. There were two tests: the navigation test and the spatial probe test. In the navigation test (6 d), the water phase was maintained at a height of 1.5 cm above the hidden platform. The escape latency period of mice, defined as the time from entering water to finding the hidden platform, indicated their spatial learning ability. The pool was divided into four quadrants, and mice facing the wall of the pool were randomly released into the water in each quadrant. Training was performed four times per day, and the results were averaged. At the end of each training, the mice were dried with paper towels, and the temperature was maintained at 37 °C under a medical heating lamp. The swimming trajectory was automatically recorded using the DigBehv-MM camera system. The maximum time allowed to find the hidden platform was 60 s. Success was determined when the mouse climbed onto the platform and remained there for 3 s. The escape latency period was recorded. The mouse was then left on the platform for 20 s to memorize the surrounding markers, which included triangle, square, circle, and rectangle markers, respectively, at four orientations. Training was subsequently completed on another mouse. The training intervals between mice were consistent. If the mouse failed to locate the hidden platform within 60 s, the mouse was manually guided to find the platform and stay there for 20 s. The corresponding escape latency period was recorded as 60 s. On the 7th day, the hidden platform was removed to perform the spatial probe test. Mice were placed in the pool at a random quadrant, and the time to cross the platform, swimming speed, and time spent swimming in the target quadrant within 60 s were recorded. ## 2.5. Tissue Sampling After the behavioral test, mice were intraperitoneally injected with $1.5\%$ pentobarbital sodium (0.1 mL/20 g, Batch No.57-33-0, Beijing Propbs Biotechnology Co., Ltd., Beijing, China). For western blot analysis, each group randomly selected 5 mice to obtain fresh hippocampus tissue, which was immediately stored at −80 °C. The other mice were perfused with 50 mL of $0.9\%$ sodium chloride through the left apex. When the lungs, intestines, and liver became white, the sodium chloride injection was changed to $4\%$ paraformaldehyde in phosphate buffered saline (PBS; 0.1 M, pH 7.4; 20 mL; fast to slow) until the organs and limbs became hard and the tail became stiff. The brain was then harvested and immersed in a $4\%$ paraformaldehyde solution for fixation for 8 h. Subsequently, the brain was exposed to $15\%$ and $30\%$ sucrose solutions, successively, until completely dehydrated and sinking. The OCT-embedded frozen tissue block was prepared and sectioned using a freezing microtome (Leica CM1950) at 12 μm thickness. The sections were allowed to stay at room temperature for more than 30 min before being restored to a −20 °C refrigerator. Immunofluorescence staining was subsequently performed. ## 2.6. Western Blot After extracting the protein, its concentration was measured by the bicinchoninic acid protein (BCA) method (Beyotime, Shanghai, China). The protein in each sample was denatured by boiling with a loading buffer (1 μL buffer per 4 μL protein sample) at 100 °C for 5 min, then transferred to a Polyvinylidene Fluoride (PVDF) membrane, with $5\%$ skim milk covered to block unspecific bindings, on a shaker at room temperature. After 2 h, the membrane was washed with Tris-buffered saline with Tween 20 (TBST). Primary rabbit anti-mouse antibodies targeting β-actin (1:1000; Cell Signaling Technology, Danvers, MA, USA), Sirt1 (1:1000; Cell Signaling Technology), BDNF (1:1000, Abcam, Cambridge County, UK), and MeCP2 (1:1000; Cell Signaling Technology) were added at 4 °C for overnight incubation. On the following day, after TBST washing, alkaline phosphatase (AP) -coupled goat anti-rabbit secondary antibodies (1:10,000; Haimingrui Biotech, Beijing, China) were added at room temperature, and incubation was completed on a shaker for 2 h. The TBST washing procedure was repeated. ECL substrate (Haimingrui Biotech, Beijing, China) was employed to develop protein bands. Protein expression was qualitatively analyzed with the β-actin as the internal reference to indicate the relative expression. The protein bands were analyzed using the Image-Pro Plus software to calculate optical density (OD). ## 2.7. Immunofluorescence The frozen tissue sections were washed 3 times with 1 × PBS. After that, the tissue was blocked with 1 × PBS, $10\%$ goat serum, and $0.3\%$ Triton at 37 °C for 2 h. Subsequently, the tissue was incubated with primary antibodies (Sirt1, 1:400, rabbit anti-mouse, Cell Signaling Technology; BDNF, 1:750, rabbit anti-mouse, Abcam Plc.; MeCP2, 1:200, rabbit anti-mouse, Cell Signaling Technology) at 4 °C overnight. On the next day, the tissue slides were washed 3 times with 1 × PBS. Then the tissue was incubated with secondary antibodies (1:1000; goat anti-rabbit IgG H&L, Abcam Plc.) in PBS in the dark for 1 h. 1 × PBS washing was performed again. The cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI), followed by 1 × PBS washing. Quenched fluorogens were added, and the slides were sealed. Images were captured under a fluorescence microscope, and the quantification of immuno-positive cells was analyzed using the Image-Pro Plus software version 6.0. ## 2.8. Statistical Analysis All data were processed using the SPSS 23.0 statistical software, and the measurement data were expressed as mean ± standard deviation (x¯±sd). The body weight and blood glucose data were analyzed by Student’s t-tests. The food intake and calorie consumption data were analyzed with the Man–Whitney U test. The data from the behavior tests, the western blot, and immunofluorescence assays were analyzed using one-way ANOVA, and the LSD test was used for those with significant differences. A value of $p \leq 0.05$ was considered to be a statistically significant difference. ## 3.1. Changes in Food Intake, Body Weight, Calorie Consumption and Blood Glucose in Aged Mice The daily food intake of the CR group was significantly less than that of the AL group (Figure 1a). Differences in body weight began to occur between the two groups in the second week ($$p \leq 0.009$$) and became significant in the third week ($p \leq 0.001$) (Figure 1b). Comparatively, the body weight of mice in the CR group was remarkably lower than that of mice in the AL group. Both groups showed relatively stable weight changes at different time points. During the 12-week intervention period, the average CR rate of the CR group was $37.3\%$ compared with the AL group (Figure 1c). Regarding blood glucose, it was slightly lower in the CR group versus the AL group, which was not statistically significant (Figure 1d). ## 3.2. Learning and Memory Ability in Aged Mice In the MWM test, the decline of learning ability was manifested by the extension of escape latency (Figure 2a). When comparing the ALC and ALS groups, the escape latency period on days 3, 5, and 6 was statistically shorter in the ALC group ($p \leq 0.05$). Additionally, the period on days 3 and 6 was significantly shorter in the CRC group than that in the CRS group ($p \leq 0.05$), and that on days 3, 4, and 5 was remarkably shorter in the CRS group than that in the ALS group (all $p \leq 0.05$). When comparing the ALC group with the CRC group, the latency period on days 2, 3, 4, and 6 was relatively prolonged, with statistically significant differences ($p \leq 0.05$). On the last day, memory was injured after surgery, as indicated by a significantly shorter aim-quadrant swimming time and fewer times crossing the platform. The aim-quadrant swimming time of the ALS group was the shortest compared with that in the other groups ($p \leq 0.05$, Figure 2b). The times of crossing the platform in the ALC and CRC groups was longer than that in the ALS and CRS groups, respectively ($p \leq 0.05$, Figure 2c). In addition, the times in the ALC group were less than those in the CRC group ($p \leq 0.05$, Figure 2c). The swimming speed of the ALC and CRC groups was slightly slower than that of the ALS and CRS groups, respectively, but there was no statistically significant difference (Figure 2d), which suggested that postoperative motor was independent from spatial learning and memory. The typical swimming patterns of four groups in the last hidden platform trial can be seen in Figure 2e. ## 3.3. Expression of Related Proteins in the Hippocampus The results of the western blot showed that the level of BDNF in the ALC group was significantly higher than that in the ALS group ($p \leq 0.05$, Figure 3b), but significantly lower than that in the CRC group ($p \leq 0.05$, Figure 3b). Compared with the CRS group, the expression of BDNF was significantly higher in the CRC group ($p \leq 0.05$, Figure 3b). The expression of Sirt1 in the ALS group was significantly lower compared with the ALC group and the CRS group ($p \leq 0.05$, Figure 3c). On the contrary, the Sirt1 expression was much higher in the CRC group compared with the CRS group and the ALC group ($p \leq 0.05$, Figure 3c). The MeCP2 level in the ALC group was significantly higher than that in the ALS group ($p \leq 0.05$, Figure 3d), as well as in the CRC group compared with the CRS group ($p \leq 0.05$, Figure 3d). The level of MeCP2 in the CRC and CRS groups was higher than that in the ALC and ALS groups, respectively ($p \leq 0.05$, Figure 3d). ## 3.4. Expression of Related Proteins in Hippocampal CA1 Region Consistent with the results of the western blot assay, the immunofluorescence staining results indicated that the optical densities of Sirt1-positive cells in the ALC group and the CRS group were significantly decreased compared with the CRC group ($p \leq 0.05$, Figure 4c). The Sirt1 level in the ALS group was significantly lower than that in the CRS and ALC groups ($p \leq 0.05$, Figure 4c). The BDNF (Figure 4d) and MeCP2 (Figure 4e) levels in the ALC group were significantly higher than those in the ALS group ($p \leq 0.05$), just as they were expressed in the CRC group compared with the CRS group ($p \leq 0.05$). The level of MeCP2 (Figure 4e) and BDNF (Figure 4d) in the CRC was higher than that in the ALC group ($p \leq 0.05$). When the ALS group was compared with the CRS group, the expression of MeCP2 remarkably decreased in the ALS group ($p \leq 0.05$, Figure 4e), while the expression of BDNF was not statistically different between the two groups. ## 4. Discussion With improving global health and a steadily increasing elderly population, the frequency of surgery for progressive elderly patients and patients with a high prevalence of complications is increasing. Of note, POCD will be common in this patient population. The complications of POCD have been associated with a longer length of stay, higher mortality, and longer-term cognitive decline [4]. Therefore, this raises the need for a preventive strategy to facilitate early interventions for POCD during the perioperative period. As the main non-genetic mechanism, CR not only extends lifespan but also exerts neuroprotective effects. In this study, we used CR and POCD models to investigate the effects of CR on aged mice. In the results, the body weight of the CR group was significantly lower compared with the AL group. The level of blood glucose was maintained simultaneously within a normal range. These results indicated that proper CR could suppress weight gain in healthy elderly mice while avoiding needless impairments to metabolism. This is consistent with the study of Quintas et al. [ 37], who discovered that CR significantly affected body weight while showing no effect on blood glucose. The model of tibial fracture fixation of POCD was modified from Terrando N [36], and the inhalation concentration of isoflurane was referred to N. Cesarovic [35]. Many kinds of surgical procedures could induce POCD, but we chose the tibial open fracture. Because this type of operation is the most common POCD model, this may be related to the high incidence of POCD in clinical orthopedic surgery [38]. In addition, orthopedic surgery is also a very common type of surgery in elderly patients. In the results of the MWM test, the swimming speed of different groups had no statistical significance, indicating that the mice recovered motor function after surgery. Other studies also demonstrated that an operation has no effect on swimming speed and the locomotor activity of the mice [39,40]. Our POCD model was valid, as evidenced by the ALS group and CRS group mice having shorter aim-quadrant swimming and times of crossing the platform than the ALC group and CRC group mice, respectively. In the behavioral test, the performance of the CRC group was better than that of the ALC group, indicating that CR could improve the learning and memory abilities in normal older mice, in accordance with those of many other studies [41,42]. The difference between the CRC group and the CRS group indicated that CR ameliorated POCD. The hippocampus plays a critical role in learning and memory. Region-specific analyses indicated that CA1 was more susceptible to aging stress, exhibiting a greater number of altered genes relative to CA3 and the dentate gyrus (DG) after CR [43]. Sirt1 has been recognized as a longevity gene and has been verified to be related to aging and disorders associated with it. Quantitative analysis showed that there was a significant reduction of Sirt1 protein levels in the hippocampus with aging [37]. While the long-term high expression of Sirt1 could completely maintain the cognitive ability in mice and delay the loss of neuronal synapse and the incidence of dysfunction [44]. The neuroprotective effect of Sirt1 may be related to the deacetylation of a variety of substrates, including peroxisome proliferator-activated receptor (PPAR)-γ Coactivator-1α (PGC-1α) and nuclear factor-κ B (NF-κ B) [45,46]. In the western blot assay and immunofluorescence test results of our research, the distinctions between the CRC group and the ALC group mice showed that CR enhanced the expression of Sirt1 in the hippocampus CA1 region of normal aged mice; the differences between the CRS group and ALS group mice indicated that CR-ameliorated POCD may be associated with the increased expression of Sirt1 in the hippocampus CA1 region. Strong evidence has shown that CR not only activates Sirt1, but also increases its expression to improve brain health [31,47]. During postnatal development, MeCP2 is necessary to maintain proper brain function. Rett syndrome (RTT), a kind of neurodevelopmental disorder that manifests as a variety of cognitive abnormalities, can result from mutations of the Mecp2 gene. Long-term memory formation in the adult hippocampus depends on MeCP2, which also maintains the chromatin features of mature CA1 neurons and can preserve the genomic responsiveness to hippocampal-dependent learning [48]. Research found that it was also emerging as a regulator of neuronal synaptic plasticity, and it could regulate the expression of BDNF via the interactions with miR-212 and miR-62132 [49,50]. Such regulation was shown to play a vital part in the functioning of the central nervous system and be key to maintaining the homeostasis of synapses and neurons [51]. The hypothesis of our study was that CR increases the expression of Sirt1, and Sirt1-mediated deacetylation of MeCP2 contributes to BDNF expression in hippocampus tissue [33]. Consistent with these findings, in the present study, the expression of MeCP2 in the CRC group was significantly higher than that in the ALC group. Combined with the behavior test and the lowest expression of MeCP2 being in the ALS group, we conclude that POCD may be alleviated by the increased expression of MeCP2 in the CA1 area of the hippocampus. In a study of stress-induced depressive mice, the expression changes of MeCP2, Sirt1, and/or neurotrophic factors in the hippocampus were consistent [52]. Widely present in the central nervous system, BDNF supports healthy neurodevelopment and is intimately linked to synaptic plasticity in neurons. In an animal study, it was discovered that exogenous BDNF could remarkably recover neuronal synapse function, whereas BDNF knockout caused major impairments in synaptic transmission and defects in the hippocampus LTP [53]. Given its function in synaptic repair, BDNF is considered a strategy for repair in neurodegenerative diseases [54]. Our results indicated that the level of BDNF was statistically lower in the ALC group compared with the CRC group. Therefore, CR-relieved cognitive function might be related to a higher level of BDNF in the hippocampal neurons. Surprisingly, this difference could not be found between the CRS group and the ALS group. But there was an increasing trend of expression in the CRS group. The factor leading to this difference may be related to our small sample size. Huang’s experimental data showed that CR could effectively promote the levels of BDNF in the hippocampus [55], which was similar to our results. By now, CR has been studied for a number of years, and there exists a variety of treatment strategies. The majority suggest that CR is beneficial for the human body, either in extending the lifespan or protecting neurological functions. All in all, this study indicated a new idea and provided a theoretical basis for CR in the prevention and treatment of POCD. Nevertheless, there are limitations of this study. First, postoperative pain and its management are closely related to the occurrence of POCD [56]. Whether the application of an anesthetic cream ($2.5\%$ lidocaine and $2.5\%$ prilocaine) could produce neuroprotective effects and reduce the incidence of POCD requires further investigation. Second, cholesterol levels were not monitored during feeding. Many studies have suggested that higher circulating TC is associated with an increased risk of cognitive impairment, especially in female subjects [57]. Therefore, further research is needed to confirm this point. Third, the behavioral test was conducted 1–7 days after surgery. The neuroprotective effects of CR for a long time after surgery demand further study. ## 5. Conclusions To conclude, our study identified that preoperative CR could not only decrease body weight with normal concentrations of blood glucose but also provide a protective effect on the learning and memory abilities in elderly mice after surgery. This beneficial action was partially regulated through increased expressions of SIRT1, MeCP2, and BDNF in the hippocampus. Further, the level of these three proteins in the CA1 region of the hippocampus may be responsible for that process. ## References 1. Evered L.A., Silbert B.S.. Postoperative cognitive dysfunction and noncardiac surgery. Obstet. Anesth. Dig. (2018) 127 496-505. DOI: 10.1213/ANE.0000000000003514 2. Liu J., Huang K., Zhu B., Zhou B., Ahmad Harb A.K., Liu L., Wu X.. Neuropsychological tests in post-operative cognitive dysfunction: Methods and applications. Front. Psychol. (2021) 12 684307. DOI: 10.3389/fpsyg.2021.684307 3. Rickenbacher M., Reinbold C.S., Herms S., Hoffmann P., Cichon S., Wueest A.S., Monsch A.U., Steiner L.A., Goettel N.. Genome-wide association study of postoperative cognitive dysfunction in older surgical patients. J. Neurosurg. Anesthesiol. (2020) 34 248-250. DOI: 10.1097/ANA.0000000000000749 4. Yang X., Huang X., Li M., Jiang Y., Zhang H.. Identification of individuals at risk for postoperative cognitive dysfunction (POCD). Ther Adv. Neurol. Disord. (2022) 15 17562864221114356. DOI: 10.1177/17562864221114356 5. Ida M., Kawaguchi M.. Postoperative cognitive dysfunction after non-cardiac surgery. Masui. Jpn. J. Anesthesiol. (2014) 63 1228-1234 6. Davidson T.L., Hargrave S.L., Swithers S.E., Sample C.H., Fu X., Kinzig K.P., Zheng W.. Inter-relationships among diet, obesity and hippocampal-dependent cognitive function. Neuroscience (2013) 253 110-122. DOI: 10.1016/j.neuroscience.2013.08.044 7. Xu T., Bo L., Wang J., Zhao Z., Xu Z., Deng X., Zhu W.. Risk factors for early postoperative cognitive dysfunction after non-coronary bypass surgery in Chinese population. J. Cardiothorac. Surg. (2013) 8 204. DOI: 10.1186/1749-8090-8-204 8. Geng Y.J., Wu Q.H., Zhang R.Q.. Effect of propofol, sevoflurane, and isoflurane on postoperative cognitive dysfunction following laparoscopic cholecystectomy in elderly patients: A randomized controlled trial. J. Clin. Anesth. (2017) 38 165. DOI: 10.1016/j.jclinane.2017.02.007 9. Yang Y., Zhang L.. The effects of caloric restriction and its mimetics in Alzheimer’s disease through autophagy pathways. Food Funct. (2020) 11 1211-1224. DOI: 10.1039/C9FO02611H 10. McCay C., Crowell M.F., Maynard L.A.. The effect of retarded growth upon the length of life span and upon the ultimate body size. J. Nutr. (1935) 10 63-79. DOI: 10.1093/jn/10.1.63 11. Zhang L., Xu H., Ding N., Li X., Chen X., Chen Z.. Beneficial effects on brain micro-environment by caloric restriction in alleviating neurodegenerative diseases and brain aging. Front. Physiol. (2021) 12 715443. DOI: 10.3389/fphys.2021.715443 12. Gillette-Guyonnet S., Vellas B.. Caloric restriction and brain function. Curr. Opin. Clin. Nutr. Metab. Care (2008) 11 686-692. DOI: 10.1097/MCO.0b013e328313968f 13. Prehn K., von Schwartzenberg R.J., Mai K., Zeitz U., Witte A.V., Hampel D., Szela A.-M., Fabian S., Grittner U., Spranger J.. Caloric restriction in older adults—Differential effects of weight loss and reduced weight on brain structure and function. Cereb. Cortex (2016) 27 1765-1778. DOI: 10.1093/cercor/bhw008 14. Izquierdo I.. The hippocampus and learning. Prog. Neurobiol. (1975) 5 37-75. DOI: 10.1016/0301-0082(75)90007-6 15. Opitz B.. Memory function and the hippocampus. Front. Neurol. Neurosci. (2014) 34 51-59. DOI: 10.1159/000356422 16. Poulose S.M., Miller M.G., Scott T., Shukitt Hale B.. Nutritional factors affecting adult neurogenesis and cognitive function. Adv. Nutr. (2017) 8 804-811. DOI: 10.3945/an.117.016261 17. Wang X., Li Y., Zhao J., Yu J., Zhang Q., Xu F., Zhang Y., Zhou Q., Yin C., Hou Z.. Activation of astrocyte Gq pathway in hippocampal CA1 region attenuates anesthesia/surgery induced cognitive dysfunction in aged mice. Front. Aging Neurosci. (2022) 14 1040569. DOI: 10.3389/fnagi.2022.1040569 18. Ginsberg S.D., Malek-Ahmadi M.H., Alldred M.J., Che S., Elarova I., Chen Y., Jeanneteau F., Kranz T.M., Chao M.V., Counts S.E.. Selective decline of neurotrophin and neurotrophin receptor genes within CA1 pyramidal neurons and hippocampus proper: Correlation with cognitive performance and neuropathology in mild cognitive impairment and Alzheimer’s disease. Hippocampus (2019) 29 422-439. DOI: 10.1002/hipo.22802 19. Gu H., Li N., Tang Y., Yan C., Shi Z., Yi S., Zhou H., Liao D., OuYang X.. Nicotinate curcumin ameliorates cognitive impairment in diabetic rats by rescuing autophagic flux in CA1 hippocampus. CNS Neurosci. Ther. (2018) 25 430-441. DOI: 10.1111/cns.13059 20. Hofer S.J., Carmona-Gutierrez D., Mueller I., Madeo M.. The ups and downs of caloric restriction and fasting: From molecular effects to clinical application. EMBO Mol. Med. (2021) 14 e14418. DOI: 10.15252/emmm.202114418 21. Mitchell S.J., Madrigal-Matute J., Scheibye-Knudsen M., Fang E., Aon M., González-Reyes J.A., Cortassa S., Kaushik S., Gonzalez-Freire M., Patel B.. Effects of sex, strain, and energy intake on hallmarks of aging in mice. Cell Metab. (2016) 23 1093-1112. DOI: 10.1016/j.cmet.2016.05.027 22. Acosta-Rodríguez V.A., de Groot M.H., Rijo-Ferreira F., Green C.B., Takahashi J.S.. Mice under caloric restriction self-impose a temporal restriction of food intake as revealed by an automated feeder system. Cell Metab. (2017) 26 267-277. DOI: 10.1016/j.cmet.2017.06.007 23. Michán S., Li Y., Chou M.M., Parrella E., Ge H., Long J.M., Allard J.S., Lewis K., Miller M., Xu W.. SIRT1 is essential for nor-mal cognitive function and synaptic plasticity. J. Neurosci. (2010) 30 9695-9707. DOI: 10.1523/JNEUROSCI.0027-10.2010 24. Sun Z., Zhao S., Suo X., Dou Y.. Sirt1 protects against hippocampal atrophy and its induced cognitive impairment in middle-aged mice. BMC Neurosci. (2022) 23. DOI: 10.1186/s12868-022-00718-8 25. Ma L., Dong W., Wang R., Li Y., Xu B., Zhang J., Zhao Z., Wang Y.. Effect of caloric restriction on the SIRT1/mTOR signaling pathways in senile mice. Brain Res. Bull. (2015) 116 67-72. DOI: 10.1016/j.brainresbull.2015.06.004 26. Sánchez-Lafuente C.L., Kalynchuk L.E., Caruncho H.J., Ausió J.. The role of MeCP2 in regulating synaptic plasticity in the context of stress and depression. Cells (2022) 11. DOI: 10.3390/cells11040748 27. Huang J., Zhang F., Su M., Li J., Yi W., Hou L., Yang S., Liu J., Zhang H., Ma T.. MeCP2 prevents age-associated cognitive decline via restoring synaptic plasticity in a senescence-accelerated mouse model. Aging Cell (2021) 20 e13451. DOI: 10.1111/acel.13451 28. Notaras M., van den Buuse M.. Brain-derived neurotrophic factor (BDNF): Novel insights into regulation and genetic variation. Neuroscientist (2019) 25 434-454. DOI: 10.1177/1073858418810142 29. Lu B., Nagappan G., Lu Y.. BDNF and synaptic plasticity, cognitive function, and dysfunction. Neurotrophic Factors (2014) 220 223-250. DOI: 10.1007/978-3-642-45106-5_9 30. Seidler K., Barrow M.. Intermittent fasting and cognitive performance—Targeting BDNF as potential strategy to optimise brain health. Front. Neuroendocr. (2021) 65 100971. DOI: 10.1016/j.yfrne.2021.100971 31. Wang Y.. Molecular Links between Caloric Restriction and Sir2/SIRT1 Activation. Diabetes Metab. J. (2014) 38 321-329. DOI: 10.4093/dmj.2014.38.5.321 32. Araya A.V., Orellana X., Espinoza J.. Evaluation of the effect of caloric restriction on serum BDNF in overweight and obese subjects: Preliminary evidences. Endocrine (2008) 33 300-304. DOI: 10.1007/s12020-008-9090-x 33. Zocchi L., Sassone-Corsi P.. SIRT1-mediated deacetylation of MeCP2 contributes to BDNF expression. Epigenetics (2012) 7 695-700. DOI: 10.4161/epi.20733 34. Veyrat-Durebex C., Quirion R., Ferland G., Dumont Y., Gaudreau P.. Aging and long-term caloric restriction regulate neuropeptide Y receptor subtype densities in the rat brain. Neuropeptides (2013) 47 163-169. DOI: 10.1016/j.npep.2013.01.001 35. Cesarovic N., Nicholls F., Rettich A., Kronen P., Hässig M., Jirkof P., Arras M.. Isoflurane and sevoflurane provide equally effective anaesthesia in laboratory mice. Lab. Anim. (2010) 44 329-336. DOI: 10.1258/la.2010.009085 36. Terrando N., Monaco C., Ma D., Foxwell B.M., Feldmann M., Maze M.. Tumor necrosis factor alpha triggers a cytokine cascade yielding postoperative cognitive decline. Proc. Natl. Acad. Sci. USA (2010) 107 20518-20522. DOI: 10.1073/pnas.1014557107 37. Quintas A., de Solís A.J., Díez-Guerra F.J., Carrascosa J.M., Bogónez E.. Age-associated decrease of SIRT1 expression in rat hippocampus: Prevention by late onset caloric restriction. Exp. Gerontol. (2012) 47 198. DOI: 10.1016/j.exger.2011.11.010 38. Wang C.-M., Chen W.-C., Zhang Y., Lin S., He H.-F.. Update on the mechanism and treatment of sevoflurane-induced postoperative cognitive dysfunction. Front. Aging Neurosci. (2021) 13 702231. DOI: 10.3389/fnagi.2021.702231 39. Ju H., Wang Y., Shi Q., Zhou Y., Ma R., Wu P., Fang H.. Inhibition of connexin 43 hemichannels improves postoperative cognitive function in aged mice. Am. J. Transl. Res. (2019) 11 2280-2287. PMID: 31105835 40. Sun L., Dong R., Xu X., Yang X., Peng M.. Activation of cannabinoid receptor type 2 attenuates surgery-induced cognitive impairment in mice through anti-inflammatory activity. J. Neuroinflammation (2017) 14 138. DOI: 10.1186/s12974-017-0913-7 41. Dong W., Wang R., Ma L.-N., Xu B.-L., Zhang J.-S., Zhao Z.-W., Wang Y.-L., Zhang X.. Autophagy involving age-related cognitive behavior and hippocampus injury is modulated by different caloric intake in mice. Int. J. Clin. Exp. Med. (2015) 8 11843-11853. PMID: 26380026 42. Ma L., Wang R., Dong W., Zhao Z.. Caloric restriction can improve learning and memory in C57/BL mice probably via regulation of the AMPK signaling pathway. Exp. Gerontol. (2018) 102 28-35. DOI: 10.1016/j.exger.2017.11.013 43. Zeier Z., Madorsky I., Xu Y., Ogle W.O., Notterpek L., Foster T.C.. Gene expression in the hippocampus: Regionally specific effects of aging and caloric restriction. Mech. Ageing Dev. (2011) 132 8-19. DOI: 10.1016/j.mad.2010.10.006 44. Corpas R., Revilla S., Ursulet S., Castro-Freire M., Kaliman P., Petegnief V., Giménez-Llort L., Sarkis C., Pallàs M., Sanfeliu C.. SIRT1 overexpression in mouse hippocampus induces cognitive enhancement through proteo-static and neurotrophic mechanisms. Mol. Neurobiol. (2017) 54 5604-5619. DOI: 10.1007/s12035-016-0087-9 45. Zhao Y., Zhang J., Zheng Y., Zhang Y., Zhang X.J., Wang H., Du Y., Guan J., Wang X., Fu J.. NAD+ improves cognitive function and reduces neuroinflammation by ameliorating mitochondrial damage and decreasing ROS production in chronic cerebral hypoperfusion models through Sirt1/PGC-1α pathway. J. Neuroinflammation. (2021) 18 207. DOI: 10.1186/s12974-021-02250-8 46. Tang X.L., Wang X., Fang G., Zhao Y.L., Yan J., Zhou Z., Sun R., Luo A.L., Li S.Y.. Resveratrol ameliorates sevoflurane-induced cognitive impairment by activating the SIRT1/NF-κB pathway in neonatal mice. J. Nutr. Biochem. (2021) 90 108579. DOI: 10.1016/j.jnutbio.2020.108579 47. Wahl D., Solon-Biet S.M., Wang Q.-P., Wali J.A., Pulpitel T., Clark X., Raubenheimer D., Senior A.M., Sinclair D.A., Cooney G.J.. Comparing the effects of low-protein and high-carbohydrate diets and caloric restriction on brain aging in mice. Cell Rep. (2018) 25 2234-2243.e6. DOI: 10.1016/j.celrep.2018.10.070 48. Gulmez Karaca K., Brito D.V.C., Zeuch B., Oliveira A.M.M.. Adult hippocampal MeCP2 preserves the genomic responsiveness to learning required for long-term memory formation. Neurobiol. Learn. Mem. (2018) 149 84-97. DOI: 10.1016/j.nlm.2018.02.010 49. Im H.I., Hollander J.A., Bali P., Kenny P.J.. MeCP2 controls BDNF expression and cocaine intake through homeostatic interactions with microRNA-212. Nat. Neurosci. (2010) 13 1120-1127. DOI: 10.1038/nn.2615 50. Su M., Hong J., Zhao Y., Liu S., Xue X.. MeCP2 controls hippocampal brain-derived neurotrophic factor expression via homeostatic interactions with microRNA-132 in rats with depression. Mol. Med. Rep. (2015) 12 5399-5406. DOI: 10.3892/mmr.2015.4104 51. Na E.S., Nelson E.D., Kavalali E.T., Monteggia L.M.. The impact of MeCP2 loss or gain of function on synaptic plasticity. Neuropsychopharmacology (2013) 38 212-219. DOI: 10.1038/npp.2012.116 52. Choi J., Kwon H., Kim Y.K., Han P.L.. Extracellular vesicles from gram-positive and gram-negative probiotics re-mediate stress-induced depressive behavior in mice. Mol. Neurobiol. (2022) 59 2715-2728. DOI: 10.1007/s12035-021-02655-9 53. Patterson S.L., Abel T., Deuel T.A., Martin K.C., Rose J.C., Kandel E.R.. Recombinant BDNF rescues deficits in basal synaptic transmission and hippocampal LTP in BDNF knockout mice. Neuron (1996) 16 1137. DOI: 10.1016/S0896-6273(00)80140-3 54. Lu B., Nagappan G., Guan X., Nathan P.J., Wren P.. BDNF-based synaptic repair as a disease-modifying strategy for neurodegenerative diseases. Nat. Rev. Neurosci. (2013) 14 401. DOI: 10.1038/nrn3505 55. Huang Y.J., Zhang L., Shi L.Y., Wang Y.Y., Yang Y.B., Ke B., Zhang T.Y., Qin J.. Caloric restriction ameliorates acrole-in-induced neurotoxicity in rats. Neurotoxicology (2018) 65 44-51. DOI: 10.1016/j.neuro.2018.01.003 56. Wang Y., Sands L.P., Vaurio L., Mullen E.A., Leung J.M.. The effects of postoperative pain and its management on postoperative cognitive dysfunction. Am. J. Geriatr. Psychiatry (2007) 15 50-59. DOI: 10.1097/01.JGP.0000229792.31009.da 57. Guo Y., Li P., Ma X., Huang X., Liu Z., Ren X., Yang Y., Halm-Lutterodt N.V., Yuan L.. Association of circulating cholesterol level with cognitive function and mild cognitive impairment in the elderly: A community-based population study. Curr. Alzheimer Res. (2020) 17 556-565. DOI: 10.2174/1567205017666200810165758
--- title: 'Sex Differences in Survival from Neuroendocrine Neoplasia in England 2012–2018: A Retrospective, Population-Based Study' authors: - Benjamin E. White - Beth Russell - Sebastiaan Remmers - Brian Rous - Kandiah Chandrakumaran - Kwok F. Wong - Mieke Van Hemelrijck - Rajaventhan Srirajaskanthan - John K. Ramage journal: Cancers year: 2023 pmcid: PMC10046836 doi: 10.3390/cancers15061863 license: CC BY 4.0 --- # Sex Differences in Survival from Neuroendocrine Neoplasia in England 2012–2018: A Retrospective, Population-Based Study ## Abstract ### Simple Summary We conducted a retrospective, population-based study comparing overall survival (OS) between males and females with neuroendocrine neoplasia (NEN). In total, 14,834 cases of NEN recorded in England’s National Cancer Registry and Analysis Service (NCRAS)), were analysed. Multivariable analysis, restricted mean survival time and mediation analysis were performed. Females displayed increased survival irrespective of the stage, morphology or level of deprivation, which was statistically significant in NEN of the lung, pancreas, rectum and stomach ($p \leq 0.001$). Stage of tumour mediated improved survival in stomach, lung, and pancreatic NEN but not in rectal NEN. Females diagnosed with NEN tend to survive longer than males, and stage at presentation only accounts for part of this effect. Future research in NEN, as well as prognostication and treatment, should consider sex as an important factor. ### Abstract Pre-clinical studies have suggested sex hormone signalling pathways may influence tumorigenesis in neuroendocrine neoplasia (NEN). We conducted a retrospective, population-based study to compare overall survival (OS) between males and females with NEN. A total of 14,834 cases of NEN diagnosed between 2012 and 2018, recorded in England’s National Cancer Registry and Analysis Service (NCRAS), were analysed. The primary outcome was OS with 5 years maximum follow-up. Multivariable analysis, restricted mean survival time and mediation analysis were performed. Appendiceal, pulmonary and early-stage NEN were most commonly diagnosed in females; stomach, pancreatic, small intestinal, colonic, rectal and later-stage NEN were more often diagnosed in males. Females displayed increased survival irrespective of the stage, morphology or level of deprivation. On average, they survived 3.62 ($95\%$ CI 1.73–5.90) to 10.26 (6.6–14.45) months longer than males; this was statistically significant in NEN of the lung, pancreas, rectum and stomach ($p \leq 0.001$). The stage mediated improved survival in stomach, lung, and pancreatic NEN but not in rectal NEN. The reasons underlying these differences are not yet understood. Overall, females diagnosed with NEN tend to survive longer than males, and the stage at presentation only partially explains this. Future research, as well as prognostication and treatment, should consider sex as an important factor. ## 1. Introduction Neuroendocrine neoplasia (NEN) are tumours arising from neuroendocrine cells; they share the traits of both nervous and endocrine cells and can release hormones in response to neuronal stimuli. NEN can be classified as well-differentiated neuroendocrine tumours (NET) or poorly differentiated neuroendocrine carcinoma (NEC), which include large and small cell differentiation. Although NEN can occur anywhere in the body, the majority arise in the gastrointestinal tract and lungs [1,2]. The symptomatology can be highly variable and is dependent on the tumour burden and hormone-secreting capacity [3]. The incidence of NEN is increasing globally [4]; theories to explain this include increased clinical awareness, more widespread availability of imaging techniques and endoscopy, and a possible ‘real’ increase. Risk factors for developing NEN include a family history of cancer, type 2 diabetes, obesity, cigarette smoking and alcohol intake. These findings are from retrospective case–control studies; high-quality prospective trials to identify risk factors have not yet been performed in NEN [5,6,7]. Survival at each NEN tumour site has improved in England over the last 25 years. This is likely due to a combination of increased detection of low-stage tumours resulting in ‘stage shift’ and the effect of treatment advances [8]. The predictors of survival at diagnosis of NEN identified so far are age, sex, organ site, stage, grade, deprivation (also known as socio-economic status) and marital status. As in other solid organ cancers, there are survival differences by sex in NEN. Population-based studies from North America that include large numbers of tumours have shown males to have statistically significant worse overall survival (OS), with HRs up to 1.26–1.27 for males compared to females [9,10]. Large cohort analyses from other population-based databases in England, Canada, Australia, Norway, Taiwan and others also demonstrated statistically significant worse OS in males in multivariable analysis [11,12,13,14,15]. Overall, sex plays an important role in survival from solid organ cancers [16]. Sex is a key modifier of pathophysiology via genetic, epigenetic and hormonal regulation. A different biological sex environment is created by genetic heterogeneity at the molecular level. Evidence shows that sex hormones affect cellular responses by modifying DNA expression, which in turn leads to different cell surface receptor expression [17]. Not only does this result in differing predispositions to and the manifestation of malignancy but it also affects the response to cancer therapy. Societal factors also play a role by influencing behaviours such as diet, smoking and physical activity, which in turn can influence health outcomes. Perceived gender also affects how a patient is treated both by society and clinicians [16]. Male predominance in solid organ cancers that affect both sexes has been observed worldwide [18]. Males are known to have greater exposure to risk factors, such as occupational, alcohol intake and smoking risk factors [19]. Survival is shorter for males across multiple solid organ cancer types [20]. As described above, these differences may be explained by sex-specific biology having effects on tumorigenesis, the stimulatory effects of androgens in male individuals and the protective effects of oestrogens in females seen in non-reproductive cancers [21], in addition to the influence of the societal, cultural or behavioural effect of gender roles. Several pre-clinical studies have shown that the expression of oestrogen and progesterone receptors is associated with favourable outcomes amongst patients with gastroenteropancreatic NEN (GEP-NEN). Pancreatic NEN (pNEN) with higher oestrogen receptor-β expression are associated with a more favourable prognosis [22]. Females, especially those who are pre-menopausal, have a lower risk of mesenteric metastasis in small intestinal NEN (SI-NEN). SI-NEN have increased oestrogen and androgen receptor expression compared to normal tissue, suggesting that sex hormone signalling pathways may modulate metastatic potential [23,24,25]. Immunohistochemical assessment of progesterone receptor (PR) status may help to identify GEP-NEN with the potential for more aggressive behaviour [26,27]. Histological differences between males and females in lung carcinoids have also been identified [28]. A clinical trial is investigating the effect of tamoxifen in well-differentiated NEN based on oestrogen and progesterone receptors being expressed in around $20\%$ of NEN [29]. We aimed to use restricted mean survival time (RMST) and mediation analysis to compare survival by sex in NEN and examine the influence of the stage on survival outcomes. To our knowledge, there are no other studies that have yet analysed sex differences in NEN in this way. ## 2.1. Data Source This work utilised data from the National Cancer Registry and Analysis Service (NCRAS) of England, which captures over $99\%$ of tumours recorded in England’s National Health Service [30,31]. The data were collected for individuals aged 16 and above who had been diagnosed with NEN between 2012 and 2018. The NCRAS database is updated as histopathological classification systems change, which presents a challenge in a rapidly evolving field such as NEN. Stage is recorded by NCRAS according to the European Neuroendocrine Tumor Society (ENETS) system for foregut [32] and mid- and hindgut [33] tumours and uses the Union for International Cancer Control tumour, node and metastasis system (UICC TNM) [34] for other sites. ## 2.2. NEN Classification and Analytic Process NEN occurring at all anatomical sites between C00 and C80 and malignant neoplasms of all sites (excluding haematological malignancy), according to the 10th edition of the WHO International Classification of Disease (ICD-10) were included. The morphology codes included 8013 (excluding lung [C34 and C78]), 8041–8045 (excluding lung), 8150–8158, 8240–8247, 8249 and 9091, according to the WHO International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) [35] in line with previously published work on NEN based on NCRAS data [13]. Large cell neuroendocrine and small cell carcinomas of the lung were excluded to enable a comparison with previous analyses and because the high incidence in this organ due to smoking would skew the results. Goblet cell adenocarcinomas (GCA) (ICD-O-3: 8243) were excluded from the dataset in view of their reclassification as non-NEN [36]. Duplicate tumours and tumours recorded as ‘death certificate only’, which made up less than $0.1\%$ of the tumours, were excluded [37,38]. Mixed neuroendocrine non-neuroendocrine neoplasms (MiNEN) (formerly termed mixed adenoneuroendocrine carcinomas (MANEC)) and Merkel cell tumours were excluded. Only tumours diagnosed from 2012 onwards were included in the main survival analysis due to markedly improved coding and classification in recent years; unclassified stage tumours ($25.6\%$) were excluded. It was decided that imputing missing data was not desirable due to a risk of bias in the resulting dataset [39]. Site groups were created from histological codes. The main sites were defined as the appendix, caecum, colon, lung, pancreas, rectum, small intestine or stomach, in line with other series [40]. Tumours with a primary site not registered as one of these ‘main’ primaries were excluded in order to clearly define the cohort and avoid inaccuracy in analysing the likely metastatic sites. We, therefore, grouped the NEN morphologically, either as well-differentiated neuroendocrine tumours (NET) or poorly differentiated neuroendocrine carcinomas (NEC), similar to other recently published work [41]. The tumours classified as NET included carcinoids of typical, atypical, tubular and other well-differentiated neoplasms such as insulinoma and glucagonoma. The NEC included all the carcinomas and tumours with large and small cell neuroendocrine differentiation. Although all tumours have a histopathological classification, the Ki-67 index was not yet available on the NCRAS database at the time of the data transfer. The available variables suitable to be included in the analysis were site, age, sex, index of multiple deprivation (IMD), morphology and stage. The IMD is a measure of relative deprivation for small areas of England (lower layer super output areas, LSOA) and is composed of seven domains with relative weights: income ($22.5\%$), employment (22.5), education ($13.5\%$), health ($13.5\%$), crime ($9.3\%$), housing ($9.3\%$) and environment ($9.3\%$). ## 2.3. Statistical Analytic Approach The categorical variables were presented as percentages; the continuous variables were reported as the median and interquartile range (IQR). The primary endpoint was OS, calculated from the date of diagnosis and censored on 31 March 2020 and calculated using the Kaplan-Meier estimator with a maximum of 5 years follow-up. The $95\%$ confidence interval ($95\%$ CI) was specified for all the results. All the variables were included in the multivariable analysis except for ethnicity. Ethnicity was excluded due to skewed data. Cox regression multivariable analysis included sex, morphology, age group, stage, site and deprivation. Of these, sex and deprivation met proportional hazards assumptions. The other variables did not strictly meet proportional hazards assumptions and were therefore included in the final multivariable model as covariates with a time-varying effect (TVC). There was no multicollinearity between the variables. The accelerated failure time (AFT) models were tested for significance against the null models (Cox) using a likelihood ratio test ($p \leq 0.001$). We aimed to use age-adjusted restricted mean survival time (RMST) as a method to compare survival between the sexes. RMST is defined as the area under the survival curve up to a specific time point and can overcome some of the limitations of proportional hazard modelling [42]. The stage at presentation might explain some of the survival differences observed in NEN between males and females. Early-stage appendiceal and lung NEN, for example, occur more frequently in females, whilst late-stage pancreatic and stomach NEN occur more frequently in males [8]. Mediation analysis can be used to further study the relationship between sex and survival and how this is influenced by the stage [43]. Mediation analysis looks at how the relationship between an exposure (e.g., sex) and an outcome (e.g., survival) might be mediated by another variable (e.g., stage) whilst adjusting for other confounding factors (e.g., morphology, deprivation). RMST and age-adjusted RMST were calculated using the strmst2 command in STATA. Mediation analysis was performed using the med4way command in STATA. For the mediation analysis, as mediators in med4way can be either continuous or dichotomous, stages I and II were classed as ‘early’ stage and stages III and IV were classed as ‘late’ stage. The statistical analyses and plots were performed using STATA/MP 16.0 (College Station, TX, USA: StataCorp LLC) and R. ## 3. Results In total, 14,834 tumours recorded on the NCRAS database between 2012 and 2018 were eligible for analysis. The largest proportion of tumours occurred in the 65–74 age group, with a median age for the cohort of 65 (IQR 53-73) (Table 1). Closely matching the ethnic mix of England, the most frequent ethnicity was White ($89\%$), followed by Asian ($2.9\%$) and Black ($2.3\%$). The most common primary site was the lung (4661; $31.4\%$ of tumours), followed by the small intestine (3201; $21.6\%$), pancreas (2183; $14.7\%$) and appendix (2146; $14.5\%$). Most tumours in the cohort were either stage I (5040; $34.0\%$ of tumours) or stage IV (5121; $34.5\%$), with stages II and III being less frequent. There were 11,080 NET ($74.7\%$) and 3754 NEC ($25.3\%$). The tumours were spread relatively evenly across deprivation quintiles ($20.5\%$ to $18.3\%$) (Table 1). Overall, there were slightly more females than males diagnosed with NEN ($51.5\%$ female vs. $48.5\%$ male) (Table 2). The median age was similar in males and females (65.5 vs. 65). The youngest age group displayed female predominance ($60.9\%$), but this disparity ceased above age 54 where the tumours became more evenly distributed. The appendix and lung sites were predominantly diagnosed in females ($61.3\%$ and $60.2\%$ female), but stomach, pancreas, small intestine, colon and rectal NEN were most frequently diagnosed in males (56.3–$61.9\%$). Stage I and II tumours showed a female preponderance ($60.4\%$ and $53.6\%$ female). However, there were more stage III and IV tumours diagnosed in males ($53.4\%$ and $55.6\%$ male). The sex distribution was relatively equal across all the deprivation quintiles. There were more NET diagnosed in females ($54.0\%$ female). However, the opposite was true in NEC, where $56.0\%$ of the diagnoses occurred in males. As expected, increasing age was associated with progressively increased hazard ratios (HR); compared to the <30 age group, the HR was 4.41 ($95\%$ CI 2.88–6.74) for the 30–54 age group, 5.35 (3.50–8.18) for the 55–65 age group, 6.13 (4.01–9.37) for the 65–74 age group and 7.72 (5.06–11.80) for those over 75. The rectum 1.27 (1.15–1.41) and stomach 1.26 (1.14–1.33) had the highest HRs of any of the main sites when compared to the appendix. Increasing stage was associated with increasing hazard, and the same pattern was observed with increasing deprivation quintile. A NEC was associated with a significantly increased HR compared to a NET (HR 1.29 (1.25–1.33)) (Table 3). The age-adjusted 5-year RMST (Table 3) of the main sites showed that females displayed a survival advantage ranging from 3.62 ($95\%$ CI 1.73 to 5.90) to 10.26 (6.6 to 14.45) months. The exception was colonic NEN, which showed a male survival advantage, but this was not statistically significant. The sites where females showed a statistically significant improved survival were the lung, pancreas, rectum and stomach (all $p \leq 0.001$). Females had an increased survival at all tumour stages: 1.2 months (0.48 to 1.92) for stage I, 3.24 months (1.76 to 4.72) for stage II, 2.23 months (0.58 to 3.89) for stage III and 3.19 months (1.84 to 4.54) for stage IV. All these results were statistically significant ($p \leq 0.001$ except Stage III $$p \leq 0.008$$). Compared to the males, females survived longer when diagnosed with both morphological groups of NET or NEC, with an HR of 2.44 (1.74 to 3.13) and 4.92 (3.46 to 6.37) months, respectively ($p \leq 0.001$). Similarly, females had a longer survival in all the deprivation quintiles (4.45 to 5.14 months, $p \leq 0.001$). Four-way decomposition mediation analysis of the four main primary sites found females to have a statistically significant survival advantage. Table 4 shows that females are less likely to be diagnosed at a later stage than males. Consistent with the age-adjusted RMST findings, females survived longer than males in all four sites according to the model for the outcome at each site. The stage mediated improved survival in females to a significant extent ($37\%$) in stomach NEN and moderately ($20\%$) in lung and pancreas NEN. The stage did not play a role ($4\%$) in mediating survival in rectal NEN (Figure 1). ## 4. Discussion This study demonstrates that females diagnosed with neuroendocrine neoplasia display a survival advantage compared to males. Sex is a statistically significant predictor of survival in multivariable analysis [8]. The survival advantage for females remained statistically significant when examining the subgroups (Table 3). When examining the primary sites, those sites with a statistically significant increase in overall survival in females are the lung, pancreas, rectum and stomach. Mediation analysis suggested that the stage is responsible for the survival advantage seen in females to different extents depending on the primary site. The stage (early or late) at diagnosis appears to explain the survival advantage in stomach NEN to a large extent, in lung and pancreas NEN moderately and not at all in rectal NEN. The reasons underlying these differences are not yet understood. The stage is, therefore, an important intermediate (i.e., a mediator) on the pathway of the association between sex and survival and might help direct investigation of underlying causes. Although we have demonstrated that the stage partially explains differential survival according to sex in NEN, it is not clear why females tend to be diagnosed earlier than males with lung NEN, whilst the reverse is true for pancreatic and stomach NEN. Females were also observed to survive longer than males, even when diagnosed with the same stage and morphological type of NEN. Suggested explanations put forward for this in the past include biological reasons (including genetic, hormonal and other factors) [17,18] and environmental reasons (including behavioural, societal and cultural factors) [16]. Other country or health system-specific factors may also play a role, such as screening programmes for cancer or prominent public health campaigns. As described previously, there is an increasing body of pre-clinical research seeking to explain the sex difference in NEN. The expression of oestrogen and progesterone receptors, or sex hormone signalling pathways, may play a role in the differing biology between the sexes and tumorigenesis [22,23,24,25,26,27]. We suggest that future research could try to explain why females are presenting with earlier-stage NEN of the lung and males are presenting with later-stage NEN of the pancreas and stomach. This could be examined by looking at the mode of presentation or diagnosis, not sufficiently complete for us to analyse. The linkage of general practice records at the patient level, which would enable a richer analysis of risk factors such as smoking and obesity, might also help to explain the difference [44]. At presentation, it would be difficult to distinguish late-stage NEN, which have been slow-growing and undiscovered but may have transformed recently from aggressive ones, which have only been present for a short time. NEN could have had differing durations of exposure to an oncogenic microenvironment, which might mean a greater impact of sex differences, for example, the duration of exposure to sex hormones. To investigate this, a study model could be devised to predict the risk of the development of NEN compared to a background rate in the population before and after menopause. Since NEN are thought to be slow-growing tumours, commonly with the tumour being present both before and after menopause, this study design may be complex [25]. The use of Ki-67 in future studies would be beneficial, allowing for the mitotic index to be taken into account when analysing the differences in survival. Ki-67 has been recorded in the NCRAS database from 2020 onwards, reflecting how histological classification has developed over time, representing a good opportunity for further investigation [45]. Another model to further characterise how sex influences survival in NEN might be to retrospectively examine the differences between patients with a co-diagnosis of NEN and prostate cancer, having or not having antiandrogen therapy, as the role of androgen deprivation in survival from these tumours is unclear [46]. The limitations of this retrospective, population-based study include a historic lack of quality recorded data before 2012, particularly regarding the stage and morphology, meaning it was not possible to accurately compare these findings to earlier time periods. We had to rely on morphology to characterise the tumours, without a grade or Ki-67, again due to the incompleteness of the data. It was not possible to accurately analyse the diagnostic imaging pathways, chemotherapy treatments or health system routes to diagnosis due to the incompleteness of the data. However, this is improving over time as the NCRAS database becomes more complete. ## 5. Conclusions We have demonstrated that females diagnosed with NEN tend to survive longer than males. The stage at presentation is only partially responsible for this difference and does not explain the underlying causes. It is not possible in this analysis to demonstrate causality with respect to sex hormones or other sex differences, such as treatment histories, which may be influencing this relationship. More research is needed to understand how sex affects presentation, disease progression and treatment response in NEN. Our research suggests that prognostication and treatment should take sex and gender into account. We suggest that future trials in NEN should consider and report on sex and gender throughout the research process. ## References 1. Cives M., Pelle’ E., Strosberg J.. Emerging Treatment Options for Gastroenteropancreatic Neuroendocrine Tumors. J. Clin. Med. (2020.0) 9. DOI: 10.3390/jcm9113655 2. Caplin M.E., Baudin E., Ferolla P., Filosso P., Garcia-Yuste M., Lim E., Oberg K., Pelosi G., Perren A., Rossi R.E.. Pulmonary neuroendocrine (carcinoid) tumors: European Neuroendocrine Tumor Society expert consensus and recommendations for best practice for typical and atypical pulmonary carcinoids. Ann. Oncol. Off. J. Eur. Soc. Med. Oncol. (2015.0) 26 1604-1620. DOI: 10.1093/annonc/mdv041 3. Clement D., Ramage J., Srirajaskanthan R.. Update on Pathophysiology, Treatment, and Complications of Carcinoid Syndrome. J. Oncol. (2020.0) 2020 8341426. DOI: 10.1155/2020/8341426 4. Das S., Dasari A.. Epidemiology, Incidence, and Prevalence of Neuroendocrine Neoplasms: Are There Global Differences?. Curr. Oncol. Rep. (2021.0) 23 43. DOI: 10.1007/s11912-021-01029-7 5. Hassan M.M., Phan A., Li D., Dagohoy C.G., Leary C., Yao J.C.. Risk factors associated with neuroendocrine tumors: A U.S.-based case-control study. Int. J. Cancer (2008.0) 123 867-873. DOI: 10.1002/ijc.23529 6. Haugvik S.-P., Hedenström P., Korsæth E., Valente R., Hayes A., Siuka D., Maisonneuve P., Gladhaug I.P., Lindkvist B., Capurso G.. Diabetes, Smoking, Alcohol Use, and Family History of Cancer as Risk Factors for Pancreatic Neuroendocrine Tumors: A Systematic Review and Meta-Analysis. Neuroendocrinology (2015.0) 101 133-142. DOI: 10.1159/000375164 7. Feola T., Puliani G., Sesti F., Modica R., Centello R., Minotta R., Cannavale G., Di Meglio S., Di Vito V., Lauretta R.. Risk factors for gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs): A three-centric case–control study. J. Endocrinol. Investig. (2022.0) 45 849-857. DOI: 10.1007/s40618-021-01715-0 8. White B.E., Rous B., Chandrakumaran K., Wong K., Bouvier C., Van Hemelrijck M., George G., Russell B., Srirajaskanthan R., Ramage J.K.. Incidence and survival of neuroendo-crine neoplasia in England 1995–2018: A retrospective, population-based study. Lancet Reg. Health-Eur. (2022.0) 23 100510. DOI: 10.1016/j.lanepe.2022.100510 9. Man D., Wu J., Shen Z., Zhu X.. Prognosis of patients with neuroendocrine tumor: A SEER database analysis. Cancer Manag. Res. (2018.0) 10 5629-5638. DOI: 10.2147/CMAR.S174907 10. Abdel-Rahman O., Fazio N.. Sex-Based Differences in Prognosis of Patients With Gastroenteropancreatic-Neuroendocrine Neoplasms: A Population-Based Study. Pancreas (2021.0) 50 727-731. DOI: 10.1097/MPA.0000000000001821 11. Hallet J., Law C.H.L., Cukier M., Saskin R., Liu N., Singh S.. Exploring the rising incidence of neuroendocrine tumors: A population-based analysis of epidemiology, metastatic presentation, and outcomes. Cancer (2015.0) 121 589-597. DOI: 10.1002/cncr.29099 12. Boyar Cetinkaya R., Aagnes B., Myklebust T.Å., Thiis-Evensen E.. Survival in neuroendocrine neoplasms; A report from a large Norwegian population-based study. Int. J. Cancer (2018.0) 142 1139-1147. DOI: 10.1002/ijc.31137 13. Genus T.S.E., Bouvier C., Wong K.F., Srirajaskanthan R., Rous B.A., Talbot D.C., Valle J.W., Khan M., Pearce N., Elshafie M.. Impact of neuroendocrine morphology on cancer outcomes and stage at diagnosis: A UK nationwide cohort study 2013–2015. Br. J. Cancer (2019.0) 121 966-972. DOI: 10.1038/s41416-019-0606-3 14. Wyld D., Wan M.H., Moore J., Dunn N., Youl P.. Epidemiological trends of neuroendocrine tumours over three decades in Queensland, Australia. Cancer Epidemiol. (2019.0) 63 101598. DOI: 10.1016/j.canep.2019.101598 15. Chang J.S., Chen L.-T., Shan Y.-S., Chu P.-Y., Tsai C.-R., Tsai H.-J.. An updated analysis of the epidemiologic trends of neuroendocrine tumors in Taiwan. Sci. Rep. (2021.0) 11 7881. DOI: 10.1038/s41598-021-86839-2 16. Mauvais-Jarvis F., Merz N.B., Barnes P.J., Brinton R.D., Carrero J.J., DeMeo D.L., De Vries G.J., Epperson C.N., Govindan R., Klein S.L.. Sex and gender: Modifiers of health, disease, and medicine. Lancet (2020.0) 396 565-582. DOI: 10.1016/S0140-6736(20)31561-0 17. Claessens F., Tilley W.. Androgen signalling and steroid receptor crosstalk in endocrine cancers. Endocr.-Relat. Cancer (2014.0) 21 E3-E5. DOI: 10.1530/ERC-14-0274 18. Clocchiatti A., Cora E., Zhang Y., Dotto G.P.. Sexual dimorphism in cancer. Nat. Rev. Cancer (2016.0) 16 330-339. DOI: 10.1038/nrc.2016.30 19. McCartney G., Mahmood L., Leyland A., Batty G.D., Hunt K.. Contribution of smoking-related and alcohol-related deaths to the gender gap in mortality: Evidence from 30 European countries. Tob. Control. (2011.0) 20 166-168. DOI: 10.1136/tc.2010.037929 20. Wagner A., Oertelt-Prigione S., Adjei A., Buclin T., Cristina V., Csajka C., Coukos G., Dafni U., Dotto G.-P., Ducreux M.. Gender medicine and oncology: Report and consensus of an ESMO workshop. Ann. Oncol. (2019.0) 30 1914-1924. DOI: 10.1093/annonc/mdz414 21. Li Y., Xu A., Jia S., Huang J.. Recent advances in the molecular mechanism of sex disparity in hepatocellular carcinoma. Oncol. Lett. (2019.0) 17 4222-4228. DOI: 10.3892/ol.2019.10127 22. Estrella J.S., Ma L.T., Milton D.R., Yao J.C., Wang H., Rashid A., Broaddus R.R.. Expression of estrogen-induced genes and estrogen receptor β in pancreatic neuroendocrine tumors: Implications for targeted therapy. Pancreas (2014.0) 43 996-1002. DOI: 10.1097/MPA.0000000000000203 23. Blažević A., Zandee W., Franssen G.J.H., Hofland J., van Velthuysen M.-L., Hofland L.J., Feelders R.A., De Herder W.W.. Mesenteric fibrosis and palliative surgery in small intestinal neuroendocrine tumours. Endocr.-Relat. Cancer (2018.0) 25 245-254. DOI: 10.1530/ERC-17-0282 24. Blažević A., Brabander T., Zandee W., Hofland J., Franssen G., van Velthuysen M.-L., Feelders R., De Herder W.. Evolution of the Mesenteric Mass in Small Intestinal Neuroendocrine Tumours. Cancers (2021.0) 13. DOI: 10.3390/cancers13030443 25. Blažević A., Iyer A.M., van Velthuysen M.-L.F., Hofland J., Oudijk L., de Herder W.W., Hofland L.J., Feelders R.A.. Sexual Dimorphism in Small-intestinal Neuroendocrine Tumors: Lower Prevalence of Mesenteric Disease in Premenopausal Women. J. Clin. Endocrinol. Metab. (2022.0) 107 e1969-e1975. DOI: 10.1210/clinem/dgac001 26. Arnason T., Sapp H.L., Barnes P.J., Drewniak M., Abdolell M., Rayson D.. Immunohistochemical expression and prognostic value of ER, PR and HER2/neu in pancreatic and small intestinal neuroendocrine tumors. Neuroendocrinology (2011.0) 93 249-258. DOI: 10.1159/000326820 27. Estrella J.S., Broaddus R.R., Mathews A., Milton D.R., Yao J.C., Wang H., Rashid A.. Progesterone Receptor and PTEN Expression Predict Survival in Patients With Low- and Intermediate-Grade Pancreatic Neuroendocrine Tumors. Arch. Pathol. Lab. Med. (2014.0) 138 1027-1036. DOI: 10.5858/arpa.2013-0195-OA 28. Daskalakis K., Kaltsas G., Öberg K., Tsolakis A.V.. Lung Carcinoids: Long-Term Surgical Results and the Lack of Prognostic Value of Somatostatin Receptors and Other Novel Immunohistochemical Markers. Neuroendocrinology (2018.0) 107 355-365. DOI: 10.1159/000493944 29. Barros M., Felismino T., de Jesus V., Mello C., Silva V., Camandaroba M., Rodrigues N., Donadio M., Nobrega E., Chinen L.. HORMONET: Study of tamoxifen in well differentiated neuroendocrine tumours and hormone receptor positive expression. Ann. Oncol. (2019.0) 30 v573. DOI: 10.1093/annonc/mdz256.024 30. Henson K.E., Elliss-Brookes L., Coupland V.H., Payne E., Vernon S., Rous B., Rashbass J.. Data Resource Profile: National Cancer Registration Dataset in England. Int. J. Epidemiol. (2020.0) 49 16-16h. DOI: 10.1093/ije/dyz076 31. Collecting and Using Data 32. Rindi G., Klöppel G., Alhman H., Caplin M., Couvelard A., de Herder W.W., Erikssson B., Falchetti A., Falconi M., Komminoth P.. TNM staging of foregut (neuro)endocrine tumors: A consensus proposal including a grading system. Virchows Arch. (2006.0) 449 395-401. DOI: 10.1007/s00428-006-0250-1 33. Rindi G., Klöppel G., Couvelard A., Komminoth P., Körner M., Lopes J.M., McNicol A.-M., Nilsson O., Perren A., Scarpa A.. TNM staging of midgut and hindgut (neuro) endocrine tumors: A consensus proposal including a grading system. Virchows Arch. (2007.0) 451 757-762. DOI: 10.1007/s00428-007-0452-1 34. Brierley J.D., Gospodarowicz M.K., Wittekind C.. TNM Classification of Malignant Tumours (2016.0) 35. Jack A., Percy C.L., Sobin L., Whelan S.. International Classification of Diseases for Oncology (ICD-O) (2013.0) 36. Nagtegaal I.D., Odze R.D., Klimstra D., Paradis V., Rugge M., Schirmacher P., Washington K.M., Carneiro F., Cree I.A.. The 2019 WHO classification of tumours of the digestive system. Histopathology (2020.0) 76 182-188. DOI: 10.1111/his.13975 37. National Cancer Intelligence Network: Cancer Survival in England by Stage 38. Berrino F.. The EUROCARE Study: Strengths, limitations and perspectives of population-based, comparative survival studies. Ann. Oncol. (2003.0) 14 v9-v13. DOI: 10.1093/annonc/mdg750 39. Hayati Rezvan P., Lee K.J., Simpson J.A.. The rise of multiple imputation: A review of the reporting and implementation of the method in medical research Data collection, quality, and reporting. BMC Med. Res. Methodol. (2015.0) 15. DOI: 10.1186/s12874-015-0022-1 40. Dasari A., Shen C., Halperin D.M., Zhao B., Zhou S., Xu Y., Shih T., Yao J.C.. Trends in the Incidence, Prevalence, and Survival Outcomes in Patients With Neuroendocrine Tumors in the United States. JAMA Oncol. (2017.0) 3 1335-1342. DOI: 10.1001/jamaoncol.2017.0589 41. Rindi G., Klimstra D.S., Abedi-Ardekani B., Asa S.L., Bosman F.T., Brambilla E., Busam K.J., De Krijger R.R., Dietel M., El-Naggar A.K.. A common classification framework for neuroendocrine neoplasms: An International Agency for Research on Cancer (IARC) and World Health Organization (WHO) expert consensus proposal. Mod. Pathol. (2018.0) 31 1770-1786. DOI: 10.1038/s41379-018-0110-y 42. Han K., Jung I.. Restricted Mean Survival Time for Survival Analysis: A Quick Guide for Clinical Researchers. Korean J. Radiol. (2022.0) 23 495-499. DOI: 10.3348/kjr.2022.0061 43. Discacciati A., Bellavia A., Lee J.J., Mazumdar M., Valeri L.. Med4way: A Stata command to investigate mediating and interactive mechanisms using the four-way effect decomposition. Int. J. Epidemiol. (2018.0) 48 15-20. DOI: 10.1093/ije/dyy236 44. Padmanabhan S., Carty L., Cameron E., Ghosh R.E., Williams R., Strongman H.. Approach to record linkage of primary care data from Clinical Practice Research Datalink to other health-related patient data: Overview and implications. Eur. J. Epidemiol. (2019.0) 34 91-99. DOI: 10.1007/s10654-018-0442-4 45. Luong T.V., Watkins J., Chakrabarty B., Wang L.M.. Standards and Datasets for Reporting Cancers Dataset for Histopathological Reporting of Neuroendocrine Neoplasms of the Gastroenteropancreatic Tract DRAFT. (2019.0) 46. Niu Y., Guo C., Wen S., Tian J., Luo J., Wang K., Tian H., Yeh S., Chang C.. ADT with antiandrogens in prostate cancer induces adverse effect of in-creasing resistance, neuroendocrine differentiation and tumor metastasis. Cancer Lett. (2018.0) 439 47-55. DOI: 10.1016/j.canlet.2018.09.020
--- title: 'Adipose-Derived Stromal Cells within a Gelatin Matrix Acquire Enhanced Regenerative and Angiogenic Properties: A Pre-Clinical Study for Application to Chronic Wounds' authors: - Nicolo Costantino Brembilla - Ali Modarressi - Dominik André-Lévigne - Estelle Brioudes - Florian Lanza - Hubert Vuagnat - Stéphane Durual - Laurine Marger - Wolf-Henning Boehncke - Karl-Heinz Krause - Olivier Preynat-Seauve journal: Biomedicines year: 2023 pmcid: PMC10046849 doi: 10.3390/biomedicines11030987 license: CC BY 4.0 --- # Adipose-Derived Stromal Cells within a Gelatin Matrix Acquire Enhanced Regenerative and Angiogenic Properties: A Pre-Clinical Study for Application to Chronic Wounds ## Abstract This study evaluates the influence of a gelatin sponge on adipose-derived stromal cells (ASC). Transcriptomic data revealed that, compared to ASC in a monolayer, a cross-linked porcine gelatin sponge strongly influences the transcriptome of ASC. Wound healing genes were massively regulated, notably with the inflammatory and angiogenic factors. Proteomics on conditioned media showed that gelatin also acted as a concentrator and reservoir of the regenerative ASC secretome. This secretome promoted fibroblast survival and epithelialization, and significantly increased the migration and tubular assembly of endothelial cells within fibronectin. ASC in gelatin on a chick chorioallantoic membrane were more connected to vessels than an empty sponge, confirming an increased angiogenesis in vivo. No tumor formation was observed in immunodeficient nude mice to which an ASC gelatin sponge was transplanted subcutaneously. Finally, ASC in a gelatin sponge prepared from outbred rats accelerated closure and re-vascularization of ischemic wounds in the footpads of rats. In conclusion, we provide here preclinical evidence that a cross-linked porcine gelatin sponge is an optimal carrier to concentrate and increase the regenerative activity of ASC, notably angiogenic. This formulation of ASC represents an optimal, convenient and clinically compliant option for the delivery of ASC on ischemic wounds. ## 1. Introduction Chronic skin wounds affect about 1–$2\%$ of the worldwide population, and up to $5\%$ of subjects older than 65 years [1,2]. The treatment landscape for the management of chronic wounds spans from dressings, debridement, negative pressure to advanced skin replacement technologies [3,4]. While the efficacy of each of these therapies has been shown in specific settings, a unique gold-standard in chronic wound management is lacking. Even in the presence of guidelines, severe ulcers are not efficiently managed with current therapies. In this respect, cell therapy based on adipose-derived stromal cells (ASC) has emerged as particularly promising [5,6]. Several in vitro and pre-clinical animal studies have demonstrated that ASC exert beneficial effects on wounds [5,7,8,9,10]. ASC promote healing [11]; suppress excessive inflammatory responses [12,13,14]; increase survival and proliferation of fibroblasts allowing ExtraCellular Matrix (ECM) remodeling [15,16]; produce anti-fibrotic factors [17]; and promote neovascularization [18]. The therapeutic effects of ASC were shown to depend mainly on paracrine mechanisms and production of extracellular vesicles [19,20]. Several human clinical trials have also been reported worldwide, most of them of limited reliability (i.e., uncontrolled studies). The appearance in recent years of a few well-conducted controlled studies has provided a new perspective to interpret the efficacy of ASC-based therapies for patients with chronic wounds [5]. Some recent and controlled trials reported that expanded ASC or the stromal vascular fraction have a superior efficacy compared with the control group, with a satisfactory safety profile. The delivery method and ASC stability in vivo remain, however, major challenges that may compromise the large-scale application of ASC therapy. The most common modality used in past and on-going trials relies on multiple intramuscular, intra-wound or peri-wound injections [21,22]. This route of administration is not well controlled spatially and it is still unclear where the ASC should be injected among the possibilities between the dermis, the adipose tissue or the adjacent muscles [23]. This route is also very painful due to frequent local inflammation. Furthermore, in the absence of scaffolds, ASC do not have the opportunity to concentrate their regenerative secretome and be regulated by their environment in a way that promotes wound healing. In addition, suspension-delivered ASC have been shown to be locally unstable due to cell migration or death, or to be rapidly trapped in the lungs before reaching the wound [24,25,26]. Collectively, these limitations due to a lack of knowledge and control of ASC delivery routes still limit standardization and progress toward a clinical reality [8,10]. In this study, we evaluated the introduction of ASC within a clinical-grade surgical sponge composed of crosslinked porcine gelatin. This formulation not only aimed at concentrating locally the ASC regenerative secretome in the wound bed, but took advantage of the pro-healing and adsorption properties of gelatin. The impact of crosslinked gelatin on ASC within this specific environment was notably investigated and showed strong regulations of the healing properties, notably angiogenesis. The ASC delivery studied here represents an easy and convenient formulation to limit the constraints of the current injection-based protocol of ASC therapy for chronic wounds. This study demonstrates that the introduction of ASC into cross-linked porcine gelatin sponges strongly influences their biological activity, in particular by regulating numerous genes involved in the wound healing process. In particular, genes related to angiogenesis were increased. When applied topically to ischemic rat wounds, the enhanced angiogenic properties of ASC by the gelatin sponge were confirmed, facilitating revascularization and wound closure more rapidly than standard treatments. Thus, cross-linked gelatin scaffolds represent a convenient, biocompatible, and effective delivery route for concentrating ASC and increasing their wound healing capacity. ## 2.1. ASC Culture and Engineering of an ASC-Enriched Patch Human ASCs were prepared from the non-ischemic subcutaneous fat of donors. The ASC lines used in this study were fully validated for their phenotype, multipotency and regenerative potential. ASC were used between passage 2–5 and were cultured in Dulbecco’s Modified Eagle Medium DMEM (4.5 g/L glucose, L-Glutamine) supplemented with $10\%$ of human platelet lysate (Stemulate, Cook Regentek, Bloomington, IN, USA), $1\%$ penicillin and streptomycin (ThermoFisher, Waltham, MA, USA) at 37 °C and under $5\%$ CO2. Rat ASC were grown in the same medium supplemented with $10\%$ of fetal calf serum (ThermoFisher, Waltham, MA, USA). To manufacture the ASC-gelatin sponge, a piece of sterile absorbable gelatin sponge USP Gelfoam (Pfizer, Brooklyn, NY, USA) was soaked in a suspension of ASC at a final density of 6000 cells/mm3. ## 2.2. Flow Cytometry and Multipotent Differentiation of ASC Cells were incubated with fluorochrome-labeled antibodies for 30 min at 4 °C in binding buffer (BD Biosciences, Allschwil, Switzerland), prior to analysis using a BD AccuriTM-B6 flow cytometer (BD Biosciences, Allschwil, Switzerland). Antibodies were as follows: (i) for human cells: mouse IgG1 anti-CD44/CD73/CD90/CD45/CD105/CD14/HLA-DR (all from Abcam, Cambridge, UK); (ii) for rat cells: Armenian hamster anti-rat CD29-APC (clone HMb1-1, ThermoFisher, Waltham, MA, USA), mouse anti-rat-CD31-PE (clone TLD-3A12, BD Biosciences, Allschwil, Switzerland), mouse anti-rat CD45-BV421 (clone OX-1, BD Biosciences, Allschwil, Switzerland) and mouse anti-rat CD90-BB515 (clone OX-7, BD Biosciences, Allschwil, Switzerland). Analysis was performed on viable cells (negative for Draq7) upon exclusion of cell doublets. Cell purity was >$98\%$. The multipotent differentiation into adipocytes, osteocytes and chondrocytes was performed by using the Human Mesenchymal Stem Cell Functional Identification Kit (R&D Systems, Minneapolis, MN, USA) according to the supplier’s instructions. ## 2.3. Immunocytochemistry and Immunofluorescence on Tissue Sections ASC were cultured on glass coverslips prior to fixation with paraformaldehyde $0.5\%$ for 15 min at RT. Cells were incubated overnight (o/n) at 4 °C in PBS containing $0.3\%$ Triton X-100 and $0.5\%$ bovine serum-albumin with the following primary antibodies: mouse IgM anti-Stro-1 (Clone STRO1, ThermoFisher, Waltham, MA, USA). Detection was achieved using an anti-mouse IgM-Alexa 555 antibody for one hour at +4 °C. Cells were stained with DAPI 1 μg/mL for 10 min prior to final washing and mounting. For histological analyses, tissues were washed in PBS and fixed with a $4\%$ paraformaldehyde solution for 20 min prior to dehydration and embedment in paraffin. Upon rehydration, slides were stained in PBS supplemented with bovine serum albumin $1\%$, Triton X-100 $0.3\%$ o/n at 4 °C with a mouse IgG anti-CD31 (Abcam, Cambridge, UK). Upon staining with anti-mouse IgG-Alexa 555 antibody, slides were counterstained with DAPI and mounted in FluorSave medium (Calbiochem, Buchs, Switzerland). Hematoxylin and Eosin staining and Masson’s trichrome staining were performed according to the standard protocol. Vessel area was computed by QuPath software as a function of CD31 staining (above a defined threshold) in at least 3 sections per condition analyzed. Immunocytochemistry and immunofluorescence applied to osteocytes, chondrocytes and adipocytes derived from ASC was performed with the reagents of the human mesenchymal stem cell functional identification kit (RnDSystems, Minneapolis, MN, USA). ## 2.4. Cytokine Measurements Cytokines were measured in the supernatants from ASC cultures and ASC-enriched patches using the human cytokine base kit A (RnDSystems, Minneapolis, MN, USA) combined with a magnetic Luminex assay (Bio-plex 200, Biorad, Hercules, CA, USA) according to the manufacturer’s instructions. ## 2.5. Transcriptomic A microarray was used as the best way to simply analyze the global cell regulation within a gelatin sponge environment. Isolation of total RNA was performed by using RNeasy kit from Qiagen (Hombrechtikon, Switzerland) according to the manufacturer’s instructions. RNA concentration was determined by a spectrometer (Thermo Scientific™ NanoDrop 2000, ThermoFisher, Waltham, MA, USA) and RNA quality was verified by 2100 bioanalyzer (Agilent, Santa Clara, CA, USA). Human and rat microarray was performed with the ClariomTM S Assay’s for human and rat (ThermoFisher, Waltham, MA, USA), respectively, using the Complete GeneChip® Instrument System, Affymetrix. Hierarchical clustering and principal component analysis were computed using TAC4.0.1.36 software (Biosystems, Muttenz, Switzerland) with default settings. Gene Set Enrichment Analysis (GSEA) was used to analyze the pattern of differential gene expression between the human ASC-patch and the monolayer condition. The Gene Ontology Biological Process (GOBP) gene set from the Molecular Signatures Database was used. The results of GSEA analysis were visualized for enrichment map using Cytoscape 3.8.2, (Moutain view, CA, USA). The enrichment of processes and pathways within the significantly upregulated or downregulated transcripts (fold change > 2, FDR < 0.01) identified in the rat ASC-path compared to rat ASC grown in monolayer was assessed using Metascape (www.metascape.org, accessed on 11 March 2022). The parameters used for the analysis were as follows: Organisms: Rattus Norvegicus, Input gene set: GO Biological Process; Min Overlap: 3; p value cutoff: 0.01; Min enrichment: 0.01. ## 2.6. Mass Spectrometry Cultured human ASC or ASC-enriched patches were incubated for 45 min with collagenase NB6 (Nordmark, Uetersen, Germany) at 0.3 U/mL, washed with a serum-free DMEM (ThermoFisher, Waltham, MA, USA) and cultured for 24 h at 37 °C in serum-free medium. Upon clarification of supernatants at 500× g for 10 min, proteins were precipitated, digested and peptides analyzed by nanoLC-MSMS using an easynLC1000 (ThermoFisher, Waltham, MA, USA) coupled with a Q Exactive HF mass spectrometer (ThermoFisher, Waltham, MA, USA). Database searches were performed with Mascot (Matrix Science, London, UK) using the Human Reference Proteome database (Uniprot). Data were analyzed and validated with Scaffold (Proteome Software, Portland, OR, USA) with $1\%$ of protein FDR and at least 2 unique peptides per protein with a $0.1\%$ of peptide FDR. ## 2.7. Chick Chorioallantoic Membrane Model To estimate the in vivo angiogenic properties of a gelatin/ASC patch, the CAM model was a first simple and rapid experimental approach. Fertilized chicken eggs were incubated at 37 °C and placed with the smaller convexity pointing upward from ED1 (Embryo Development day) to ED4. At ED4, a hole was drilled through the smaller convexity pointing of the shell. At ED 7, the eggs were opened with scissors through the hole and the inner membrane to create a round window with approximate 1 cm diameter. The developing chorioallantoic membrane was then irritated through creation of a micro-hemorrhage. With ASC in suspension, a silicon ring with a 4 mm inner diameter was placed on the site of the generated hemorrhage. The ASC-enriched patched, fibroblast-enriched patches or control empty patches were deposited directly on the site of the generated hemorrhage. After implantation, the window in the eggshell was covered with a paraffin film and placed in the incubator at 37 °C. The number of vessel connections to the patch were counted manually under a binocular loop. ## 2.8. Migration and Tubulogenesis of HUVEC Migration and tubulogenesis of HUVEC was an efficient way to discover the functional influence of the ASC/gelatin secretome on endothelial cells. HUVEC were purchased and cultured in complete endothelial cell medium 2 (both from Sigma, Buchs, Switzerland). Migration of HUVEC was analyzed by using the endothelial cells migration assay (Sigma, Buchs, Switzerland) according to the manufacturer’s instructions. Briefly, HUVEC were starved for 15 h in the endothelial cell medium 2 without serum and supplement and introduced in a Boyden chamber with a hemi-permeable membrane coated with fibronectin or Bovine Serum Albumin (BSA) used as a control at the bottom. Migration through the fibronectin layer towards supplement-free endothelial cell medium 2 conditioned 48 h by ASC was measured by cell coloration (crystal violet) and extraction of the dye having migrated outside the Boyden chamber (via measurement of the absorbance of the extract at 540 nm). The migration was calculated as the difference between the absorbance with fibronectin and the absorbance with control BSA. For the tubulogenesis assay, serum/supplement-free endothelial cell medium 2 was conditioned for 48 h with ASC. The analysis of tubular assembly of HUVEC was performed in the presence of each conditioned medium by using the angiogenesis assay kit (Abcam, Cambridge, UK) according to the manufacturer’s instructions. Briefly, HUVEC were plated in their conditioned medium on a fibronectin-containing gel for 24 h, prior to cell coloration by a fluorescent dye and analysis of tubes via the Cytation 5 cell imaging reader (Agilent, Santa Clara, CA, USA). ## 2.9. Animal Experiments For stability/tumorigenicity assays, the method recommended by the European Pharmacopoeia (EMEA/$\frac{149995}{2008}$) was used. ASC-enriched patches or Hela cells (5 × 105 cells) were subcutaneously transplanted in the right flank of 10 Nu/Nu mice, followed for 12 weeks for tumor palpation and necropsy. At week five, 9 out of 10 mice that received Hela cells developed palpable tumors, confirming the validity of the test. The model of ischemic wound in the rat was the best available animal model of ischemic wounds and performed as previously described [25,27]. Wistar female rats of 250–300 g were pre-anesthetized by inhalation of isoflurane $5\%$, and anesthetized at the dose of $2\%$. Hairs were removed from the inguinal region using a mechanical shaver. All surgical procedures were performed under an operating microscope. Through a longitudinal incision made in the upper part of the left thigh, the external iliac and femoral arteries were dissected free along their entire length, from the common iliac to the saphenous artery, and one cm-length artery was removed. Immediately after the arterial resection, a wound was created on the dorsal aspect of the feet in all animals by removing a full-thickness skin area of 1.2 × 0.8 cm. Treatments were applied a day after the surgery. Rat ASC were generated from the inguinal non-ischemic fat of control rats two months after the induction of paw ischemia. Treatments were applied a day after the wound creation. To maintain the patches on the wound, a gutter of perforated silicone interface (Mepitel, Molnlycke, Singapore, Singapore) was covered with a thin sheet of polyurethane (Opsite, SmithNephew, London, UK) and sutured around the wound. The patches were removed at day 7 and the wound covered with polyurethane until full recovery. Daily macroscopic evaluation of the limbs and feet as well as wounds planimetry were performed until complete wound healing. ## 2.10. Statistical Analysis Statistical analysis was performed using GraphPad Prism version 6.0 (Graphpad Software, La Jolla, CA, USA). p-values less than 0.05 were considered statistically significant, and were indicated as follows: : $p \leq 0.05$; : $p \leq 0.01$; *: $p \leq 0.001$ (non-parametric Mann–Whitney t test). ## 3.1. Culture of ASC in a Sponge Made of Porcine Crosslinked Gelatin ASC were introduced and cultured in a sponge made of crosslinked porcine gelatin (Figure S1A). It was indeed desirable to study a scaffold whose structure, functions and mechanical properties are similar to those of healthy skin and compatible with healing [28]. A number of scaffolds have been widely used in the field of tissue engineering and all emphasize the importance of hydrophilicity, biodegradability and biocompatibility. Collagen is an important component of the skin and provides strength. Gelatin is a hydrophilic, biocompatible, and inexpensive collagen-derived product and was considered to have the desired characteristics to promote ASC survival, adhesion, and activity. Several ASC lines used in this study were prepared from the adipose tissue of donors. The ASC identity was confirmed by their phenotype (CD44+/CD73+/CD90+/CD45−/CD105+/CD14−/HLA-DR−) and ability to be differentiated towards chondrocytes, osteoblasts and adipocytes under appropriated differentiation conditions and according to international standards [29]. Inter-donor variability, as assessed by computing the coefficient of variation of the whole transcriptome among the different ASC lines, was minimal (mean ± SD of 5.7 ± $4.0\%$). The* generated ASC-gelatin sponge was easy to handle (Figure 1A, left), and characterized histologically by a dense cellular tissue (purple) interspersed between gelatin trabeculae (dark pink) (Figure 1A, middle). A green halo around gelatin trabeculae suggested collagen dissolution as assessed by Masson’s trichrome staining (Figure 1A, right). ASC that clustered in pores organized into a compact tissue composed of collagen fibers from their own secretion. This tissue resulted from an ASC-dependent secretion of ECM and contraction of the gelatin sponge, as only isolated cells in a rarefied gelatin mesh were observed at the beginning of the culture (Figure S1B). A low cellular density was instead obtained upon parallel culture of ASCs within a conventional collagen gel (Figure S1C). ASC retained their stromal identity within the gelatin sponge, as shown by sustained and widespread expression of the stromal marker Stro-1 (Figure 1B). Furthermore, cells obtained upon enzymatic dissociation of the ASC-enriched patch had a transcript profile compatible with undifferentiated ASC as defined by current guidelines [30] (Figure 1C). Confirming the ASC identity, cells extracted from the ASC-enriched patch retained multipotent capacity, being able to differentiate towards osteoblasts (expressing osteocalcin), adipocytes (FAB4) and chondrocytes (aggrecan) in appropriate culture conditions (Figure 1D). Thus, ASC could efficiently be included and grown within a clinical-grade and crosslinked porcine gelatin sponge to generate an undifferentiated ASC-enriched compact tissue, which has physical properties compatible with its use as a cellular patch. ## 3.2. ASC in a Gelatin Sponge Enhanced Their Regenerative Transcriptome We next investigated whether ASC have modified their gene expression capabilities because of their interaction with porcine gelatin in spongious conditions. To this aim, the transcriptome of cells extracted from three independent ASC-gelatin sponges was compared with ASC from the same batch, but grown in parallel in monolayers. This latter condition was performed in line with the standard protocol used to produce ASC for injection-based ASC therapy [8]. Hierarchical clustering and principal-component analysis revealed that ASC strongly modified their global transcriptome when cultured within the gelatin sponge, compared to ASC in monolayers (Figure S2A,B, respectively). ASC grown in monolayers for 7 days maintained a transcriptome profile-like cells prior to culture (freshly isolated ASC vs. monolayer (d7), Figure S2B). ASC within the patch did not differentiate in fibroblasts nor embryoid bodies (Figure S2B). To explore the characteristics of the genes expressed in ASC-patches compared to ASC grown in monolayers, we performed a threshold-free gene set enrichment analysis (GSEA). The 50 most highly differentially expressed genes (top 25 upregulations and top 25 downregulations) are shown in Figure 2A. The most up-regulated genes and pathways were linked to critical components of the healing process, namely immune function, morphogenesis, and vascular growth. Down-regulations were linked to DNA regulation and mitochondrial functions (Figure 2B). Smaller clusters are shown in Figure S2C. Analysis of the 191 genes reported to be most implicated in the wound regeneration process (list available upon request) resulted in the identification of 51 significantly regulated transcripts (fold change > 2, $p \leq 0.005$): 30 strongly up-regulated (4× to 60×), 13 minimally up-modulated (<4×), five minimally down-modulated (<4×) and three modestly down-regulated (4× to 10×) (Figure 2C). The strongest up-regulated transcripts observed were CXCL8 (IL-8) (potent angiogenic, chemotactic and inflammatory cytokine), CXCL6 (angiogenic, chemotactic, anti-microbial), IL-6 (angiogenic, pro-inflammatory), and CXCL5 (angiogenic, matrix remodeling factor, pro-inflammatory). Other angiogenic factors were up-regulated (VEGF, ANGPTL2, ADM), as well as matrix remodeling factors (MMP16), collagens, cell growth factors (HGF) and several chemokines. Keratinocytes Growth factor (KGF or FGF-7), promoting epithelialization during skin wound healing, was similarly upregulated. Together, these observations show that the interaction of ASC with gelatin within resulted in a general enhancement of their regenerative transcriptome. ## 3.3. The Secreted Proteome of ASC Is Absorbed by Gelatin The composition of the proteome secreted from ASC in a gelatin sponge was profiled by mass spectrometry and compared to that of control ASC grown in monolayers, or the empty sponge. Hierarchical clustering analysis allowed for the distinction of the three conditions (Figure 3A). The number of proteins found in the supernatants from ASC-gelatin was less than the sum of the proteins found in the other two conditions (Figure 3B), suggesting protein adsorption within the patch. Accordingly, the protein diversity was slightly higher in supernatants from ASC grown in monolayers if compared to ASC-gelatin (Figure 3A,C). Overall, $3.1\%$ of proteins were down-regulated and $2.7\%$ up-regulated in ASC-patches as compared to the monolayers’ supernatants (The complete list is available upon request). Quantification by bio-arrays of cytokines, growth and angiogenic factors having a key role in wound healing confirmed their relative concentration in the supernatant from ASC-patches (Figure S3). Analysis of the supernatants from empty patches revealed that the gelatin dissolved, releasing 49 proteins on average, most being collagens and keratins (Figure S4). We next assessed the ability of the gelatin of the patch to adsorb proteins secreted by ASC. Three empty gelatin sponges were incubated with a serum-free medium or a serum-free medium previously conditioned by ASC. Mass spectrometry analysis was performed on the mixture after complete dissolution of the sponge. Twelve vs. one hundred and eighty-one proteins were found in sponges exposed to an empty serum-free medium vs. an ASC-conditioned media, confirming that gelatin had adsorbed several factors produced by ASC (Figure 3D,E). Interestingly, fibronectin-1 (FN1), a molecule important for the healing process [31], was the most abundant protein derived from ASC adsorbed on gelatin. The complete list is available upon request. Together, these proteomic data indicate that the ASC-gelatin sponge actively released factors that are derived from both ASC and gelatin. In addition, our results indicate that the patch functioned as a reservoir of healing factors. ## 3.4. The ASC Gelatin Sponge Promotes Angiogenesis Neo-vascularization being a most critical process for wound healing, we next investigated the ability of the ASC-gelatin sponge to promote angiogenesis in vivo by using the chicken chorioallantoic membrane (CAM) model (Figure 4A). ASC in gelatin were superior to all conditions tested, as a significantly higher number of vessel branches sprouting from the dressing were observed (Figure 4A). Notably, new blood vessels could only be visualized in the ASC-gelatin sponge by macroscopic analysis, but not the other conditions. A single cell suspension of ASC derived from standard culture (contained by a silicon ring) did not induce significant new vessel branches. The pro-angiogenic activity of the ASC-gelatin sponge was probably due to synergistic mechanisms between ASC and gelatin, since it was not merely the addition of the effect of its isolated constituents. In complementary in vitro experiments, conditioned media from the ASC-gelatin sponge increased the migration of Human Umbilical Endothelial Vein Cells (HUVEC) through a fibronectin-containing hemi-permeable membrane and better promoted tubular assembly of HUVEC as compared to media from the same number of ASC alone or empty sponge (Figure 4B). None of the condition tested interfered with HUVEC proliferation. Finally, healthy human keratinocytes seeded on top of the ASC-patch developed a fully stratified epidermis with an intact basement membrane (Figure S5A), and healthy human fibroblasts exposed to ASC-patch conditioned media showed an increased in vitro survival compared to empty sponge-conditioned media (Figure S5B). Together, we show that the ASC in a gelatin sponge harbor more angiogenic properties than ASC alone, promoting tubular assembly and endothelial cell migration, and does not inhibit the growth of key cutaneous cellular components. ## 3.5. ASC in Gelatin Sponge Stability and Tumorigenicity In Vivo Since ASC in a gelatin sponge acquired a pro-angiogenic profile, we tested if they could promote tumorigenesis in immunodeficient nude mice. None of the 10 transplanted mice developed signs of palpable tumors near the transplant or in peripheral tissues, whereas 10 mice transplanted with Hela control cells developed tumors. Notably, all transplanted ASC-gelatin sponges showed macroscopic vascularization at week 5 (Figure 5A, left). Histological assessment of the transplanted ASC-patches revealed the presence of a tissue-like cell-rich gelatin mesh (Figure 5B). Culture of cells extracted upon enzymatic digestion of the transplanted ASC sponges had the morphology of ASC and expressed the Human Nuclear Antigen (HNA) (Figure 5C). Flow cytometric analysis of the secondary ASC line confirmed the maintenance of the ASC phenotype (positivity for CD73, CD90, CD105, and CD44 and absence of CD45, HLA-DR and CD14) (Figure 5D). Finally, the size of the transplanted ASC-patches was stable in vivo until week 3, and was completely resorbed by week 12 (Figure 5A, right). Together, these in vivo experiments indicated that ASC in gelatin lacked any tumorigenic activity and allowed the stabilization of the ASC for at least 5 weeks, while the whole product was resorbed within 12 weeks in immunodeficient setting. The ASC-patch thus had essential pre-clinical safety requirements for further development in humans. ## 3.6. ASC in Gelatin Sponge Accelerates the Healing and Induces Early Neo-Angiogenesis in a Rat Model of Ischemic Wound Next, the in vivo efficacy of ASC in a gelatin sponge in a pre-clinical rat model of an ischemic wound was investigated [25]. Wounds were created on the dorsal part of the hind paws of Wistar rats by removing a full-thickness skin area. To create paw ischemia, 1 cm of the femoral artery was surgically removed prior to the wound creation. In these experiments, ASC-patches were prepared from non-consanguineous rat ASC (rASC-patch). The identity of rat ASC was confirmed by flow cytometry. Rat ASC grown within the patch generated a dense tissue-like structure (Figure S6A). Transcriptomics revealed the presence of differences in the rat ASC-patch as compared to rat ASC monolayer cultures (Figure S5B). Metascape-based enrichment analysis using the GO-Biological Process gene set showed that the most enriched pathways and processes identified within the significantly up-regulated genes (fold change > 2, FDR < 0.05) included extracellular matrix deposition, response to growth factors and angiogenesis. Downregulations were instead linked to DNA regulation and cell cycle (Figure S5C). These results are in line with the observations in humans, considering species-specific differences, and indicate that rat ASC enhance their regenerative transcriptome when grown within the patch. Of note, rats treated with rat ASC-patches healed faster than rats treated with empty gelatin patches or standard silicone/polyurethane dressings (Figure 6A,B). A granulation tissue was macroscopically visible from day 9 in the ASC-patch treated group, while tendons were still exposed in the control groups (Figure 6A). All rats treated with the ASC-enriched patches reached a complete wound closure before day 17, compared to $67\%$ of rats treated with the empty gelatin sponge (Figure 6B). The complete wound closure was confirmed histologically (Figure 6C). In line with the observed effects on angiogenesis, staining for the endothelial marker CD31 revealed an increased vascularization in the healing tissue of rats treated with ASC-patches (Figure 6D). In this condition, angiogenesis was faster and sustained over time, leading to vessels characterized by a higher diameter and organized a denser network. Quantification of the total vessel area confirmed the superiority of the ASC-patch over the empty gelatin sponge and standard dressing (Figure 6D). Finally, we analyzed the ASC survival within the ASC-patch in vivo. Rat-ASC stably transduced with firefly luciferase (FLuc) under the control of the ubiquitous promoter EFS (rASCEFS FLuc) were used to generate rat ASC-patches (FLuc-rASC-Patch). Intraperitoneal injection of D-luciferin allowed the monitoring of ASC survival in vivo by using the live imaging system IVIS Spectrum (Perkin Elmer, Waltham, MA, USA). Luminescence in the FLuc-ASC-Patch was detected until removal of the treatment at day 7, confirming the survival of ASC in vivo (Figure 6E,F). Luminescence was maintained after treatment removal until day 17, indicating that some ASC were engrafted within the healing wound. Additional quantifications during the entire period of treatment indicated a fast increase of the luminescence until day 6, an observation in favor of rapid exchanges of the patch with the biological fluids and the increased neovascularization previously observed (Figure 6F). No ectopic tissue formation was observed outside the wound area (Figure S7). Together, we provide evidence that the ASC within the gelatin sponge were stable and viable in vivo for at least 17 days. Notably, the ASC-patch accelerates wound healing in a rat model of ischemic wounds, resulting in increased neo-vascularization and emergence of a granulation tissue earlier in the healing process. ## 4. Discussion In this study, we demonstrate that ASC introduced within a porcine crosslinked-gelatin sponge not only concentrated their secretome, but also regulated their regenerative activity in favor of regeneration and angiogenesis. The pre-clinical efficacy/safety of a cellular patch composed of ASC in gelatin sponge is demonstrated in a preclinical model of ischemic wound healing. The adipose tissue was preferred to other sources for the preparation of mesenchymal stromal cells (MSC) due to its easier accessibility, which guarantees a higher translational relevance in possible clinical applications. Extensive molecular and biochemical characterization revealed that induction of angiogenesis, in the absence of tumorigenesis, is one of the most important mechanisms of action of this approach. The patch formulation represents an optimal non-invasive delivery route that maximizes the local effects of ASC within the wound. Compared to intra- or peri-wound injection of ASC, the ASC-gelatin sponge guarantees an extended local stability and viability of ASC and the absence of extra-wound migration. This study validates the patch approach in the pre-clinical setting and paves the way for its use in first-in-human studies. The effect of the ASC-patch mainly relies on the combined interaction of ASC with gelatin. ASC are known for their ability to enhance the healing process by secretion of soluble healing factors [32,33,34,35]. Here, we demonstrate that ASC within the patch not only increase their own regenerative potential, but also locally concentrate pro-healing factors. ASC exposed to gelatin microcryogels have been shown to increase their expression of VEGF, HGF, FGF, and PDGF [36]. The formation of a three-dimensional tissue organization within the patch enables the optimization of cell–cell interactions, paracrine events and gas exchange/oxygen supply [37,38]. The gelatin sponge itself may promote the healing process independently of ASC. Gelatin acts as a chemotactic agent [39], promotes the formation of a granulation tissue [40] and absorbs exudates present in the wound bed [41]. In line with this data, gelatin was shown to directly enhance wound closure [39], particularly by increasing angiogenesis, keratinocyte and fibroblast proliferation/migration and myofibroblast differentiation [42,43,44]. Interestingly, the local administration mesenchymal stem cells in a collagen scaffold led to better regeneration of soft gingival tissues in rabbits through enhanced gingival vascularization and epithelization with a clear positive correlation between vascular growth and epithelial response [45]. Additionally, epithelialization and angiogenesis were linked in another study where diabetic wounds received mesenchymal stem cells activated with LPS: the granulation tissue of treated wounds had higher pronounced epithelialization and associated vascularization compared with controls [46]. Besides the direct action of gelatin, we showed that the gelatin patch dissolves in aqueous conditions, releasing soluble collagens and cytokeratins known to favor wound healing through enhancement of proliferation/migration of fibroblasts and keratinocytes [47,48,49,50,51,52,53]. Previous studies have shown the possibility to include MSCs in acellular scaffolds in vitro [54]. A dermic substitute of collagen or atelocollagen loaded with MSC, was also tested in animals [55] or patients [56]. The efficacy was confirmed in an animal model considered to be relevant for clinical translatability, although not fully recapitulating all feature of the human disease. Of note, the in vivo experiments described in this study were performed in an allogeneic setting, as outbred non-consanguineous rats were used. Despite a head-to-head experiment is lacking, the ASC-patch approach promoted a faster wound closure when compared to ASC locally injected in peri-wound area, as assessed in our previous independent experiments performed using the same rat model [25]. The advantages of using a crosslinked gelatin as a scaffold for ASC delivery are the following: it is a regulatory approved, clinical-grade support to ensure an easier and faster clinical translation. Gelatin is indeed a biodegradable and biocompatible scaffold that is already widely used in clinical settings without any antigenicity/toxicity. Compared with collagens or atelocollagens, gelatin is cheaper than collagen, an essential feature for pharmaceutical development. It is also more hydrophilic, an important property ensuring the maintenance of a moist environment on the wound. The porosity of the sponge is particularly attractive to allow ASC integration, survival, simultaneously permitting the biological fluids to circulate, as well as the colonization by host cells and vessels. In line with this, we have observed a rapid irrigation of the sponge when applied on ischemic wounds and a colonization of the patch by host cells. The malleability and physical stability of gelatin sponges allows their spatial adaptation to the wound bed, an important prerequisite for clinical use. Finally, the adsorption and local concentration of the ASC secretome are beneficial to accelerate healing compared to cell suspensions. Thus, the gelatin-based delivery studied here represents an attractive, non-invasive route, which maximizes the local efficacy of ASC therapy for chronic wound care without toxicity/tumorigenicity. The risk of side effects of ASC/gelatin therapy are considered to be limited because (i) clinical-grade gelatin is widely used in surgery worldwide (as hemostatic sponges), and (ii) ASC have been extensively studied and administrated through many delivery routes, showing an excellent tolerance in humans [5]. First-in-human studies are planned soon to clinically validate the use of the ASC-patch solution as a simple and effective treatment improving patient compliance and physician acceptance. The possibility to use the ASC-patch in an allogenic setting in clinical practice is an option that still requires further validation. Indeed, an allogeneic approach would have several advantages over an autologous approach, including the use of batches of certified and validated ASC, reduced logistical constraints, greater scalability, and substantial cost reductions. ## 5. Conclusions The scientific novelty of this study is the demonstration that a gelatin-based matrix modifies ASC in favor of the secretion of factors increasing wound healing and vascularization. This regulation is complemented by the ability of the matrix to locally concentrate and adsorb cell secretions for optimal delivery to the wound bed. This comprehensive preclinical study demonstrates the safety and efficacy of a gelatin-based stem cell patch and is a prerequisite for a future clinical study in humans. ## 6. Patents NB, OPS, KHK and WHB are the inventors of patent application PCT/EP$\frac{2020}{076083}$ covering the culture of ASC within the gelatin support. ## References 1. Guest J.F., Vowden K., Vowden P.. The health economic burden that acute and chronic wounds impose on an average clinical commissioning group/health board in the UK. J. Wound Care (2017) 26 292-303. DOI: 10.12968/jowc.2017.26.6.292 2. Sen C.K., Gordillo G.M., Roy S., Kirsner R., Lambert L., Hunt T.K., Gottrup F., Gurtner G.C., Longaker M.T.. Human skin wounds: A major and snowballing threat to public health and the economy. Wound Repair Regen. (2009) 17 763-771. DOI: 10.1111/j.1524-475X.2009.00543.x 3. Han G., Ceilley R.. Chronic Wound Healing: A Review of Current Management and Treatments. Adv. Ther. (2017) 34 599-610. DOI: 10.1007/s12325-017-0478-y 4. Piaggesi A., Lauchli S., Bassetto F., Biedermann T., Marques A., Najafi B., Palla I., Scarpa C., Seimetz D., Triulzi I.. Advanced therapies in wound management: Cell and tissue based therapies, physical and bio-physical therapies smart and IT based technologies. J. Wound Care (2018) 27 S1-S137. DOI: 10.12968/jowc.2018.27.Sup6a.S1 5. Brembilla N.C., Vuagnat H., Boehncke W.H., Krause K.H., Preynat-Seauve O.. Adipose-Derived Stromal Cells for Chronic Wounds: Scientific Evidence and Roadmap Toward Clinical Practice. Stem Cells Transl. Med. (2023) 12 17-25. DOI: 10.1093/stcltm/szac081 6. DesJardins-Park H.E., Gurtner G.C., Wan D.C., Longaker M.T.. From Chronic Wounds to Scarring: The Growing Health Care Burden of Under- and Over-Healing Wounds. Adv. Wound Care (2022) 11 496-510. DOI: 10.1089/wound.2021.0039 7. Czerwiec K., Zawrzykraj M., Deptula M., Skoniecka A., Tyminska A., Zielinski J., Kosinski A., Pikula M.. Adipose-Derived Mesenchymal Stromal Cells in Basic Research and Clinical Applications. Int. J. Mol. Sci. (2023) 24. DOI: 10.3390/ijms24043888 8. Gadelkarim M., Abushouk A.I., Ghanem E., Hamaad A.M., Saad A.M., Abdel-Daim M.M.. Adipose-derived stem cells: Effectiveness and advances in delivery in diabetic wound healing. Biomed. Pharmacother. (2018) 107 625-633. DOI: 10.1016/j.biopha.2018.08.013 9. Henriksen J.L., Sorensen N.B., Fink T., Zachar V., Porsborg S.R.. Systematic Review of Stem-Cell-Based Therapy of Burn Wounds: Lessons Learned from Animal and Clinical Studies. Cells (2020) 9. DOI: 10.3390/cells9122545 10. Mamsen F.P., Munthe-Fog L., Kring M.K.M., Duscher D., Taudorf M., Katz A.J., Kolle S.T.. Differences of embedding adipose-derived stromal cells in natural and synthetic scaffolds for dermal and subcutaneous delivery. Stem Cell Res. Ther. (2021) 12 68. DOI: 10.1186/s13287-020-02132-5 11. Silachev D.N., Goryunov K.V., Shpilyuk M.A., Beznoschenko O.S., Morozova N.Y., Kraevaya E.E., Popkov V.A., Pevzner I.B., Zorova L.D., Evtushenko E.A.. Effect of MSCs and MSC-Derived Extracellular Vesicles on Human Blood Coagulation. Cells (2019) 8. DOI: 10.3390/cells8030258 12. Chen L., Tredget E.E., Wu P.Y., Wu Y.. Paracrine factors of mesenchymal stem cells recruit macrophages and endothelial lineage cells and enhance wound healing. PLoS ONE (2008) 3. DOI: 10.1371/journal.pone.0001886 13. Stout R.D.. Editorial: Macrophage functional phenotypes: No alternatives in dermal wound healing?. J. Leukoc. Biol. (2010) 87 19-21. DOI: 10.1189/jlb.0509311 14. Zhang Q.Z., Su W.R., Shi S.H., Wilder-Smith P., Xiang A.P., Wong A., Nguyen A.L., Kwon C.W., Le A.D.. Human gingiva-derived mesenchymal stem cells elicit polarization of m2 macrophages and enhance cutaneous wound healing. Stem Cells (2010) 28 1856-1868. DOI: 10.1002/stem.503 15. Smith A.N., Willis E., Chan V.T., Muffley L.A., Isik F.F., Gibran N.S., Hocking A.M.. Mesenchymal stem cells induce dermal fibroblast responses to injury. Exp. Cell Res. (2010) 316 48-54. DOI: 10.1016/j.yexcr.2009.08.001 16. Yates C.C., Rodrigues M., Nuschke A., Johnson Z.I., Whaley D., Stolz D., Newsome J., Wells A.. Multipotent stromal cells/mesenchymal stem cells and fibroblasts combine to minimize skin hypertrophic scarring. Stem Cell Res. Ther. (2017) 8 193. DOI: 10.1186/s13287-017-0644-9 17. Zhu M., Hua T., Ouyang T., Qian H., Yu B.. Applications of Mesenchymal Stem Cells in Liver Fibrosis: Novel Strategies, Mechanisms, and Clinical Practice. Stem Cells Int. (2021) 2021 6546780. DOI: 10.1155/2021/6546780 18. Maacha S., Sidahmed H., Jacob S., Gentilcore G., Calzone R., Grivel J.C., Cugno C.. Paracrine Mechanisms of Mesenchymal Stromal Cells in Angiogenesis. Stem Cells Int. (2020) 2020 4356359. DOI: 10.1155/2020/4356359 19. Guillamat-Prats R.. The Role of MSC in Wound Healing, Scarring and Regeneration. Cells (2021) 10. DOI: 10.3390/cells10071729 20. Yu B., Zhang X., Li X.. Exosomes derived from mesenchymal stem cells. Int. J. Mol. Sci. (2014) 15 4142-4157. DOI: 10.3390/ijms15034142 21. Holm J.S., Toyserkani N.M., Sorensen J.A.. Adipose-derived stem cells for treatment of chronic ulcers: Current status. Stem Cell Res. Ther. (2018) 9 142. DOI: 10.1186/s13287-018-0887-0 22. Huang Y.Z., Gou M., Da L.C., Zhang W.Q., Xie H.Q.. Mesenchymal Stem Cells for Chronic Wound Healing: Current Status of Preclinical and Clinical Studies. Tissue Eng. Part B Rev. (2020) 26 555-570. DOI: 10.1089/ten.teb.2019.0351 23. Krasilnikova O.A., Baranovskii D.S., Lyundup A.V., Shegay P.V., Kaprin A.D., Klabukov I.D.. Stem and Somatic Cell Monotherapy for the Treatment of Diabetic Foot Ulcers: Review of Clinical Studies and Mechanisms of Action. Stem Cell Rev. Rep. (2022) 18 1974-1985. DOI: 10.1007/s12015-022-10379-z 24. Rustad K.C., Gurtner G.C.. Mesenchymal Stem Cells Home to Sites of Injury and Inflammation. Adv. Wound Care (2012) 1 147-152. DOI: 10.1089/wound.2011.0314 25. Kallmeyer K., Andre-Levigne D., Baquie M., Krause K.H., Pepper M.S., Pittet-Cuenod B., Modarressi A.. Fate of systemically and locally administered adipose-derived mesenchymal stromal cells and their effect on wound healing. Stem Cells Transl. Med. (2020) 9 131-144. DOI: 10.1002/sctm.19-0091 26. Oh E.J., Lee H.W., Kalimuthu S., Kim T.J., Kim H.M., Baek S.H., Zhu L., Oh J.M., Son S.H., Chung H.Y.. In vivo migration of mesenchymal stem cells to burn injury sites and their therapeutic effects in a living mouse model. J. Control. Release (2018) 279 79-88. DOI: 10.1016/j.jconrel.2018.04.020 27. Tobalem M., Levigne D., Modarressi A., Atashi F., Villard F., Hinz B., Pittet-Cuenod B.. Hyperglycemia Interacts with Ischemia in a Synergistic Way on Wound Repair and Myofibroblast Differentiation. Plast. Reconstr. Surg. Glob. Open (2015) 3 e471. DOI: 10.1097/GOX.0000000000000443 28. Shan Y., Li C., Wu Y., Li Q., Liao J.. Hybrid cellulose nanocrystal/alginate/gelatin scaffold with improved mechanical properties and guided wound healing. RSC Adv. (2019) 9 22966-22979. DOI: 10.1039/C9RA04026A 29. Bourin P., Bunnell B.A., Casteilla L., Dominici M., Katz A.J., March K.L., Redl H., Rubin J.P., Yoshimura K., Gimble J.M.. Stromal cells from the adipose tissue-derived stromal vascular fraction and culture expanded adipose tissue-derived stromal/stem cells: A joint statement of the International Federation for Adipose Therapeutics and Science (IFATS) and the International Society for Cellular Therapy (ISCT). Cytotherapy (2013) 15 641-648. DOI: 10.1016/j.jcyt.2013.02.006 30. Dominici M., Le Blanc K., Mueller I., Slaper-Cortenbach I., Marini F., Krause D., Deans R., Keating A., Prockop D., Horwitz E.. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy (2006) 8 315-317. DOI: 10.1080/14653240600855905 31. Patten J., Wang K.. Fibronectin in development and wound healing. Adv. Drug Deliv. Rev. (2021) 170 353-368. DOI: 10.1016/j.addr.2020.09.005 32. Ajit A., Ambika Gopalankutty I.. Adipose-derived stem cell secretome as a cell-free product for cutaneous wound healing. 3 Biotech (2021) 11 413. DOI: 10.1007/s13205-021-02958-7 33. Gentile P., Garcovich S.. Systematic Review: Adipose-Derived Mesenchymal Stem Cells, Platelet-Rich Plasma and Biomaterials as New Regenerative Strategies in Chronic Skin Wounds and Soft Tissue Defects. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22041538 34. Lombardi F., Palumbo P., Augello F.R., Cifone M.G., Cinque B., Giuliani M.. Secretome of Adipose Tissue-Derived Stem Cells (ASCs) as a Novel Trend in Chronic Non-Healing Wounds: An Overview of Experimental In Vitro and In Vivo Studies and Methodological Variables. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20153721 35. Zou M.L., Liu S.Y., Sun Z.L., Wu J.J., Yuan Z.D., Teng Y.Y., Feng Y., Yuan F.L.. Insights into the role of adipose-derived stem cells: Wound healing and clinical regenerative potential. J. Cell. Physiol. (2021) 236 2290-2297. DOI: 10.1002/jcp.30019 36. Zeng Y., Zhu L., Han Q., Liu W., Mao X., Li Y., Yu N., Feng S., Fu Q., Wang X.. Preformed gelatin microcryogels as injectable cell carriers for enhanced skin wound healing. Acta Biomater. (2015) 25 291-303. DOI: 10.1016/j.actbio.2015.07.042 37. Petrenko Y., Sykova E., Kubinova S.. The therapeutic potential of three-dimensional multipotent mesenchymal stromal cell spheroids. Stem Cell Res. Ther. (2017) 8 94. DOI: 10.1186/s13287-017-0558-6 38. Mytsyk M., Cerino G., Reid G., Sole L.G., Eckstein F.S., Santer D., Marsano A.. Long-Term Severe In Vitro Hypoxia Exposure Enhances the Vascularization Potential of Human Adipose Tissue-Derived Stromal Vascular Fraction Cell Engineered Tissues. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22157920 39. Naomi R., Bahari H., Ridzuan P.M., Othman F.. Natural-Based Biomaterial for Skin Wound Healing (Gelatin vs. Collagen): Expert Review. Polymers (2021) 13. DOI: 10.3390/polym13142319 40. Gonzalez A.C., Costa T.F., Andrade Z.A., Medrado A.R.. Wound healing—A literature review. An. Bras. Dermatol. (2016) 91 614-620. DOI: 10.1590/abd1806-4841.20164741 41. Gaspar-Pintiliescu A., Stanciuc A.M., Craciunescu O.. Natural composite dressings based on collagen, gelatin and plant bioactive compounds for wound healing: A review. Int. J. Biol. Macromol. (2019) 138 854-865. DOI: 10.1016/j.ijbiomac.2019.07.155 42. Garcia-Orue I., Santos-Vizcaino E., Etxabide A., Uranga J., Bayat A., Guerrero P., Igartua M., de la Caba K., Hernandez R.M.. Development of Bioinspired Gelatin and Gelatin/Chitosan Bilayer Hydrofilms for Wound Healing. Pharmaceutics (2019) 11. DOI: 10.3390/pharmaceutics11070314 43. Nikpasand A., Parvizi M.R.. Evaluation of the Effect of Titatnium Dioxide Nanoparticles/Gelatin Composite on Infected Skin Wound Healing; An Animal Model Study. Bull. Emerg. Trauma (2019) 7 366-372. DOI: 10.29252/beat-070405 44. Xue M., Jackson C.J.. Extracellular Matrix Reorganization During Wound Healing and Its Impact on Abnormal Scarring. Adv. Wound Care (2015) 4 119-136. DOI: 10.1089/wound.2013.0485 45. Kulakov A., Kogan E., Brailovskaya T., Vedyaeva A., Zharkov N., Krasilnikova O., Krasheninnikov M., Baranovskii D., Rasulov T., Klabukov I.. Mesenchymal Stromal Cells Enhance Vascularization and Epithelialization within 7 Days after Gingival Augmentation with Collagen Matrices in Rabbits. Dent. J. (2021) 9. DOI: 10.3390/dj9090101 46. Wang K., Chen Z., Jin L., Zhao L., Meng L., Kong F., He C., Kong F., Zheng L., Liang F.. LPS-pretreatment adipose-derived mesenchymal stromal cells promote wound healing in diabetic rats by improving angiogenesis. Injury (2022) 53 3920-3929. DOI: 10.1016/j.injury.2022.09.041 47. Chen J., Gao K., Liu S., Wang S., Elango J., Bao B., Dong J., Liu N., Wu W.. Fish Collagen Surgical Compress Repairing Characteristics on Wound Healing Process In Vivo. Mar. Drugs (2019) 17. DOI: 10.3390/md17010033 48. Dang Q.F., Liu H., Yan J.Q., Liu C.S., Liu Y., Li J., Li J.J.. Characterization of collagen from haddock skin and wound healing properties of its hydrolysates. Biomed. Mater. (2015) 10 015022. DOI: 10.1088/1748-6041/10/1/015022 49. El Masry M.S., Chaffee S., Das Ghatak P., Mathew-Steiner S.S., Das A., Higuita-Castro N., Roy S., Anani R.A., Sen C.K.. Stabilized collagen matrix dressing improves wound macrophage function and epithelialization. FASEB J. (2019) 33 2144-2155. DOI: 10.1096/fj.201800352R 50. Jridi M., Bardaa S., Moalla D., Rebaii T., Souissi N., Sahnoun Z., Nasri M.. Microstructure, rheological and wound healing properties of collagen-based gel from cuttlefish skin. Int. J. Biol. Macromol. (2015) 77 369-374. DOI: 10.1016/j.ijbiomac.2015.03.020 51. Pal P., Srivas P.K., Dadhich P., Das B., Maity P.P., Moulik D., Dhara S.. Accelerating full thickness wound healing using collagen sponge of mrigal fish (. Int. J. Biol. Macromol. (2016) 93 1507-1518. DOI: 10.1016/j.ijbiomac.2016.04.032 52. Wiegand C., Buhren B.A., Bunemann E., Schrumpf H., Homey B., Frykberg R.G., Lurie F., Gerber P.A.. A novel native collagen dressing with advantageous properties to promote physiological wound healing. J. Wound Care (2016) 25 713-720. DOI: 10.12968/jowc.2016.25.12.713 53. Feroz S., Muhammad N., Ranayake J., Dias G.. Keratin—Based materials for biomedical applications. Bioact. Mater. (2020) 5 496-509. DOI: 10.1016/j.bioactmat.2020.04.007 54. Ye D., Wu S., Zhang B., Hong C., Yang L.. Characteristics and clinical potential of a cellularly modified gelatin sponge. J. Appl. Biomater. Funct. Mater. (2021) 19 22808000211035061. DOI: 10.1177/22808000211035061 55. Markowicz M., Koellensperger E., Neuss S., Koenigschulte S., Bindler C., Pallua N.. Human bone marrow mesenchymal stem cells seeded on modified collagen improved dermal regeneration in vivo. Cell Transplant. (2006) 15 723-732. DOI: 10.3727/000000006783464408 56. Yoshikawa T., Mitsuno H., Nonaka I., Sen Y., Kawanishi K., Inada Y., Takakura Y., Okuchi K., Nonomura A.. Wound therapy by marrow mesenchymal cell transplantation. Plast. Reconstr. Surg. (2008) 121 860-877. DOI: 10.1097/01.prs.0000299922.96006.24 57. Zhou Z.Q., Chen Y., Chai M., Tao R., Lei Y.H., Jia Y.Q., Shu J., Ren J., Li G., Wei W.X.. Adipose extracellular matrix promotes skin wound healing by inducing the differentiation of adiposederived stem cells into fibroblasts. Int. J. Mol. Med. (2019) 43 890-900. DOI: 10.3892/ijmm.2018.4006 58. Borowczyk J., Buerger C., Tadjrischi N., Drukala J., Wolnicki M., Wnuk D., Modarressi A., Boehncke W.H., Brembilla N.C.. IL-17E (IL-25) and IL-17A Differentially Affect the Functions of Human Keratinocytes. J. Investig. Dermatol. (2020) 140 1379-1389.e2. DOI: 10.1016/j.jid.2019.12.013
--- title: Developing a Fluorescent Inducible System for Free Fucose Quantification in Escherichia coli authors: - Samantha Nuñez - Maria Barra - Daniel Garrido journal: Biosensors year: 2023 pmcid: PMC10046853 doi: 10.3390/bios13030388 license: CC BY 4.0 --- # Developing a Fluorescent Inducible System for Free Fucose Quantification in Escherichia coli ## Abstract L-Fucose is a monosaccharide abundant in mammalian glycoconjugates. In humans, fucose can be found in human milk oligosaccharides (HMOs), mucins, and glycoproteins in the intestinal epithelium. The bacterial consumption of fucose and fucosylated HMOs is critical in the gut microbiome assembly of infants, dominated by Bifidobacterium. Fucose metabolism is important for the production of short-chain fatty acids and is involved in cross-feeding microbial interactions. Methods for assessing fucose concentrations in complex media are lacking. Here we designed and developed a molecular quantification method of free fucose using fluorescent Escherichia coli. For this, low- and high-copy plasmids were evaluated with and without the transcription factor fucR and its respective fucose-inducible promoter controlling the reporter gene sfGFP. E. coli BL21 transformed with a high copy plasmid containing pFuc and fucR displayed a high resolution across increasing fucose concentrations and high fluorescence/OD values after 18 h. The molecular circuit was specific against other monosaccharides and showed a linear response in the 0–45 mM range. Adjusting data to the Hill equation suggested non-cooperative, simple regulation of FucR to its promoter. Finally, the biosensor was tested on different concentrations of free fucose and the supernatant of Bifidobacterium bifidum JCM 1254 supplemented with 2-fucosyl lactose, indicating the applicability of the method in detecting free fucose. In conclusion, a bacterial biosensor of fucose was validated with good sensitivity and precision. A biological method for quantifying fucose could be useful for nutraceutical and microbiological applications, as well as molecular diagnostics. ## 1. Introduction L-Fucose is an important monosaccharide that exerts functional roles in multiple biological processes [1]. L-Fucose is a deoxy hexose sugar characterized by missing a hydroxyl group at C-6 [1]. It is commonly present in mammalian mucins, human milk oligosaccharides (HMOs), and glycoconjugates of the intestinal epithelium [2,3], such as glycolipids, N-glycans, and O-glycans. In these glycoconjugates, fucose is usually found at terminal positions in α- linkages (such as α1-2, α1-3, α1-4, and α1-6; [4]). The main enzyme responsible for these fucosylations is α-1,2-fucosyltransferase (Fut2), which is expressed in epithelial cells and links fucose to the terminal β-D-galactose of mucosal glycans [5]. Fucose is abundant in the gastrointestinal tract (GT) and influences the complex microbial ecosystem that inhabits there. Several microorganisms are equipped with α-fucosidases targeting all existing fucose linkages [6,7,8]. Therefore, gut microbes can release fucose from dietary glycans, which is used as a microbial carbon source [9]. In addition, fucose may promote the growth of beneficial bacteria in the gut, such as Bifidobacterium and Bacteroides [10]. Fucose is finally a common exchange molecule involved in multiple microbial cross-feeding interactions [11,12,13]. In addition to serving as an energy source for some microbes, fucose is involved in diverse metabolic pathways, including the regulation of quorum sensing and suppression of virulence genes in pathogens [14,15]. Detecting low levels of free fucose in biological samples could be a valuable indicator of infection or inflammation [16]. Some pathogens can use fucose as a signaling molecule regulating pathogenesis [17]. Fucose and fucose-containing oligosaccharides usually act as a decoy, preventing the binding of some viral and bacterial pathogens [18]. High levels of fucose in urine have been associated with cirrhosis and certain types of cancer [19,20]. Finally, loss of function mutations of fucosyl-transferase Fut2 have been associated with Crohn’s disease [21]. Free fucose is usually quantified by HPLC [22,23,24,25] and enzymatic assays with L-fucose dehydrogenase [19,26]. Recently, lectin-based microfluidic detection assays have been developed [27,28], as well as fluorescence-based assays with probes and electrochemical sensors [20,29]. Shin et al. [ 30] developed a molecular biosensor for quantifying 2-fucosyllactose (2FL) in breast milk samples. Their circuit contained a constitutive α-fucosidase expressed in an E. coli strain mutant for lactose consumption. Therefore, the detection of 2FL and emission of fluorescence were coupled to cell growth and 2FL degradation [30]. Bacterial biosensors are genetically modified organisms that detect an input, usually a substance or the changes in the concentrations of a specific molecule, which are sensed and internally translated into a genetic output that emits a quantifiable signal [31,32]. Bacterial biosensors are usually constructed of transcription factors and their corresponding promoters. The most common outputs are fluorescent proteins such as Green Fluorescent Protein (GFP). An excellent candidate to develop a bacterial biosensor is Escherichia coli, a model widely used in biotechnological research and development since its genome and metabolic pathways are fully known [30,32,33] Escherichia coli K12 can use multiple sugars as a carbon source for its growth, including fucose [34,35]. Fucose can induce the expression of genes allowing its transport and metabolism, a genetic system known as the fucose regulon. It consists of six genes: L-fucose permease (fucP), L-fucose isomerase (fucI), L-fuculose kinase (fucK), L-fuculose phosphate aldolase (fucA), L-1,2-propanediol oxidoreductase (fucO) and the transcription regulator of the regulon (fucR) [36,37]. These genes are clustered into three operons, fucPIK, fucA (which is transcribed in a clockwise direction), and fucO (which is transcribed counterclockwise). FucR is an activator [37], which is induced by fuculose 1-phosphate, an intermediate molecule from fucose metabolism. FucR also shows positive autoregulation [38]. In this study, we used the fuc molecular system for developing a method of quantifying free L-fucose, using FucR and the fuc promoter triggering the induction of sfGFP in E. coli. We first compared the detection of fucose in high- and low-copy plasmids with or without fucR. The best system was evaluated for specificity, and calibration curves were obtained at low (0–3 mM) and high concentrations (0–50 mM) with good resolution. The biosensor was successfully applied to quantify fucose in bacterial supernatants. This molecular biosensor could be further studied to quantify free fucose in complex biological samples with good resolution and specificity. ## 2. Materials and Methods Mediums, reagents, and sugars. Miller Luria-Bertani (LB) liquid and agar medium was obtained from Merck (Boston, MA, USA) and autoclaved at 121 °C for 15 min. Minimum medium MM9 was prepared with KH2PO4 ($15\%$ w/v), NaCl ($2.5\%$ w/v), Na2HPO4 ($33.9\%$ w/v), and NH4Cl ($5\%$ w/v). These reagents were obtained from Sigma Aldrich (St. Louis, MO, USA). The liquid medium ZMB was prepared according to Medina et al. [ 39]. Solid media contained $1.5\%$ w/v agar. Carbohydrates used were L-fucose, 2-O-fucosyllactose, 3-O-fucosyllactose, and sialic acid (Neu5ac), which were kindly donated by Glycom (Hørshol, Denmark). Mannose, glucose, galactose, and lactose were obtained from Sigma Aldrich (St. Louis, MO, USA). Carbohydrate solutions were prepared with Milli-Q water and then filtered with Millex-gv filters (0.22 μm). Plasmid construction. The in-silico plasmid construction was carried out in the SnapGene program, obtaining all the constructs and the primers (Table 1). Fucose biosensors were created in a high copy plasmid backbone (pTAC_sfGFP ColE1) and a low copy plasmid backbone (pTAC_sfGFP SC101). The high copy plasmid contains an ampicillin resistance gene as a selection marker and superfolder GFP (sfGFP) as a reporter molecule [40], while the low copy plasmid contains a chloramphenicol resistance and sfGFP controlled by pTac. These plasmids were a kind donation from Dr. Tal Danino (Columbia University). The fucose-induced promoter (pFuc) was synthesized as a gBlock from Integrated DNA Technologies, Inc. (IDT), including the PstI and EcoRI restriction sites at the 5′ and 3′ ends, respectively. The sequence was obtained from the E. coli K12 MG1655 genome, specifically from the fucose fucPIK operon [38]. Both DNA fragments were digested with the PstI-HF and EcoRI-HF restriction enzymes for 1 h at 37 °C, gel purified, and ligated with T4 DNA ligase at room temperature for 1 h (New England Biolabs, Inc., Ipswich, MA, USA). The resulting plasmids were named pFUC_sfGFP_colE1 and pFUC_sfGFP_SC101. Cloning of FucR. Later, the transcription factor gene (fucR) was obtained from the genome of E. coli K12 MG1655 by PCR with the primers 5′-tctcatACCGGTacgcccgcc-3′ and 5′-ctatCCCGGGtcaggctgttaccaaagaag-3′. These primers contain restriction sites for the enzymes AgeI and XmaI. PCR reactions were performed with Q5 high-fidelity polymerase (New England BioLabs, Ipswich, MA, USA) using manufacturer instructions. Exceptions were an annealing temperature of 70 °C and an extension time of 20 s, using 0.5 μM of the primers and 1 U of polymerase Q5. PCR products were recovered from a $1\%$ agarose gel with the Zymoclean Gel DNA Recovery Zymo research kit (Irvine, CA, USA). The pFUC_sfGFP_colE1 plasmid containing the fuc promoter was digested with the same enzymes and amplified with the primers 5′-TGAcgctagaactagtggatcc-3′ and 5′-tcagACCGGtagaccgagatagggttgag-3. PCR products were recovered (FucR and the fucose-induced promoter in the high-copy plasmid (Table 1)). Digestions were carried out for 16 h at 37 °C with AgeI-HF and XmaI enzymes following manufacturer instructions (New England Biolabs; Ipswich, MA, USA). Digested plasmids and fragments were ligated with a T4 DNA ligase (New England Biolabs; Ipswich, MA, USA) at room temperature for 16 h. Bacterial transformations. All plasmids were transformed into chemically competent E. coli strains. Plasmids were stored in DH5α, and biosensors were produced in the BL21 strain. Two microliters of ligation mixture or Gibson Assembly Master Mix were added to 50 µL of cells and incubated on ice for 30 min. Heat shock was performed at 42 °C for 50 s, followed by 2 min on ice. One ml of SOC media was added, and the bacteria were incubated at 37 °C with shaking at 200 rpm for 1 h. The transformation volume was plated onto LB agar plates with the corresponding antibiotic. Single colonies were picked and cultured in LB media with the antibiotic for stock preparation and miniprep. Carbenicillin was used at 100 µg/mL, and chloramphenicol was prepared in ethanol at 25 µg/mL. Correct insertion of genes of interest was verified through plasmid sequencing at Macrogen Inc (Seoul, Republic of Korea). Fluorescence kinetics. Four colonies were selected per plate and cultured in 2 mL of LB-antibiotic broth with 200-rpm agitation for 16 h at 37 °C. Filtered fucose (100 mM) was used to prepare 200 μL triplicate reactions with decreasing monosaccharide concentrations, inoculated at $1\%$ w/v with fresh LB-antibiotic medium. All kinetics were performed on black with transparent bottom Nunc™ F96 MicroWell™ 96-well polystyrene plates (Thermo Scientific, Waltham, MA, USA)), in a Synergy H1 Biotek multi-plate reader (Winooski, VT, USA). Growth curves were monitored for 24 h with agitation, measuring OD600 and fluorescence every 30 min with excitation at 485 nm and emission at 510 nm. The Gen5 3.09 software was used for absorbance and fluorescence measurements and data analysis. B. bifidum culture and fucose quantification. B. bifidum JCM 1254 was inoculated in de Mann Rogose Sharp (MRS) broth supplemented with cysteine $0.05\%$ for 48 h in an anaerobic jar at 37 °C with an anaerobic GasPak EZ patch (Becton Dickson, Franklin Lakes, NJ, USA). Cells were centrifuged at 12,000× g for 1 min after 48 h and resuspended in reduced mZMB broth [39] with no carbon source. B. bifidum was then cultured at $4\%$ w/v in 5 mL of mZMB supplemented with 2FL (81 mM) or with 3FL (20.4 mM) for 40 h under anaerobic conditions as above. Supernatants were recovered at 0, 12, 16, 20, 24, and 40 h. All supernatants were filtered with Millex-gv 0.22 μm filters (Sigma Aldrich, St. Louis, MO, USA), and pH was adjusted to 7 using NaOH 1 M. Supernatants were later analyzed using thin layer chromatography (TLC) in parallel to biosensor detection. Standards of fucose, 2FL, 3FL, lactose, galactose (all at $1\%$ w/v), and B. bifidum supernatants JCM1254 were used. TLC DC-Fertigfolien ALUGRAM Xtra Silica Gel 0.20 mm plates were used (Macherey-Nagel, Allentown, PA, USA), with 1 μL of each sample. A run solution was prepared with $50\%$ v/v of n-butanol and $25\%$ v/v of acetic acid in distilled water. Two ascents were performed to improve resolution. A staining solution was prepared with $0.5\%$ w/v naphthol and $5\%$ v/v sulfuric acid in ethanol. Statistical analysis. All curves represent the average of triplicates, and the standard deviation is shown. Statistical analyses, including linear regressions, were performed in GraphPad Prism 9. To determine the sensitivity of the regulation and potential cooperativity, the Hill equation for an activator was fitted to fluorescence/OD values [41]. Hill equation parameters were minimized to experimental data using Solver in Excel. β is the maximum expression rate, K represents the dissociation constant, [S] is the substrate concentration and n is the Hill cooperativity coefficient [42]. [ 1]d F/ODd t=βKnKn+sn ## 3.1. Biosensor Properties and Functions The three plasmids constructed in this study are shown in Table 1 and depicted in Supplementary Figure S1. The reporter gene sfGFP is controlled by pFuc and is a marker of selection that gives resistance to antibiotics. Low- and high-copy plasmids were evaluated. The transcription factor gene fucR was also cloned upstream of pFuc, with constitutive expression. The plasmids were used for chemical transformations of the BL21 and DH5α E. coli strains and tested with increasing fucose concentrations. E. coli BL21 transformed only with pFuc displayed good behavior with increased F/OD ratio to higher fucose concentrations (Figure 1A). This molecular system did not fluoresce with 0 mM fucose, suggesting a tight control (Figure 1A). Interestingly, the cloning of fucR and pFuc into the high copy plasmid increased fucose detection values by nearly $50\%$, with good resolution and increasing F/OD ratios in response to higher fucose concentrations in BL21 (Figure 1C). The low-copy plasmid biosensor (SC101) emitted smaller fluorescence values than high-copy plasmids (Figure 1B; $p \leq 0.0001$ at 50 mM fucose). The three plasmids transformed in E. coli DH5α generated F/OD curves that did not correlate well with fucose concentrations (Figure 1D–F). ## 3.2. Comparison of Biosensor Specificities High-copy biosensors with pFuc and pFuc+FucR in BL21 were preliminary evaluated for non-specific cross-detection of other monosaccharides. The system with only pFuc showed a crossed response with galactose, irrespective of its concentration (Figure 2A). This result is in part explained by the vigorous growth on galactose (Figure 2E). Low concentrations of glucose and mannose (5 and 10 mM), but not higher, also triggered GFP production (Figure 2B,D). Sialic acid appeared not to induce GFP expression (Figure 2C). These results indicate that the sole inclusion of the pFuc promoter is insufficient to provide a specific response to fucose. Interestingly, the inclusion of the transcription factor increased biosensor specificity (Figure 3). No positive F/OD values were obtained in the presence of galactose, glucose, sialic acid, or mannose (Figure 3). Specificity to these carbohydrates was highlighted by the good growth the biosensor showed in these sugars, with no fluorescence emitted (Figure 3). Finally, rhamnose is another 6-deoxyhexose sugar that could interfere with fucose sensing. A small crossed response was observed for rhamnose, indicating the molecular system needs further improvements in its specificity (Supplementary Figure S2). ## 3.3. Calibration Curves The biosensor E. coli BL21 pFuc + FucR high copy (colE1) was evaluated in a range of 0 to 3 mM of fucose to assess its performance under low concentrations (Figure 4). Even at 0.4 mM fucose, the system generated a measurable output (Figure 4A). It can be observed that from 15 h and after, fucose concentrations were well differentiated, with a positive linear correlation between fucose amounts and F/OD values (Figure 4B). Applying a linear regression to these parameters (Supplementary Table S1), the best correlation (higher R squared value) was obtained at 15.5 h (Figure 4B, Supplementary Table S1). ## 3.4. Sensitivity of the E. coli BL21 pFuc+FucR colE1 Biosensor The behavior of the biosensor was later evaluated in a broader range to be used for measurements of free fucose, from 0 mM to 45 mM (Figure 5A). As expected, increasing F/OD values were obtained. These data were used to determine the regulatory parameters of the Hill equation ([42]; see methods). F/OD values at 15.5 h were used to fit experimental data to the equation. Modeling results indicate a Hill coefficient value n of 1, which suggests that FucR regulates its promoter via simple non-cooperative regulation. This suggests that FucR binds only one fucose molecule upon binding its DNA. K is a dissociation constant, and a small value was obtained (5 mM). K indicates the affinity of FucR for its promoter, representing the concentration of fucose required to activate $50\%$ of the maximal response. ## 3.5. Measuring Fucose in the Supernatant of B. bifidum JCM1254 Finally, the biosensor was used to measure fucose concentrations from a bacterial supernatant (Figure 6). B. bifidum can ferment HMOs, especially 2FL and 3FL, as carbon sources. This microorganism displays extracellular α1-2 and α1-3 fucosidase activities, releasing free fucose in the medium and allowing the bacterium to use lactose [29]. The bacterium was cultured anaerobically for 40 h in a medium supplemented with either 2FL or 3FL. Samples were taken regularly (Figure 6) and incubated with the E. coli biosensor. OD and fluorescence measurements were taken for 24 h at an interval of 30 min. Figure 6A shows normalized F/OD values of the supernatants obtained from B. bifidum growing on 2FL. Supernatants from time points at 12–24 h generated low but increasing F/OD values in time (Figure 6A). A strong fucose signal was detected at 40 h. These results correlated well with a visual assessment of carbohydrates in TLC (Figure 6E), where a strong band with the same migration as fucose was observed. Finally, no major differences in growth were observed for the biosensors using the supernatants from multiple time points, suggesting they were not inhibitory. In the case of B. bifidum supernatants with 3FL, a similar result was observed compared to 2FL (Figure 6B). The released fucose concentration at 40 h was lower than in 2FL (Figure 6B), and the supernatant sample at 24 h also showed a significant fluorescent output. Similar to 2FL supernatants, only the 40 h sample showed a strong fucose band in the TLC, which correlated with fluorescence data. At 15.5 h of incubation, the biosensor incubated with the 2FL supernatant at 40 h presented a normalized F/OD value average of 8206.12, while for 3FL at 40 h was 3717.66 after 15.5 h. These values were used in a calibration curve obtained from the linear regression analysis (Figure 5A). The extrapolation of fucose concentrations in these supernatants was 42.4 mM for 2FL and 6.47 mM for 3FL. These values correlated well with TLC band intensity and appeared in the correct range compared with fucose standards of 1 and 10 mM (Figure 6E). ## 4. Discussion In this study, we constructed a biosensor for quantifying fucose in biological samples, using a molecular promoter and transcription factor naturally occurring in E. coli and using sfGFP as output. E. coli is well characterized by its L-fucose utilization mechanism [36]. A permease allows fucose entrance, and a feeder pathway allows L-fucose conversion into lactate and 1,2-propanediol, generating NADH and FADH [35]. An intermediate in this pathway, fuculose-1-phosphate, is the ligand recognized by the system regulator, FucR [37]. Therefore, our biosensor is expected to sense fuculose-1-phosphate and not directly fucose. The system requires that any external fucose sensed be first metabolized to generate an output. Fucose is not among the most preferred carbon sources for E. coli, compared to glucose, galactose, or arabinose [43,44]. It shows a slow growth in this substrate in minimal media [35]. Catabolic repression exerted by CRP on the fucose promoter is also complemented with small RNA regulation via Spot42 [45]. The biosensor developed here is based on the activation role of FucR, which binds its promoter in the presence of fuculose-1-phosphate and allows the expression of sfGFP. We were able to determine in this study that FucR displays simple regulation, showing no cooperativity and suggesting it acts as a monomer. It is known that rhamnose appears to induce the operon [46], and fucose can also activate the galactose galETK system in E. coli [43]. These findings indicate that crossed regulatory responses of fucose and FucR are common in E. coli and might alter biosensor specificity. Results in this study showed a high increase in specificity attributed to the presence of FucR. pFuc alone showed little specificity, indicating that other molecules can still induce leaky expression. The cloning of additional copies of FucR dramatically reduced crossed regulatory responses, probably increasing the threshold of fucose activation and resulting in a much tighter response. Finally, a much better resolution for strain BL21 compared to DH5α could be explained by the mutation in the lon protease in BL21, which allows a smaller reporter protein degradation and increased half-life [47]. The biosensor characterized showed a good linear response in the low concentration range (0–3 mM) or higher (0–45 mM). Some applications of the biosensor are as a diagnostic tool. Fucose is metabolized in the liver, and excess fucose is secreted in the urine [2,48]. A rare genetic disorder is fucosidosis, where fucose found in glycoconjugates cannot be removed and accumulates in the body resulting in severe consequences [49]. Therefore, quantifying fucose in urine could be of interest, especially since there is an increase in fucose concentrations in certain liver diseases [19,50]. Another field of application of biosensors is in GIT and gut microbiome research [21]. The rapid quantification of HMOs in breast milk samples is desirable, especially fucosylated molecules. Similarly, there is great interest in the enzymatic biosynthesis of these molecules, which requires quantifying fucose [51]. Finally, several gut microbes display α-fucosidase activities and use fucose as a carbon and energy source [6,9,52]. Bifidobacterium and Bacteroides species are well known for their extracellular activities, which release fucose from HMO, mucins, other glycoproteins, or glycolipids [3,10]. Therefore, free fucose can be expected to be detected in GIT contents in mammals. Free fucose is also known to participate in cross-feeding interactions, where the fucose released by one microorganism is imported and used by another. This has been observed during the consumption of mucin glycans and HMO, for example, between B. bifidum and Bifidobacterium breve [12,13,53]. Therefore, an accurate and inexpensive method for quantifying fucose could show how this monosaccharide is shared between species. In this study, the developed biosensor displayed a good performance in quantifying fucose derived from 2FL utilization by B. bifidum and could be used in studying cross-feeding interactions. ## 5. Conclusions A fluorescent quantification method of fucose was developed in this study in E. coli with a high copy plasmid containing a reporter sfGFP, a fucose promoter, and FucR. The biosensor showed good sensitivity and specificity, showing a linear response to increasing fucose concentrations from 0 to 45 mM, a range within physiological concentrations. A validation to quantify fucose in a bacterial supernatant during HMO utilization was achieved. This method could be coupled to other enzymes (fucosidases, endoglycosidases, peptidases) to determine the concentration of fucosylated glycoconjugates. This assay could be an important tool for clinical research (serum, urine), foods, bioprocessing in the production of fucosylated glycoproteins, or detecting fucose in bacterial supernatants and quantifying fucose in fecal samples or bioreactors. ## References 1. Schneider M., Al-Shareffi E., Haltiwanger R.S.. Biological functions of fucose in mammals. Glycobiology (2017) 27 601-618. DOI: 10.1093/glycob/cwx034 2. Becker D.J., Lowe J.B.. Fucose: Biosynthesis and biological function in mammals. Glycobiology (2003) 13 41R-53R. DOI: 10.1093/glycob/cwg054 3. González-Morelo K.J., Vega-Sagardía M., Garrido D.. Molecular Insights Into O-Linked Glycan Utilization by Gut Microbes. Front. Microbiol. (2020) 11 591568. DOI: 10.3389/fmicb.2020.591568 4. Varki A., Lowe J.B., Varki A., Cummings R., Esko J.. Biological roles of glycans. Essentials of Glycobiology (2009) 5. Goto Y., Uematsu S., Kiyono H.. Epithelial glycosylation in gut homeostasis and inflammation. Nat. Immunol. (2016) 17 1244-1251. DOI: 10.1038/ni.3587 6. Sela D.A., Garrido D., Lerno L., Wu S., Tan K., Eom H.-J., Joachimiak A., Lebrilla C.B., Mills D.A.. Bifidobacterium longum subsp. infantis ATCC 15697 α-fucosidases are active on fucosylated human milk oligosaccharides. Appl. Environ. Microbiol. (2012) 78 795-803. DOI: 10.1128/AEM.06762-11 7. Ojima M.N., Asao Y., Nakajima A., Katoh T., Kitaoka M., Gotoh A., Hirose J., Urashima T., Fukiya S., Yokota A.. Diversification of a Fucosyllactose Transporter within the Genus Bifidobacterium. Appl. Environ. Microbiol. (2022) 88 e01437-21. DOI: 10.1128/AEM.01437-21 8. Marcobal A., Barboza M., Sonnenburg E.D., Pudlo N., Martens E.C., Desai P., Lebrilla C.B., Weimer B.C., Mills D.A., German J.B.. Bacteroides in the Infant Gut Consume Milk Oligosaccharides via Mucus-Utilization Pathways. Cell Host Microbe (2011) 10 507-514. DOI: 10.1016/j.chom.2011.10.007 9. Ruiz-Moyano S., Totten S.M., Garrido D.A., Smilowitz J.T., Bruce German J., Lebrilla C.B., Mills D.A.. Variation in consumption of human milk oligosaccharides by infant gut-associated strains of bifidobacterium breve. Appl. Environ. Microbiol. (2013) 79 6040-6049. DOI: 10.1128/AEM.01843-13 10. Hooper L.V., Xu J., Falk P.G., Midtvedt T., Gordon J.I.. A molecular sensor that allows a gut commensal to control its nutrient foundation in a competitive ecosystem. Proc. Natl. Acad. Sci. USA (1999) 96 9833-9838. DOI: 10.1073/pnas.96.17.9833 11. Gotoh A., Katoh T., Sakanaka M., Ling Y., Yamada C., Asakuma S., Urashima T., Tomabechi Y., Katayama-Ikegami A., Kurihara S.. Sharing of human milk oligosaccharides degradants within bifidobacterial communities in faecal cultures supplemented with Bifidobacterium bifidum. Sci. Rep. (2018) 8 13958. DOI: 10.1038/s41598-018-32080-3 12. Tsukuda N., Yahagi K., Hara T., Watanabe Y., Matsumoto H., Mori H., Higashi K., Tsuji H., Matsumoto S., Kurokawa K.. Key bacterial taxa and metabolic pathways affecting gut short-chain fatty acid profiles in early life. ISME J. (2021) 15 2574-2590. DOI: 10.1038/s41396-021-00937-7 13. Bunesova V., Lacroix C., Schwab C.. Mucin Cross-Feeding of Infant Bifidobacteria and Eubacterium hallii. Microb. Ecol. (2018) 75 228-238. DOI: 10.1007/s00248-017-1037-4 14. Pacheco A.R., Munera D., Waldor M.K., Sperandio V., Ritchie J.M.. Fucose sensing regulates bacterial intestinal colonization. Nature (2012) 492 113-117. DOI: 10.1038/nature11623 15. Pickard J.M., Chervonsky A.V.. Intestinal Fucose as a Mediator of Host–Microbe Symbiosis. J. Immunol. (2015) 194 5588-5593. DOI: 10.4049/jimmunol.1500395 16. Pickard J.M., Maurice C.F., Kinnebrew M.A., Abt M.C., Schenten D., Golovkina T.V., Bogatyrev S.R., Ismagilov R.F., Pamer E.G., Turnbaugh P.J.. Rapid fucosylation of intestinal epithelium sustains host-commensal symbiosis in sickness. Nature (2014) 514 638-641. DOI: 10.1038/nature13823 17. Furniss R.C.D., Clements A.. Regulation of the locus of enterocyte effacement in attaching and effacing pathogens. J. Bacteriol. (2018) 200 e00336-17. DOI: 10.1128/JB.00336-17 18. Gibson G.R., McCartney A.L., Rastall R.A.. Prebiotics and resistance to gastrointestinal infections. Br. J. Nutr. (2005) 93 S31-S34. DOI: 10.1079/BJN20041343 19. Yamauchi M., Nishikawa F., Kimura K., Maezawa Y., Ohata M., Toda G.. Urinaryl-fucose: A new marker for the diagnosis of chronic liver disease. Int. Hepatol. Commun. (1993) 1 222-227. DOI: 10.1016/0928-4346(93)90044-G 20. Tsiafoulis C.G., Prodromidis M.I., Karayannis M.I.. Development of an amperometric biosensing method for the determination of L-fucose in pretreated urine. Biosens. Bioelectron. (2004) 20 620-627. DOI: 10.1016/j.bios.2004.03.012 21. McGovern D.P.B., Jones M.R., Taylor K.D., Marciante K., Yan X., Dubinsky M., Ippoliti A., Vasiliauskas E., Berel D., Derkowski C.. Fucosyltransferase 2 (FUT2) non-secretor status is associated with Crohn’s disease. Hum. Mol. Genet. (2010) 19 3468-3476. DOI: 10.1093/hmg/ddq248 22. Wang H., Walaszczyk E.J., Li K., Chung-Davidson Y.-W., Li W.. High-performance liquid chromatography with fluorescence detection and ultra-performance liquid chromatography with electrospray tandem mass spectrometry method for the determination of indoleamine neurotransmitters and their metabolites in sea lamprey plasma. Anal. Chim. Acta (2012) 721 147-153. DOI: 10.1016/j.aca.2012.01.025 23. Chen Y., Yao Q., Zeng X., Hao C., Li X., Zhang L., Zeng P.. Determination of monosaccharide composition in human serum by an improved HPLC method and its application as candidate biomarkers for endometrial cancer. Front. Oncol. (2022) 12 1014159. DOI: 10.3389/fonc.2022.1014159 24. Suzuki J., Kondo A., Kato I., Hase S., Ikenaka T.. Assay of urinary free fucose by fluorescence labeling and high-performance liquid chromatography. Clin. Chem. (1992) 38 752-755. DOI: 10.1093/clinchem/38.5.752 25. Chaturvedi P., Warren C.D., Altaye M., Morrow A.L., Ruiz-Palacios G., Pickering L.K., Newburg D.S.. Fucosylated human milk oligosaccharides vary between individuals and over the course of lactation. Glycobiology (2001) 11 365-372. DOI: 10.1093/glycob/11.5.365 26. Sakai T., Yamamoto K., Yokota H., Hakozaki-Usui K., Hino F., Kato I.. Rapid, simple enzymatic assay of free L-fucose in serum and urine, and its use as a marker for cancer, cirrhosis, and gastric ulcers. Clin. Chem. (1990) 36 474-476. DOI: 10.1093/clinchem/36.3.474 27. Erlandsson P.G., Åström E., Påhlsson P., Robinson N.D.. Determination of Fucose Concentration in a Lectin-Based Displacement Microfluidic Assay. Appl. Biochem. Biotechnol. (2019) 188 868-877. DOI: 10.1007/s12010-018-02944-5 28. Gao Z., Chen S., Du J., Wu Z., Ge W., Gao S., Zhou Z., Yang X., Xing Y., Shi M.. Quantitative analysis of fucosylated glycoproteins by immobilized lectin-affinity fluorescent labeling. RSC Adv. (2023) 13 6676-6687. DOI: 10.1039/D3RA00072A 29. Thurl S., Müller-Werner B., Sawatzki G.. Quantifbation of individual oligosaccharide compounds from human milk using high-pH anion-exchange chromatography. Anal. Biochem. (1996) 235 202-206. DOI: 10.1006/abio.1996.0113 30. Shin J., Park M., Kim C., Kim H., Park Y., Ban C., Yoon J.W., Shin C.S., Lee J.W., Jin Y.S.. Development of fluorescent Escherichia coli for a whole-cell sensor of 2′-fucosyllactose. Sci. Rep. (2020) 10 10514. DOI: 10.1038/s41598-020-67359-x 31. Della Corte D., van Beek H.L., Syberg F., Schallmey M., Tobola F., Cormann K.U., Schlicker C., Baumann P.T., Krumbach K., Sokolowsky S.. Engineering and application of a biosensor with focused ligand specificity. Nat. Commun. (2020) 11 4851. DOI: 10.1038/s41467-020-18400-0 32. Barra M., Danino T., Garrido D.. Engineered Probiotics for Detection and Treatment of Inflammatory Intestinal Diseases. Front. Bioeng. Biotechnol. (2020) 8 265. DOI: 10.3389/fbioe.2020.00265 33. Sharma S., Kumar R., Jain A., Kumar M., Gauttam R., Banerjee R., Mukhopadhyay J., Tyagi J.S.. Functional insights into Mycobacterium tuberculosis DevR-dependent transcriptional machinery utilizing Escherichia coli. Biochem. J. (2021) 478 3079-3098. DOI: 10.1042/BCJ20210268 34. Pierre M., Julien P., Fabien L.. Functional Analysis of Deoxyhexose Sugar Utilization in Escherichia coli Reveals Fermentative Metabolism under Aerobic Conditions. Appl. Environ. Microbiol. (2021) 87 e00719-21. DOI: 10.1128/AEM.00719-21 35. Kim J., Cheong Y.E., Jung I., Kim K.H.. Metabolomic and transcriptomic analyses of Escherichia coli for efficient fermentation of L-Fucose. Mar. Drugs (2019) 17. DOI: 10.3390/md17020082 36. Bartkus J.M., Mortlock R.P.. Isolation of a mutation resulting in constitutive synthesis of L-fucose catabolic enzymes. J. Bacteriol. (1986) 165 710-714. DOI: 10.1128/jb.165.3.710-714.1986 37. Chen Y.-M., Zhu Y., Lin E.C.C.. The organization of the fuc regulon specifying l-fucose dissimilation in Escherichia coli K12 as determined by gene cloning. Mol. Gen. Genet. MGG (1987) 210 331-337. DOI: 10.1007/BF00325702 38. Zhu Y., Lin E.C.. A mutant crp allele that differentially activates the operons of the fuc regulon in Escherichia coli. J. Bacteriol. (1988) 170 2352-2358. DOI: 10.1128/jb.170.5.2352-2358.1988 39. Medina D.A., Pinto F., Ovalle A., Thomson P., Garrido D.. Prebiotics mediate microbial interactions in a consortium of the infant gut microbiome. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18102095 40. Pédelacq J.-D., Cabantous S., Tran T., Terwilliger T.C., Waldo G.S.. Engineering and characterization of a superfolder green fluorescent protein. Nat. Biotechnol. (2006) 24 79-88. DOI: 10.1038/nbt1172 41. Madar D., Dekel E., Bren A., Alon U.. Negative auto-regulation increases the input dynamic-range of the arabinose system of Escherichia coli. BMC Syst. Biol. (2011) 5. DOI: 10.1186/1752-0509-5-111 42. Alon U.. An Introduction to Systems Biology: Design Principles of Biological Circuits (2006) 43. Mangan S., Itzkovitz S., Zaslaver A., Alon U.. The Incoherent Feed-forward Loop Accelerates the Response-time of the gal System of Escherichia coli. J. Mol. Biol. (2006) 356 1073-1081. DOI: 10.1016/j.jmb.2005.12.003 44. Aidelberg G., Towbin B.D., Rothschild D., Dekel E., Bren A., Alon U.. Hierarchy of non-glucose sugars in Escherichia coli. BMC Syst. Biol. (2014) 8. DOI: 10.1186/s12918-014-0133-z 45. Beisel C.L., Storz G.. The Base-Pairing RNA Spot 42 Participates in a Multioutput Feedforward Loop to Help Enact Catabolite Repression in Escherichia coli. Mol. Cell (2011) 41 286-297. DOI: 10.1016/j.molcel.2010.12.027 46. Chen Y.M., Tobin J.F., Zhu Y., Schleif R.F., Lin E.C.. Cross-induction of the L-fucose system by L-rhamnose in Escherichia coli. J. Bacteriol. (1987) 169 3712-3719. DOI: 10.1128/jb.169.8.3712-3719.1987 47. Gottesman S.. Proteases and their targets in Escherichia coli. Annu. Rev. Genet. (1996) 30 465-506. DOI: 10.1146/annurev.genet.30.1.465 48. Coffey J.W., Miller O.N., Sellinger O.Z.. The Metabolism of l-Fucose in the Rat. J. Biol. Chem. (1964) 239 4011-4017. DOI: 10.1016/S0021-9258(18)91124-5 49. Stepien K.M., Ciara E., Jezela-Stanek A.. Fucosidosis—Clinical Manifestation, Long-Term Outcomes, and Genetic Profile—Review and Case Series. Genes (2020) 11. DOI: 10.3390/genes11111383 50. Wang H., Zhang X., Peng Y., Pan B., Wang B., Peng D.H., Guo W.. A LC-MS/MS method to simultaneously profile 14 free monosaccharides in biofluids. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. (2022) 1192 123086. DOI: 10.1016/j.jchromb.2021.123086 51. Li M., Luo Y., Hu M., Li C., Liu Z., Zhang T.. Module-Guided Metabolic Rewiring for Fucosyllactose Biosynthesis in Engineered Escherichia coli with Lactose De Novo Pathway. J. Agric. Food Chem. (2022) 70 14761-14770. DOI: 10.1021/acs.jafc.2c05909 52. Crost E.H., Tailford L.E., Le Gall G., Fons M., Henrissat B., Juge N.. Utilisation of Mucin Glycans by the Human Gut Symbiont Ruminococcus gnavus Is Strain-Dependent. PLoS ONE (2013) 8. DOI: 10.1371/journal.pone.0076341 53. Everard A., Belzer C., Geurts L., Ouwerkerk J.P., Druart C., Bindels L.B., Guiot Y., Derrien M., Muccioli G.G., Delzenne N.M.. Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity. Proc. Natl. Acad. Sci. USA (2013) 110 9066-9071. DOI: 10.1073/pnas.1219451110
--- title: 'SARS-CoV-2 Neutralizing Antibodies in Mexican Population: A Five Vaccine Comparison' authors: - Fernando Alcorta-Nuñez - Diana Cristina Pérez-Ibave - Carlos Horacio Burciaga-Flores - Miguel Ángel Garza - Moisés González-Escamilla - Patricia Rodríguez-Niño - Juan Francisco González-Guerrero - Adelina Alcorta-Garza - Oscar Vidal-Gutiérrez - Genaro A. Ramírez-Correa - María Lourdes Garza-Rodríguez journal: Diagnostics year: 2023 pmcid: PMC10046906 doi: 10.3390/diagnostics13061194 license: CC BY 4.0 --- # SARS-CoV-2 Neutralizing Antibodies in Mexican Population: A Five Vaccine Comparison ## Abstract Neutralizing antibodies (NAs) are key immunological markers and are part of the humoral response of the adaptive immune system. NA assays determine the presence of functional antibodies to prevent SARS-CoV-2 infection. We performed a real-world evidence study to detect NAs that confer protection against SARS-CoV-2 after the application of five vaccines (Pfizer/BioNTech, AstraZeneca, Sinovac, Moderna, and CanSino) in the Mexican population. Side effects of COVID-19 vaccines and clinical and demographic factors associated with low immunogenicity were also evaluated. A total of 242 SARS-CoV-2-vaccinated subjects were recruited. Pfizer/BioNTech and Moderna proved the highest percentage of inhibition in a mono-vaccine scheme. Muscular pain, headache, and fatigue were the most common adverse events. None of the patients reported severe adverse events. We found an estimated contagion-free time of 207 (IQR: 182–231) and 187 (IQR: 184–189) days for Pfizer/BioNTech and CanSino in 12 cases in each group. On the basis of our results, we consider that the emerging vaccination strategy in Mexico is effective and safe. ## 1. Introduction Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged in China in December 2019 and caused a worldwide severe respiratory disease (COVID-19) [1], which was declared a pandemic on 11 March 2020 by the WHO (World Health Organization) [2]. Three years after the novel coronavirus SARS-CoV-2 shook the world with a global health crisis, it remains a public health problem. COVID-19 has caused 619,161,228 cumulative infections and more than 6 million deaths worldwide as of 11 Oct 2022, according to the WHO COVID-19 Dashboard [3]. Vaccines were developed and administrated to prevent the novel coronavirus’s spread and protect at-risk populations. More than 100 vaccines have been developed, and according to the WHO, 26 COVID-19 vaccines have been evaluated in phase III clinical trials [4,5]. There are different types of COVID-19 vaccines, including the adenovirus-vectored vaccines, such as the CanSino and Oxford University/AstraZeneca COVID-19 vaccines (Ad5-CoV and ChAdOx1 nCoV-19, respectively); the mRNA vaccines from Pfizer/BioNTech and Moderna (BNT162b1 and mRNA-1273); and the inactivated vaccine from Sinovac (PiCoVacc-CoronaVac), among others [1,6,7]. To date, the Food and Drug Administration (FDA), the World Health Organization (WHO), and the European Medicine Agency (EMA) have approved four vaccines, including Pfizer/BioNTech, Moderna, and Novavax (NVX-CoV2373). Additionally, the EMA has approved Valneva, and the WHO has approved Sinopharm (BBIBP-CorV), Sinovac CoronaVac (J07BX03), Bharat Biotech Covaxin (BBV152), and CanSino. In Mexico, the vaccination campaign started on December 2020. On the basis of the probability of infection and lethality, the government decided to start vaccination for health workers and people aged 60 years or older with or without comorbidities, followed by people a decade younger until completing the rest of the population [8,9]. Until November 2022, the regulatory agency in Mexico (the Federal Commission of Protection of Sanitary Risks, COFEPRIS) approved the following vaccines for administration in the Mexican population: Pfizer/BioNTech, AstraZeneca, Sputnik V, Sinovac, CanSino, COVAX, Moderna, Sinopharm, and Abdala [10]. Up to December 2022, Mexico had administered 225,063,079 doses of COVID vaccines ($82.2\%$ of the population with two doses), with good acceptance and proper coverage independent of the type of vaccine [4,11]. Several published studies have suggested that COVID-19 vaccines are effective and well-tolerated. However, it has been observed that the effectiveness of vaccines varies according to the sample characteristics of the population [5,12]. Effectiveness refers to how well the vaccine performs in the real world [5]. On the other hand, Mexican population data are scarce. Knowing whether and to what extent vaccine effectiveness wanes is crucial to vaccine policy decisions, such as the need for, timing, and target populations for booster doses [13]. The BNT162b2 vaccine, commercially known as the Pfizer/BioNTech COVID-19 vaccine, was the first approved by FDA and has demonstrated efficacy against the original strain of COVID-19 and other variants. It is a lipid nanoparticle-formulated, nucleoside-modified RNA (modRNA) vaccine [14] with two proline mutations to lock it in the prefusion conformation and has the ability to encode the trimerized receptor-binding domain of the spike protein of SARS-CoV-2, which allows it to mimic intact virus infection [14,15]. CanSino Biologics Inc., in collaboration with the Beijing Institute of Biotechnology, developed a vaccine that is not currently approved by the FDA; however, the vaccine received a nod from the WHO on 11 May 2022 and has been approved in over 10 countries, including China and Mexico. It is a non-replicating viral vector vaccine against SARS-CoV-2 called Adenovirus Type 5 Vector (Ad5-nCoV). Ad5-nCoV encodes the full spike protein of SARS-CoV-2 and has shown enough immunogenicity in human clinical trials [16]. Neutralizing antibody (NA) assays determine the presence of functional antibodies to prevent SARS-CoV-2 infection [17,18]. Vaccine effectiveness is measured by the quality of performance outside clinical trials [5,19]. Neutralizing antibodies are valuable tools to analyze the performance of vaccines, and they are cheap and easy to test outside clinical trials. This real-world evidence study shows the neutralization antibody titers that confer protection against SARS-CoV-2 after the application of five vaccines (BNT162b2, AZD1222ChAdOx1, Ad5-nCoV, mRNA-1273, and CoronaVac) in the Mexican population. We also evaluated the side effects of COVID-19 vaccines and the clinical and demographic factors associated with low immunogenicity. ## 2.1. Study Population This study was conducted following the guidelines of the Declaration of Helsinki. We collected serum samples from 242 subjects vaccinated against COVID-19 with a full vaccination scheme (143 vaccinated with Pfizer, 49 with CanSino, 21 with Sinovac, 17 with AstraZeneca, and 17 with Moderna). Vaccination schemes were considered completed if the patients had at least 2 doses of Pfizer/BioNTech, 1 dose of CanSino, 2 doses of AstraZeneca, 2 doses of Sinovac, or 2 doses of Moderna. The subjects were recruited from March to December 2021 at the Centro Universitario Contra el Cáncer (CUCC) of the Universidad Autónoma de Nuevo León (U.A.N.L.) in Monterrey, Nuevo León, México. All the subjects were ≥18 years old, signed an informed consent letter, and answered a questionnaire with clinical and demographic information (including their history of SARS-CoV-2 infection, vaccine certificates, doses, vaccine-associated side effects, etc.). Peripheral blood samples were taken to obtain serum and stored at −80 °C until analysis. The institutional ethics committee of the university hospital (Comité de Ética en Investigación del Hospital Universitario “Dr. José Eleuterio González”) approved the protocol with the registration number ON21-00028. ## 2.2. Survey and Data Collection Patients were invited to participate in the study, and an informed consent form was signed after an interview. Clinical and demographic data were collected using a 15 min questionnaire in the SurveyMonkey® platform through electronic devices (smartphones and tablets) [20]. ## 2.3. Neutralizing Antibodies Detection Blood samples were centrifuged at 3750 rpm for 10 min at room temperature to obtain serum in an Eppendorf 5804R Refrigerated Centrifuge (Hamburg, Germany). The quantification of neutralizing antibodies was performed with the cPass™ SARS-CoV-2 Neutralization Antibody Detection Kit (GenScript, Piscataway Township, NJ, USA) according to the manufacturer’s instructions. To make a semi-quantitative analysis, we added a standard curve using a monoclonal NA (MAB), SARS-CoV-2 NA, as previously described [16]. The samples and standards were read at 450 nm in a Cytation™ 3 Cell Imaging Multi-Mode Reader (BioTek®, Winooski, VT, USA). The following formula was used to calculate the level of signal inhibition: signal inhibition (%) = (1 − OD value of sample/OD value of negative control) × 100. The results were interpreted as follows: positive results, ≥$31\%$ of inhibition, and negative results, ≤$30\%$ of inhibition. ## 2.4. Statistical Analysis Data analysis was performed using the IBM SPSS Statistics for Windows version 25.0 software (Armonk, NY, USA: IBM Corp.). Graphs, frequency tables, and crossed tables were constructed for categorical variables. For quantitative variables, we perform descriptive statistics, such as the mean, standard deviation, median, variance, and box plots. For comparisons of means, we used Student’s t-test. Data with non-parametric distributions were represented as medians with interquartile ranges (IQRs); to compare groups, we used the Mann–Whitney U-test, Kruskal–Wallis test, and Wilcoxon signed-rank test. The significance level was set at $p \leq 0.05.$ We used the Kaplan–Meier method on Pfizer/BioNTech and CanSino for survival analysis. ## 3.1. Description of Study Groups A total of 242 SARS-CoV-2 vaccinated subjects were recruited. There were 165 females ($68\%$), and the median age was 32 years (IQR: 25–42) (Table 1). The most frequent vaccines were Pfizer/BioNTech and CanSino, with $58\%$ ($$n = 140$$) and $19\%$ ($$n = 49$$), respectively (Table 2). Among comorbidities, we found overweight in $34\%$ ($$n = 83$$), obesity in $24\%$ ($$n = 59$$), hypertension in $9\%$ ($$n = 22$$), and diabetes mellitus in $2\%$ ($$n = 5$$) (Table 1). We analyzed patients with COVID-19 before and after the vaccination scheme. There was no statistical significance in antibody titers in patients with SARS-CoV-2 infection prior to vaccination, and patients with COVID-19 after vaccination had higher antibody titers than patients without infection ($$p \leq 0.001$$) (Table 2). ## 3.2. Quantification of Neutralizing Antibodies An exploratory analysis was carried out by the type of vaccine and their level of inhibition against SARS-CoV-2; Pfizer/BioNTech, Moderna, and CanSino showed the highest percentage of inhibition with medians of $97.23\%$, $97.61\%$, and $97.23\%$ (IQR: 94.24–97.70, 97.23–97.94, and 73.21–97.48), respectively (Table 3). The neutralizing activity of the vaccines over time (12 months) was also measured, with a clear decrease in neutralizing antibodies after six months (Figure 1). Pfizer/BioNTech achieved better immunogenicity than CanSino ($100\%$ vs. $85\%$), with a significance $p \leq 0.001$ (Supplementary data). We found differences in immunogenicity; women had a higher percentage of inhibition than males. No statistical differences were found in any of the comorbidities analyzed. ## 3.3. Survival Analysis We performed a survival analysis on 36 patients with COVID-19 after vaccination with Pfizer/BioNTech ($$n = 12$$) and CanSino ($$n = 12$$). The patients were RT-PCR positive, and the global survival was estimated at 187 (IQR: 184–189) days with immune protection. Pfizer/BioNTech was the most effective vaccine, with 207 (182–231) days, and CanSino had an estimation of 187 (184–189) days (Figure 2). ## 3.4. Vaccine-Associated Side Effects The most common side effects reported were muscular pain, headache, fatigue, fever, joint pain, and shivers (Figure 3). For the first and second vaccination doses, muscular pain and headache were the most frequent side effects ($52\%$ and $41\%$ for the first dose ($$n = 242$$) and $62\%$ and $44\%$ for the second dose ($$n = 156$$), respectively). ## 4. Discussion In Mexico, more than 209 million doses have been administered, with a ratio of 162.62 total doses per 100 population [3]. Different vaccination schemes were administrated according to vaccines available, as in other developed countries; this contrasted with countries such as the United Kingdom, the United States, China, and Russia, who developed vaccines such as AstraZeneca, Pfizer/BioNTech, CanSino, and Sputnik, which were openly available to the entire population prior to other countries [21]. In Latin America (Mexico, Argentina, Chile, Ecuador, Brazil, Colombia, and Peru), the Chinese CanSino vaccine was one of the vaccines used by health authorities in the middle of the pandemic crisis, even though there was not enough information about its efficacy and security [16,22]. The combination of vaccines for generating immunity against SARS-CoV-2 has proven to be effective and safe as a vaccination strategy [23,24,25]. This allowed developing countries to complete vaccination schemes on time to reduce the cumulative incidence of COVID-19 and subsequently reinforce the vaccination scheme as the availability of vaccines increased. In our study, Pfizer/BioNTech and Moderna showed the highest percentage of inhibition in the mono-vaccine scheme. Still, compared with an exploratory group of a heterologous vaccination scheme of CanSino (CanSino X or any other vaccine), the inhibition levels were boosted to achieve similar mono-vaccine schemes. In addition, Pfizer/BioNTech had the longest titer inhibition period, similar to that of the heterologous CanSino vaccination scheme (Table S1) [7,16,24,26]. A higher generation of antibodies is associated with a longer duration of seroprotection [7]. According to other studies, the Pfizer/BioNTech vaccine has known global efficacy and safety with two doses of 30 μg of BNT162b2, with $91.3\%$ ($95\%$ confidence interval (CI), 89.0 to 93.2) and specifically $96.7\%$ ($95\%$ CI, 80.3 to 99.9) efficacy against severe and moderate symptomatic disease caused by SARS-CoV-2 infection in a period of 6 months without previous infection. Across different countries, in a wide range of age groups of both sexes with a diverse spectrum of risk factors, the efficacy was 86 to $100\%$. This efficacy showed a decline over time. In South Africa, for the B.1.351 variant, the efficacy is $100\%$ ($95\%$ CI, 53.5 to 100) [14,27]. Pfizer/BioNTech showed a broad immune response with SARS-CoV-2 S-specific neutralizing antibodies for the BNT162b2 vaccine, inducing poly-specific CD4+ and CD8+ T cells for the original strain [28], with around $83.3\%$ with neutralizing antibodies 14 days after the first dose and up to 56 days after the second dose [29]. In the phase III study for CanSino, they found that one dose of Ad5-nCoV showed a $57.5\%$ ($95\%$ CI 39.7–70.0, $$p \leq 0.0026$$) efficacy against symptomatic PCR-confirmed COVID-19 infection 28 days or more after vaccination (21,250 participants; 45 days median duration of follow-up (IQR 36–58)). It was $63.7\%$ efficacious against symptomatic PCR-confirmed COVID-19 infection beginning 14 days after vaccination. The Ad5-nCoV vaccine was $91.7\%$ effective against severe disease 28 days after vaccination and $96.0\%$ effective 14 days after vaccination. Their findings were from different countries around the world, including Mexico, Russia, and Pakistan, in participants of both sexes over 18 years old [30]. Additionally, we found a higher percentage of inhibition in women; this was previously reported in other studies, in which men had lower inhibition levels [31]. This may be explained by different interacting factors, such as environmental, genetic, and hormone factors, which differ between sexes and vary throughout life, with a general understanding that adult females mount stronger innate and adaptive immune responses than males [32]. Regarding comorbidities, including obesity, we found no statistical differences; this is important to mention because other studies have found discrepant data; some proposed a potential association between obesity and low NA titers [33], and others found that obesity had a probable booster effect in the generation of NAs [31]. This is still controversial since Mexico ranks eighth in the world for obesity; more studies should be conducted to clarify this [34]. We found an estimated median contagion-free time of 207 (IQR: 182–231), and 187 (IQR: 184–189) days for Pfizer/BioNTech and CanSino, respectively ($$n = 24$$). In accordance with Dr. Maria Krutik et al., who reported a seroconversion within 90–180 days, after this period, the level of NAs began to decrease [35]. The early contagion after vaccination with Pfizer/BioNTech vaccine could be explained by the fact that most of the individuals were healthcare staff on the front lines of pandemic combat who were exposed to a higher risk of contagion. CanSino was primarily administrated by personnel working in lower-risk areas of the hospital [35]. We identified seven patients ($15\%$) vaccinated with CanSino in the single-dose modality who were negative for NAs. This agrees with another study that found similar results, with a prevalence of $11\%$ seronegative for IgG-type antibodies [7]. Vaccine seronegative responders should be further studied to explain factors associated with poor antibody response. Compared with other studies in which a more significant generation of neutralizing antibodies has been associated with positive COVID-19 patients before or after the vaccination scheme, our results revealed a significant generation of antibody titers in positive patients after the vaccination scheme, considering the lack of population to verify both theories with certainty as a limiting factor [36,37]. We found that muscular pain ($52\%$), headache ($41\%$), and fatigue ($35\%$) were among the most common adverse events; specifically, participants who received Pfizer/BioNTech ($70\%$) and CanSino ($73\%$) reported at least one minor side effect. None of the vaccines had severe adverse events. International data have shown that with two doses of Pfizer/BioNTech of 30 μg, only $27\%$ of the patients have minimal adverse effects, characterized by short-term, mild-to-moderate pain at the injection site, fatigue, and headache within 14 days after the second dose. After six months of the vaccination, $30\%$ had any adverse events, and only $1.2\%$ had severe events. In adolescents, only $6\%$ had related adverse events, and $0.6\%$ had severe events [27,38,39]. For CanSino, in the primary safety analysis, only $0.1\%$ of participants reported serious adverse events. In the extended safety cohort, 63·$5\%$ reported adverse events, of which headache was the most common systemic adverse event ($44\%$). In addition, $59\%$ reported pain at the injection site [30]. We found a higher percentage of minor side effects; this can be explained by the fact that our population was primarily young adults (median 32 years, IQR: 25–42) with similar comorbidities with an expected competent immune response. In addition, the survey allowed participants to self-report symptoms in a practical way. As expected in a real-world data study, investigator bias was not present; the results tend to be more attached to reality. We did not find any major side effects, as reported by other studies [40,41]. The main disadvantages of this study were the small sample and the vaccine diversity. Access to vaccines has been limited in Mexico. Even with this drawback, this exploratory study of five vaccine types allowed us to have an overview of neutralizing antibody titers in our population. An advantage of this study is that we had semi-quantitative determinations to make the correlations between antibody levels and their neutralizing capacity. The blocking of neutralizing antibodies is equivalent to the gold standard neutralization test [42,43]. ## 5. Conclusions We found high NA titers in all vaccines. Pfizer/BioNTech and Moderna produced the highest antibody titers and longer immuno-protection. Pfizer/BioNTech and CanSino vaccines were tolerated and generated NAs in most participants. CanSino non-responders should be studied in greater detail to identify determinant factors, including clinical and sociodemographic characteristics involved in the low response to vaccination. The CanSino combination scheme proved to be effective and safe. As in other developed countries, the Mexican government’s strategy attempted to apply several vaccines to reach a broad approach. This finding could be useful to define criteria for the application of vaccines in developing countries with difficult access to vaccination, such as Mexico. On the basis of our results, we consider that the emerging vaccination strategy in Mexico is effective and safe. ## References 1. Liu L.D., Lian C., Yeap L.S., Meng F.L.. The development of neutralizing antibodies against SARS-CoV-2 and their common features. J. Mol. Cell. Biol. (2020.0) 12 980-986. DOI: 10.1093/jmcb/mjaa070 2. Coronavirus Disease (COVID-19) Pandemic 3. WHO Coronavirus (COVID-19) Dashboard 4. COVID-19 Vaccine Tracker and Landscape 5. Fiolet T., Kherabi Y., MacDonald C.J., Ghosn J., Peiffer-Smadja N.. Comparing COVID-19 vaccines for their characteristics, efficacy and effectiveness against SARS-CoV-2 and variants of concern: A narrative review. Clin. Microbiol. Infect. (2022.0) 28 202-221. DOI: 10.1016/j.cmi.2021.10.005 6. Alter G., Yu J., Liu J., Chandrashekar A., Borducchi E.N., Tostanoski L.H., McMahan K., Jacob-Dolan C., Martinez D.R., Chang A.. Immunogenicity of Ad26.COV2.S vaccine against SARS-CoV-2 variants in humans. Nature (2021.0) 596 268-272. DOI: 10.1038/s41586-021-03681-2 7. Guzman-Martinez O., Guardado K., de Guevara E.L., Navarro S., Hernandez C., Zenteno-Cuevas R., Montero H.. IgG Antibodies Generation and Side Effects Caused by Ad5-nCoV Vaccine (CanSino Biologics) and BNT162b2 Vaccine (Pfizer/BioNTech) among Mexican Population. Vaccines (2021.0) 9. DOI: 10.3390/vaccines9090999 8. Priorización inicial y consecutiva para la vacunación contra SARS-CoV-2 en la población mexicana. Recomendaciones preliminares. Salud. Publica Mex (2020.0) 63 288-309. DOI: 10.21149/12399 9. Mongua-Rodriguez N., Rodriguez-Alvarez M., De-la-Rosa-Zamboni D., Jimenez-Corona M.E., Castaneda-Cediel M.L., Miranda-Novales G., Cruz-Pacheco G., Ferreira-Guerrero E., Ferreyra-Reyes L., Delgado-Sanchez G.. Knowledge, attitudes, perceptions, and COVID-19 hesitancy in a large public university in Mexico city during the early vaccination rollout. BMC Public Health (2022.0) 22. DOI: 10.1186/s12889-022-14225-2 10. Vacunas COVID-19 Autorizadas 11. Vacunas Contra COVID-19 12. He X., Su J., Ma Y., Zhang W., Tang S.. A comprehensive analysis of the efficacy and effectiveness of COVID-19 vaccines. Front. Immunol. (2022.0) 13 945930. DOI: 10.3389/fimmu.2022.945930 13. Feikin D.R., Higdon M.M., Abu-Raddad L.J., Andrews N., Araos R., Goldberg Y., Groome M.J., Huppert A., O’Brien K.L., Smith P.G.. Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: Results of a systematic review and meta-regression. Lancet (2022.0) 399 924-944. DOI: 10.1016/S0140-6736(22)00152-0 14. Walsh E.E., Frenck R.W., Falsey A.R., Kitchin N., Absalon J., Gurtman A., Lockhart S., Neuzil K., Mulligan M.J., Bailey R.. Safety and Immunogenicity of Two RNA-Based Covid-19 Vaccine Candidates. N. Engl. J. Med. (2020.0) 383 2439-2450. DOI: 10.1056/NEJMoa2027906 15. Taylor S.C., Hurst B., Martiszus I., Hausman M.S., Sarwat S., Schapiro J.M., Rowell S., Lituev A.. Semi-quantitative, high throughput analysis of SARS-CoV-2 neutralizing antibodies: Measuring the level and duration of immune response antibodies post infection/vaccination. Vaccine (2021.0) 39 5688-5698. DOI: 10.1016/j.vaccine.2021.07.098 16. Hernandez-Bello J., Morales-Nunez J.J., Machado-Sulbaran A.C., Diaz-Perez S.A., Torres-Hernandez P.C., Balcazar-Felix P., Gutierrez-Brito J.A., Lomeli-Nieto J.A., Munoz-Valle J.F.. Neutralizing Antibodies against SARS-CoV-2, Anti-Ad5 Antibodies, and Reactogenicity in Response to Ad5-nCoV (CanSino Biologics) Vaccine in Individuals with and without Prior SARS-CoV-2. Vaccines (2021.0) 9. DOI: 10.3390/vaccines9091047 17. Nam M., Seo J.D., Moon H.W., Kim H., Hur M., Yun Y.M.. Evaluation of Humoral Immune Response after SARS-CoV-2 Vaccination Using Two Binding Antibody Assays and a Neutralizing Antibody Assay. Microbiol. Spectr. (2021.0) 9 e0120221. DOI: 10.1128/Spectrum.01202-21 18. Liu J., Mao Q., Wu X., He Q., Bian L., Bai Y., Wang Z., Wang Q., Zhang J., Liang Z.. Considerations for the Feasibility of Neutralizing Antibodies as a Surrogate Endpoint for COVID-19 Vaccines. Front. Immunol. (2022.0) 13 814365. DOI: 10.3389/fimmu.2022.814365 19. Mexico Situation 20. Momentive.ai. Survey Monkey. Get Answers with Surveys 21. Al-Amer R., Maneze D., Everett B., Montayre J., Villarosa A.R., Dwekat E., Salamonson Y.. COVID-19 vaccination intention in the first year of the pandemic: A systematic review. J. Clin. Nurs. (2022.0) 31 62-86. DOI: 10.1111/jocn.15951 22. Taborda A.R., Murillo D.A., Moreno C.L., Taborda P.A.R., Fuquen M., Diaz P.A., Londono D.. Analysis of budgetary impact of COVID-19 vaccination in Latin AmericaAnalise do impacto orcamentario da vacinacao contra a COVID-19 na America Latina. Rev. Panam. Salud. Publica (2022.0) 46 e5. DOI: 10.26633/RPSP.2022.5 23. Au W.Y., Cheung P.P.. Effectiveness of heterologous and homologous covid-19 vaccine regimens: Living systematic review with network meta-analysis. BMJ (2022.0) 377 e069989. DOI: 10.1136/bmj-2022-069989 24. Chiu N.C., Chi H., Tu Y.K., Huang Y.N., Tai Y.L., Weng S.L., Chang L., Huang D.T., Huang F.Y., Lin C.Y.. To mix or not to mix? A rapid systematic review of heterologous prime-boost covid-19 vaccination. Expert Rev. Vaccines (2021.0) 20 1211-1220. DOI: 10.1080/14760584.2021.1971522 25. Voysey M., Costa Clemens S.A., Madhi S.A., Weckx L.Y., Folegatti P.M., Aley P.K., Angus B., Baillie V.L., Barnabas S.L., Bhorat Q.E.. Single-dose administration and the influence of the timing of the booster dose on immunogenicity and efficacy of ChAdOx1 nCoV-19 (AZD1222) vaccine: A pooled analysis of four randomised trials. Lancet (2021.0) 397 881-891. DOI: 10.1016/S0140-6736(21)00432-3 26. Pormohammad A., Zarei M., Ghorbani S., Mohammadi M., Razizadeh M.H., Turner D.L., Turner R.J.. Efficacy and Safety of COVID-19 Vaccines: A Systematic Review and Meta-Analysis of Randomized Clinical Trials. Vaccines (2021.0) 9. DOI: 10.3390/vaccines9050467 27. Thomas S.J., Moreira E.D., Kitchin N., Absalon J., Gurtman A., Lockhart S., Perez J.L., Perez Marc G., Polack F.P., Zerbini C.. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine through 6 Months. N. Engl. J. Med. (2021.0) 385 1761-1773. DOI: 10.1056/NEJMoa2110345 28. Sahin U., Muik A., Vogler I., Derhovanessian E., Kranz L.M., Vormehr M., Quandt J., Bidmon N., Ulges A., Baum A.. BNT162b2 vaccine induces neutralizing antibodies and poly-specific T cells in humans. Nature (2021.0) 595 572-577. DOI: 10.1038/s41586-021-03653-6 29. Favresse J., Gillot C., Di Chiaro L., Eucher C., Elsen M., Van Eeckhoudt S., David C., Morimont L., Dogne J.M., Douxfils J.. Neutralizing Antibodies in COVID-19 Patients and Vaccine Recipients after Two Doses of BNT162b2. Viruses (2021.0) 13. DOI: 10.3390/v13071364 30. Halperin S.A., Ye L., MacKinnon-Cameron D., Smith B., Cahn P.E., Ruiz-Palacios G.M., Ikram A., Lanas F., Lourdes Guerrero M., Munoz Navarro S.R.. Final efficacy analysis, interim safety analysis, and immunogenicity of a single dose of recombinant novel coronavirus vaccine (adenovirus type 5 vector) in adults 18 years and older: An international, multicentre, randomised, double-blinded, placebo-controlled phase 3 trial. Lancet (2022.0) 399 237-248. DOI: 10.1016/S0140-6736(21)02753-7 31. Levin E.G., Lustig Y., Cohen C., Fluss R., Indenbaum V., Amit S., Doolman R., Asraf K., Mendelson E., Ziv A.. Waning Immune Humoral Response to BNT162b2 Covid-19 Vaccine over 6 Months. N. Engl. J. Med. (2021.0) 385 e84. DOI: 10.1056/NEJMoa2114583 32. Klein S.L., Flanagan K.L.. Sex differences in immune responses. Nat. Rev. Immunol. (2016.0) 16 626-638. DOI: 10.1038/nri.2016.90 33. Faizo A.A., Qashqari F.S., El-Kafrawy S.A., Barasheed O., Almashjary M.N., Alfelali M., Bawazir A.A., Albarakati B.M., Khayyat S.A., Hassan A.M.. A potential association between obesity and reduced effectiveness of COVID-19 vaccine-induced neutralizing humoral immunity. J. Med. Virol. (2022.0) 95 e28130. DOI: 10.1002/jmv.28130 34. Global Obesity Observatory/Mexico 35. Krutikov M., Palmer T., Tut G., Fuller C., Azmi B., Giddings R., Shrotri M., Kaur N., Sylla P., Lancaster T.. Prevalence and duration of detectable SARS-CoV-2 nucleocapsid antibodies in staff and residents of long-term care facilities over the first year of the pandemic (VIVALDI study): Prospective cohort study in England. Lancet Healthy Longev. (2022.0) 3 e13-e21. DOI: 10.1016/S2666-7568(21)00282-8 36. Krammer F., Srivastava K., Alshammary H., Amoako A.A., Awawda M.H., Beach K.F., Bermudez-Gonzalez M.C., Bielak D.A., Carreno J.M., Chernet R.L.. Antibody Responses in Seropositive Persons after a Single Dose of SARS-CoV-2 mRNA Vaccine. N. Engl. J. Med. (2021.0) 384 1372-1374. DOI: 10.1056/NEJMc2101667 37. Ontanon J., Blas J., de Cabo C., Santos C., Ruiz-Escribano E., Garcia A., Marin L., Saez L., Beato J.L., Rada R.. Influence of past infection with SARS-CoV-2 on the response to the BNT162b2 mRNA vaccine in health care workers: Kinetics and durability of the humoral immune response. EBioMedicine (2021.0) 73 103656. DOI: 10.1016/j.ebiom.2021.103656 38. Polack F.P., Thomas S.J., Kitchin N., Absalon J., Gurtman A., Lockhart S., Perez J.L., Perez Marc G., Moreira E.D., Zerbini C.. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N. Engl. J. Med. (2020.0) 383 2603-2615. DOI: 10.1056/NEJMoa2034577 39. Frenck R.W., Klein N.P., Kitchin N., Gurtman A., Absalon J., Lockhart S., Perez J.L., Walter E.B., Senders S., Bailey R.. Safety, Immunogenicity, and Efficacy of the BNT162b2 Covid-19 Vaccine in Adolescents. N. Engl. J. Med. (2021.0) 385 239-250. DOI: 10.1056/NEJMoa2107456 40. Kaur R.J., Dutta S., Bhardwaj P., Charan J., Dhingra S., Mitra P., Singh K., Yadav D., Sharma P., Misra S.. Adverse Events Reported From COVID-19 Vaccine Trials: A Systematic Review. Indian J. Clin. Biochem. (2021.0) 36 427-439. DOI: 10.1007/s12291-021-00968-z 41. Ghasemiyeh P., Mohammadi-Samani S., Firouzabadi N., Dehshahri A., Vazin A.. A focused review on technologies, mechanisms, safety, and efficacy of available COVID-19 vaccines. Int. Immunopharmacol. (2021.0) 100 108162. DOI: 10.1016/j.intimp.2021.108162 42. Galipeau Y., Greig M., Liu G., Driedger M., Langlois M.A.. Humoral Responses and Serological Assays in SARS-CoV-2 Infections. Front. Immunol. (2020.0) 11 610688. DOI: 10.3389/fimmu.2020.610688 43. Meyer B., Reimerink J., Torriani G., Brouwer F., Godeke G.J., Yerly S., Hoogerwerf M., Vuilleumier N., Kaiser L., Eckerle I.. Validation and clinical evaluation of a SARS-CoV-2 surrogate virus neutralisation test (sVNT). Emerg. Microbes. Infect. (2020.0) 9 2394-2403. DOI: 10.1080/22221751.2020.1835448
--- title: Extracellular Lactic Acidosis of the Tumor Microenvironment Drives Adipocyte-to-Myofibroblast Transition Fueling the Generation of Cancer-Associated Fibroblasts authors: - Elena Andreucci - Bianca Saveria Fioretto - Irene Rosa - Marco Matucci-Cerinic - Alessio Biagioni - Eloisa Romano - Lido Calorini - Mirko Manetti journal: Cells year: 2023 pmcid: PMC10046917 doi: 10.3390/cells12060939 license: CC BY 4.0 --- # Extracellular Lactic Acidosis of the Tumor Microenvironment Drives Adipocyte-to-Myofibroblast Transition Fueling the Generation of Cancer-Associated Fibroblasts ## Abstract Lactic acidosis characterizes the tumor microenvironment (TME) and is involved in the mechanisms leading to cancer progression and dissemination through the reprogramming of tumor and local host cells (e.g., endothelial cells, fibroblasts, and immune cells). Adipose tissue also represents a crucial component of the TME which is receiving increasing attention due to its pro-tumoral activity, however, to date, it is not known whether it could be affected by the acidic TME. Now, emerging evidence from chronic inflammatory and fibrotic diseases underlines that adipocytes may give rise to pathogenic myofibroblast-like cells through the adipocyte-to-myofibroblast transition (AMT). Thus, our study aimed to investigate whether extracellular acidosis could affect the AMT process, sustaining the acquisition by adipocytes of a cancer-associated fibroblast (CAF)-like phenotype with a pro-tumoral activity. To this purpose, human subcutaneous adipose-derived stem cells committed to adipocytes (acADSCs) were cultured under basal (pH 7.4) or lactic acidic (pH 6.7, 10 mM lactate) conditions, and AMT was evaluated with quantitative PCR, immunoblotting, and immunofluorescence analyses. We observed that lactic acidosis significantly impaired the expression of adipocytic markers while inducing myofibroblastic, pro-fibrotic, and pro-inflammatory phenotypes in acADSCs, which are characteristic of AMT reprogramming. Interestingly, the conditioned medium of lactic acidosis-exposed acADSC cultures was able to induce myofibroblastic activation in normal fibroblasts and sustain the proliferation, migration, invasion, and therapy resistance of breast cancer cells in vitro. This study reveals a previously unrecognized relationship between lactic acidosis and the generation of a new CAF-like cell subpopulation from adipocytic precursor cells sustaining tumor malignancy. ## 1. Introduction Extracellular acidosis characterizes the microenvironment of most solid tumors which generally show a pH ranging from 6.4 to 7.0 [1]. The acidification of the tumor microenvironment (TME) is indeed the direct consequence of the high glycolytic metabolism of cancer cells, which prefer glycolysis instead of phosphorylative oxidation even in presence of oxygen (Warburg effect), with a final advantage in cell proliferation but accompanied by an overproduction and subsequent release of lactic acid in the extracellular milieu [2,3]. Besides that, the impaired lymphatic circulation and the high interstitial pressure typical of solid cancer tissues further exacerbate the phenomenon [4]. The acidic TME has been recognized as a hallmark of cancer and a crucial contributor to tumor progression. Notably, extracellular acidosis has been demonstrated to reprogram both tumor and host stromal cellular components toward disease advancement, actively participating in every step of cancer dissemination [5,6,7]. Much evidence to date has revealed the effects of the acidic TME on cancer cells, which, once adapted to such environmental conditions, gain an extremely plastic phenotype endowed with an increased ability to invade the surrounding tissues, enter into and survive in the blood/lymphatic circulation (anoikis resistance), and evade immune surveillance systems, as well as resist chemotherapy, radiation therapy, immunotherapy, and molecularly targeted therapies, overall accounting for an increased metastatic potential [5,8,9,10]. The acidic TME exerts pro-tumoral effects by also acting on non-tumor cellular components, either within or surrounding the tumor mass, or even at distant sites—in a sort of paracrine/endocrine way, involving, for instance, extracellular vesicles that can travel through all body fluids, conveying oncogenic bioactive molecules and potentially preparing every tissue for metastatic colonization [11]. Indeed, the ability of extracellular acidosis to drive the pro-tumoral reprogramming of fibroblasts [12,13,14], endothelial cells [15,16], and immune cells [17,18,19,20] has been extensively demonstrated. To date, no evidence has been collected instead on the possible effects of extracellular acidosis on the adipocytic cell compartment and their possible switching through the adipocyte-to-myofibroblast transition (AMT). Notably, very recent findings underline the importance of AMT, a process by which pre-adipocytes/adipocytes transdifferentiate into pathogenic myofibroblast-like cells, observed in systemic sclerosis and other fibrotic diseases, where it was deemed as relevant in the formation of the pro-fibrotic myofibroblasts responsible for excessive extracellular matrix synthesis/deposition and remodeling [21,22,23,24]. Adipocytes and, more precisely, cancer-associated adipocytes (CAAs) are then being extensively investigated, and evidence supports their active contribution to cancer progression, especially in mammary tumors [20,25,26,27,28,29], where they represent $90\%$ of the breast tissue [30] and were found to give rise to the generation of cancer-associated fibroblast (CAF)-like cells with pro-tumoral activity [31]. To date, it has never been explored whether extracellular lactic acidosis might favor AMT. Here, indeed, we demonstrated for the first time that lactic acidosis promotes the AMT process, resulting in the generation of myofibroblast/CAF-like cells that in turn sustain proliferation, migration, and invasion, as well as the therapy resistance of breast cancer cells. ## 2.1. Cell Culture and Treatment Three lines of normal human subcutaneous adipose-derived stem cells (ADSCs) purchased from Lonza (catalog no. PT-5006; Lonza, Basel, Switzerland) were maintained in ADSC complete medium (ADSC Growth Medium BulletKit; catalog no. PT-4505; Lonza) at 37 °C in a $5\%$ CO2 incubator. To commit ADSCs to adipocytes, cells at low passage (passages 2–4) and at $90\%$ confluence were seeded in Petri dishes and incubated for 10 days in a preadipocyte differentiation medium (PGM-2 Preadipocyte Growth Medium-2 BulletKit; catalog no. PT-8002; Lonza) containing PBM-2 basal medium (catalog no. PT-8202; Lonza) and PGM-2 SingleQuots supplements (catalog no. PT-9502; Lonza). The obtained adipocyte-committed ADSCs (acADSCs) were then cultured for another 3 days under different conditions to obtain three different experimental points, namely basal acADSCs, acidic acADSCs, and transforming growth factor β1 (TGFβ1)-treated acADSCs. Basal acADSCs were cultured in PBM-2 basal medium with $2\%$ fetal bovine serum (FBS) at pH 7.4, acidic acADSCs were subjected to chemically induced acidosis by the direct administration of 1 N hydrogen chloride (HCl) in PBM-2 medium with $2\%$ FBS to reach pH 6.7 in the presence of 10 mM lactate (Sigma-Aldrich, Milan, Italy), while TGFβ1-treated acADSCs were incubated in PBM-2 medium with $2\%$ FBS (pH 7.4) and 10 ng/mL recombinant human TGFβ1 (PeproTech, Rocky Hill, NJ, USA). pH 6.7 was chosen as the representative acidic condition following preliminary experiments where a pH range from 7.4 to 6.4 was assessed in our cell culture model. pH was monitored using an Orion 520A1 pH meter (Thermo Fisher Scientific, Waltham, MA, USA). An additional experimental condition of acADSCs grown for 3 days in the basal medium at pH 7.4 in the presence of 10 mM lactate was used as an internal control in the initial experiments on AMT. Phase-contrast photomicrographs of acADSCs were captured under a Leica inverted microscope (Leica Microsystems, Mannheim, Germany). Cells with intracytoplasmic lipid droplets were counted in 10 randomly chosen microscopic high-power fields (hpf; 40× original magnification) per sample by two independent observers (B.S.F. and M.M.) who were blinded with regard to the sample classification. The final result was the mean of the two different observations for each sample. For selected experiments, conditioned media from basal and acidic acADSCs (basal acADSC-cm and acidic acADSC-cm, respectively) were collected after maintaining cells in fresh PBM-2 basal medium supplemented with $2\%$ FBS for an additional 3 days. Three lines of normal human dermal fibroblasts (NHDFs) purchased from Sigma-Aldrich (catalog no. C-12302) and cultured in Dulbecco’s modified Eagle medium (DMEM), 4.5 g/L glucose (Euroclone, Milan, Italy) supplemented with 2 mM L-glutamine and $10\%$ FBS were used for the experimental procedures between the third and seventh passages. MCF7 and MDA-MB-231 breast cancer cells were obtained from American Type Culture Collection (ATCC) and cultured in DMEM 4.5 g/L glucose supplemented with $10\%$ FBS. In selected experiments, MCF7 and MDA-MB-231 were treated for 72 h with 0.1 µM and 1 µM doxorubicin (MedChemExpress, Sollentuna, Sweden), respectively. ## 2.2. Annexin V/Propidium Iodide Flow Cytometer Assay ADSCs were seeded into 6-well plates until $90\%$ confluence, committed to adipocyte differentiation for 10 days, and subsequently grown for 72 h under basal (pH 7.4) or lactic acidic (pH 6.7, 10 mM lactate) conditions. Culture media were collected and acADSCs were harvested with Accutase (Euroclone), and subsequently collected in flow cytometer tubes. After 5 min centrifugation at 300× g, cell pellets were incubated for 15 min at 4 °C in the dark with 100 μL of annexin binding buffer (100 mM HEPES, 140 mM NaCl, 25 mM CaCl2, pH 7.4), 1 μL of 100 μg/mL propidium iodide (PI; Sigma-Aldrich) working solution, and 3 μL annexin V APC-conjugated (Immunotools, Friesoythe, Germany). Each sample was added with annexin binding buffer to reach 500 μL volume/tube. Samples were then analyzed at BD FACSCanto II (BD Biosciences, Milan, Italy). Cellular distribution depending on annexin V and/or PI positivity allowed the measurement of the percentage of viable cells (annexin V− and PI−), early apoptosis (annexin V+ and PI−), late apoptosis (annexin V+ and PI+), and necrosis (annexin V− and PI+). A minimum of 10,000 events were collected. ## 2.3. MTT Assay After 1 × 104 ADSCs were seeded into a 96-well plate and committed to adipocyte differentiation for 10 days, the obtained acADSCs were grown for 72 h under basal (pH 7.4) or lactic acidic (pH 6.7, 10 mM lactate) conditions. Cells were subsequently incubated for 2 h at 37 °C in the dark with 0.5 mg/mL MTT-containing medium without phenol red. MTT was removed and cells were lysed in 100 µL dimethyl sulfoxide (DMSO). Absorbance values were recorded at 595 nm with an automatic plate reader (Bio-Rad, Hercules, CA, USA). Similarly, 0.5 × 104 MCF7 or MDA-MB-231 cells were seeded into a 96-well plate and grown in basal acADSC-cm or acidic acADSC-cm for 72 h or 96 h as specified in the results section and figure legends. Following the 2 h incubation at 37 °C in 0.5 mg/mL MTT-containing medium in the absence of phenol red, cells were lysed in 100 µL DMSO, and 595 nm absorbance values were measured with the automatic plate reader. ## 2.4. RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction (qPCR) Total RNA was prepared from acADSCs using Tri Reagent (Merck Life Science, Milan, Italy), agarose gel checked for integrity, and quantified with the NanoDrop 8000 Spectrophotometer (Thermo Fisher Scientific). In selected experiments, RNA was also purified from NHDFs cultured in basal acADSC-cm or acidic acADSC-cm for 24 h. Reverse transcription was performed with the iScript cDNA Synthesis Kit (Bio-Rad, Milan, Italy) according to the manufacturer’s instructions. For gene expression quantification, SYBR Green Real-Time PCR was performed using the StepOnePlus Real-Time PCR System (Applied Biosystems, Milan, Italy) with melting curve analysis. Predesigned oligonucleotide primer pairs were employed (QuantiTect Primer Assay; Qiagen, Hilden, Germany). The assay IDs are shown in Table 1. The PCR mixture was composed of 1 μL cDNA, 0.5 μM sense, and antisense primers, 10 μL 2× QuantiTect SYBR Green PCR Master Mix containing SYBR Green I dye, ROX passive reference dye, HotStarTaq DNA Polymerase, dNTP mix, and MgCl2 (Qiagen). Amplification was performed according to a standard protocol recommended by the manufacturer. Non-specific signals produced by primer dimers or genomic DNA were excluded by dissociation curve analysis, non-template controls, and samples without enzyme in the reverse transcription step. In all samples, 18S ribosomal RNA (Hs_RRN18S_1_SG; catalog no. QT00199367; Qiagen) was measured as an endogenous control to normalize the amounts of loaded cDNA. Differences were calculated with the threshold cycle (Ct) and comparative Ct method for relative quantification. ## 2.5. Western Blotting acADSCs, grown for 3 days under basal (pH 7.4) or lactic acidic (pH 6.7, 10 mM lactate) conditions were lysed in radioimmunoprecipitation assay (RIPA) lysis buffer (Millipore by Sigma-Aldrich) added with Pierce Protease Inhibitor Tablets (Thermo Fisher Scientific) for protein isolation. The protein concentration was measured with Bradford reagent (Millipore by Sigma-Aldrich). After the addition of the Laemmli sample buffer (Bio-Rad) and β-mercaptoethanol, equal amounts of proteins were boiled at 100 °C for 5 min, electrophoresed on precast polyacrylamide gels (4-$15\%$ Mini-Protean TGX Gels; Bio-Rad), and blotted onto nitrocellulose membranes (Bio-Rad). The membranes were blocked for 30 min at room temperature in $5\%$ w/v milk buffer ($5\%$ w/v non-fat dried milk, 50 mM Tris, 200 mM NaCl, $0.2\%$ Tween-20) and subsequently incubated overnight at 4 °C with the following primary antibodies diluted in $5\%$ w/v milk buffer: rabbit monoclonal anti-fatty acid-binding protein 4 (FABP4) (1:1000 dilution; catalog no. ab92501; Abcam, Cambridge, UK), mouse monoclonal anti-adiponectin (1:1000 dilution; catalog no. ab22554; Abcam), rabbit polyclonal anti-perilipin-1 (1:500 dilution; catalog no. ab3526; Abcam), mouse monoclonal anti-α-smooth muscle actin (α-SMA) (1:300 dilution; catalog no. ab7817; Abcam), rabbit monoclonal anti-COL1A1 (1:1000 dilution; catalog no. # 39952; Cell Signaling Technology, Danvers, MA, USA), and goat polyclonal anti-GPR81 (1:1000 dilution; catalog no. ab106942; Abcam). Mouse monoclonal anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (1:5000 dilution; catalog no. ab8245; Abcam) and rabbit polyclonal anti-α-actinin (1:1000 dilution; catalog no. # 3134; Cell Signaling Technology) antibodies were used for normalization. The immunoblots were washed three times in Tris-buffered saline (Bio-Rad) with $0.1\%$ Tween-20, and then incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (Cell Signaling Technology). The proteins were visualized by an enhanced chemiluminescence method (Clarity Western ECL Substrate; Bio-Rad) and analyzed by ChemiDoc Touch Imaging System (Bio-Rad). Band intensities were quantified using the free-share ImageJ software (NIH, Bethesda, MD, USA; online at http://rsbweb.nih.gov/ij, accessed on 23 May 2022) and values were normalized to GAPDH or α-actinin, as needed. ## 2.6. Immunofluorescence For immunofluorescence microscopy, 2.5 × 105 ADSCs were seeded onto glass coverslips in 6-well plates and committed to adipocyte differentiation for 10 days. The resulting acADSCs were grown for 72 h under basal (pH 7.4) or lactic acidosis (pH 6.7, 10 mM lactate) conditions. Cells were then fixed for 30 min at 4 °C with $3.7\%$ buffered paraformaldehyde and permeabilized for 15 min with PBS $0.1\%$ Triton X-100 at room temperature. Slides were then blocked with $1\%$ bovine serum albumin in PBS for 1 h at room temperature, and finally incubated overnight at 4 °C with the following primary antibodies: rabbit polyclonal anti-perilipin-1 (1:80 dilution; catalog no. ab3526; Abcam), mouse monoclonal anti-α-SMA (1:100 dilution; catalog no. ab7817; Abcam), mouse monoclonal anti-adiponectin (1:100 dilution; catalog no. ab22554; Abcam), and rabbit monoclonal anti-COL1A1 (1:300 dilution; catalog no. # 39952; Cell Signaling Technology). Irrelevant isotype-matched and concentration-matched mouse and rabbit IgG (Sigma-Aldrich) were used as negative controls. The day after, cells were incubated for 45 min at room temperature in the dark with Alexa Fluor-488-conjugated and Rhodamine Red-X-conjugated secondary antibodies at 1:200 dilution (Invitrogen, Carlsbad, CA, USA). Following nuclei staining with 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific) for 10 min at room temperature in the dark, the immunostained cells were mounted onto glass slides and visualized at the Leica DM4000 B microscope (Leica Microsystems). Fluorescence images were captured with a Leica DFC310 FX 1.4-megapixel digital color camera equipped with the Leica software application suite LAS V3.8 (Leica Microsystems). Immunostained cells were counted in 10 randomly chosen microscopic hpf (40× original magnification) per sample. Only the cells with well-defined DAPI-positive nuclei were counted. Counting was performed by two independent observers (I.R. and M.M.) who were blinded with regard to the sample classification. The final result was the mean of the two different observations for each sample. ## 2.7. Colony Assay The 0.2 × 103 MCF7 or MDA-MB-231 cells were seeded into 6-well plates and allowed to grow in the presence of the basal acADSC-cm or acidic acADSC-cm for 10 days. Formed colonies were then fixed in $3.7\%$ paraformaldehyde and stained with a crystal violet solution. ## 2.8. Migration and Invasion Assays Migration and invasion assays on MCF7 and MDA-MB-231 cells were performed following a 24 h treatment with basal acADSC-cm or acidic acADSC-cm. For the migration assay, 12 mm diameter Millicell cell culture inserts with 8 µm diameter-pore polycarbonate filters (Sigma-Aldrich) were placed into 24-well plates, and 5 × 104 cancer cells were seeded in the upper compartment and allowed to migrate for 6 h without any FBS gradient toward the lower compartment. Following a 1 h fixation in methanol at 4 °C, non-invading cells on the upper side of the filters were mechanically wiped off with a cotton swab, while invasive cells on the lower side of the filters were stained with Diff-Quik dye (BD Biosciences). Cells were then visualized and counted using an optical microscope. For the invasion assay, an analog experimental procedure was performed with the difference that the polycarbonate filters of the inserts were pre-coated overnight with 0.25 µg/µL Matrigel (Corning by Sigma-Aldrich) before cell seeding. ## 2.9. Statistical Analysis All data were obtained based on at least three independent experiments and analyzed with GraphPad Prism 8 software. After assessing the normality of data by the Kolmogorov–Smirnov test, statistical analysis between the two groups was performed using an unpaired Student’s t-test. In the case of the comparative analysis of three groups, one-way analysis of variance (ANOVA) was performed followed by Tukey’s post hoc test. Values are presented as the mean of independent experiments ± standard error of the mean (SEM). Values of $p \leq 0.05$ were considered statistically significant. ## 3.1. Lactic Acidosis Induces an Adipocyte-to-Myofibroblast Transition in Adipocyte-Committed Subcutaneous Adipose-Derived Stem Cells Preliminary experiments performed on ADSCs allowed us to choose pH 6.7, out of a pH range from 7.4 to 6.4, as the representative acidic condition for our experimental model, ensuring that cell viability under the conditions of pH 6.7 was not impaired (Supplementary Figure S1). By cultivating acADSCs under basal conditions (pH 7.4) or lactic acidosis (pH 6.7 + lactate), we observed that the acidic microenvironment induced deep morphologic changes in these cells. Indeed, compared to basal conditions, acADSCs grown in a lactic acidic medium showed a substantial loss of intracytoplasmic lipid droplets while displaying a mesenchymal-like elongated shape (Figure 1). This observation suggests that, under such acidic conditions, acADSCs undergo phenotypic remodeling toward a mesenchymal phenotype. The AMT process has already been identified in several fibrotic diseases as the mechanism by which pre-adipocytes/adipocytes transdifferentiate into myofibroblasts, and TGFβ1 was found to induce such a phenotypic switch [22,23,24]. Thereby, acADSCs cultured in standard pH in the presence of 10 ng/mL TGFβ1 represented a positive control of AMT. In fact, acADSCs treated with TGFβ1, compared to those grown under basal conditions, displayed a mesenchymal-like, elongated shape—even further marked compared to lactic acidosis—accompanied by a considerable decrease in intracytoplasmic lipid droplets. As shown in Figure 1, the number of cells containing cytoplasmic lipid droplets was significantly decreased in both acADSCs cultured under lactic acidosis and those stimulated with TGFβ1 compared to cultures maintained in basal (pH 7.4) medium. In contrast, we did not observe any significant variation in acADSC viability or proliferation when grown in basal conditions, lactic acidosis, or in the presence of TGFβ1 (Figure 2A–C). Indeed, annexin V/PI assay revealed that the percentage of viable cells at 72 h was approximately $90\%$ in all experimental conditions (Figure 2A,B). Besides viability, cell proliferation evaluated by MTT assay was also not altered, despite a slight but non-significant decreasing tendency observed in acADSCs subjected to lactic acidosis or TGFβ1 treatment (Figure 2C). The morphologic changes displayed by acADSCs subjected to the lactic acidic microenvironment were accompanied by the loss of the expression of typical adipogenic/adipocytic markers and a concomitant increase in the expression of myofibroblast markers (Figure 3A–H). Quantitative real-time PCR analysis revealed that the mRNA expression of the adipocyte-related genes FABP4, CEBPA (i.e., gene encoding C/EBPα), PPARG (i.e., gene encoding PPARγ), and PLIN1 (i.e., gene-encoding perilipin-1) was almost halved in acADSCs subjected to the lactic acidic microenvironment, while a further decrease was observed for ADIPOQ (i.e., gene encoding adiponectin) mRNA expression levels (Figure 3A–E). On the contrary, lactic acidosis increased the expression levels of typical myofibroblast-related genes, such as ACTA2 (i.e., gene encoding α-SMA), COL1A1, and COL1A2, which were 2.5-fold, 3-fold, and 4-fold increased, respectively, compared to the basal conditions (Figure 3F–H). As expected, the treatment with TGFβ1, compared to the basal condition, significantly reduced the expression levels of all the adipocyte-related genes tested and concomitantly boosted the expression of myofibroblastic genes (Figure 3A–H). The expression levels of the 18S rRNA reference gene were stable under the three different experimental conditions (Figure 3I). The data obtained in real-time PCR were validated with Western blot analysis, which confirmed the trend of the expression of adipocyte/myofibroblast markers (Figure 4A–E). Briefly, we observed a significant decrease in FABP4, adiponectin, and perilipin-1 protein expression both in acADSCs grown under lactic acidosis and in those treated with TGFβ1 compared to basal conditions (Figure 4A–C). On the contrary, the protein expression levels of the myofibroblastic markers α-SMA and COL1A1 were significantly increased by approximately 2-fold in acADSCs cultured in a lactic acidic medium, as well as in the presence of TGFβ1, compared to those grown under basal conditions (Figure 4D,E). The immunofluorescence analysis further strengthened the evidence of the AMT process that the acADSCs underwent upon exposure to an acidic microenvironment (Figure 5). Indeed, we observed a prominent loss of perilipin-1-coated cytoplasmic lipid droplets and the accumulation of intracellular COL1A1 and expression of α-SMA-positive stress fibers characteristic of myofibroblasts in acidic cell cultures, which was similar to that found in cultures stimulated with TGFβ1 (Figure 5). In particular, the percentages of perilipin-1-positive and adiponectin-positive cells/hpf were approximately three times decreased in acADSCs grown under lactic acidosis, as well as following TGFβ1 treatment, compared to those maintained in basal (pH 7.4) medium (Figure 5). Under the same experimental conditions (i.e., lactic acidosis or TGFβ1), the percentages of α-SMA-positive and COL1A1-positive cells/hpf were significantly increased compared to those under basal conditions (Figure 5). Of note, acADSCs grown for 3 days in a pH 7.4 medium in the presence of 10 mM lactate did not show any relevant change in cell morphology, as well as in the expression of perilipin-1, adiponectin, α-SMA, and COL1A1 compared to the cells grown under basal conditions (Supplementary Figure S2A,B). Therefore, the exposure of acADSCs to lactate alone, without the acidification of the culture medium, was not able to induce the AMT process. As far as the expression of the lactate receptor GPR81 is concerned, no difference was detected between basal acADSCs and those grown under lactic acidosis or at pH 7.4 in the presence of 10 mM lactate (Supplementary Figure S3). ## 3.2. Lactic Acidosis Induces an Adipocyte-to-Myofibroblast Transition in Adipocyte-Committed Subcutaneous Adipose-Derived Stem Cells Characterized by a Pro-inflammatory Phenotype We next observed that the exposure to lactic acidic conditions in vitro induced the acquisition of a pro-inflammatory and pro-fibrotic phenotype by acADSCs. Indeed, acADSCs cultured in a pH 6.7 medium in the presence of 10 mM lactate showed a 1.5-fold increase in mRNA expression of IL1B and IL6 genes (i.e., genes encoding the pro-inflammatory cytokines interleukin (IL)-1β and IL-6, respectively) compared to those maintained under basal (pH 7.4) conditions (Figure 6A,B). The acADSC treatment with 10 ng/mL TGFβ1 was able to stimulate IL1B and IL6 gene expression at a similar level to lactic acidosis (Figure 6A,B). In line with the above-described acADSC switch toward a myofibroblastic-like phenotype under lactic acidosis, we observed that acADSCs exposed to pH 6.7 in the presence of 10 mM lactate showed a modest but significant increase in the mRNA expression of TGFB1 (i.e., gene encoding TGFβ1) compared to acADSCs grown under basal conditions (Figure 6C). A similar trend—even though not significant—was observed for the mRNA expression of TGFB2 (i.e., gene encoding TGFβ2) (Figure 6D). ## 3.3. Pro-Tumoral Activity of Adipocyte-Committed Subcutaneous Adipose-Derived Stem Cells Exposed to the Acidic Microenvironment Based on the latter findings, we decided to verify the ability of the acidic acADSCs to promote the reprogramming of naïve stromal cells toward a CAF-like phenotype. At first, we evaluated the effects of the conditioned media collected from acADSCs grown either under basal (pH 7.4) conditions or under lactic acidosis (basal acADSC-cm or acidic acADSC-cm, respectively) on NHDFs. We observed that NHDFs increased the mRNA expression of FAP (i.e., gene encoding fibroblast activation protein), ACTA2, COL1A1, and COL1A2 genes by approximately 1.5 times when treated with acidic acADSC-cm compared to basal acADSC-cm (Figure 7). These data suggest that acADSCs exposed to a lactic acidic microenvironment, following the AMT process, could be responsible for the activation of quiescent fibroblasts toward a CAF-like phenotype, thereby enabling the hypothesis of their possible involvement in the pro-tumor sustainment. Next, we further investigated the pathogenic effects of acidic acADSC-cm, by focusing the attention at this point on breast cancer cells, particularly on the estrogen receptor (ER)-positive MCF7 and the triple-negative MDA-MB-231 cells. Notably, both MCF7 and MDA-MB-231 cells showed a boosted colony formation ability (Figure 8A) and an increased proliferation (Figure 8B) when exposed to acidic acADSC-cm compared to basal acADSC-cm, suggesting the release of pro-tumoral soluble factors by acidic acADSCs in their conditioned media. Moreover, both the migratory and invasive abilities of MCF7 and MDA-MB-231 cells were significantly increased following the 24 h treatment with acidic acADSC-cm compared to basal acADSC-cm (Figure 9A,B). More precisely, by treating cancer cells with acidic acADSC-cm, we observed an approximately 1.3-fold increase in the number of MCF7 and MDA-MB-231 cells able to migrate (Figure 9A). In parallel, we detected a 1.3-fold increase in the number of invasive MCF7 cells when treated with the acidic acADSC-cm compared to basal acADSC-cm (Figure 9B). Such an increase was even further evident in the MDA-MB-231 cell line, where the number of invading cells was 1.8 times higher following acidic the acADSC-cm treatment than the basal acADSC-cm (Figure 9B). Finally, by treating cancer cells with doxorubicin, we observed a slight but significant increase in cell survival, revealed by the higher rate of proliferation of both MCF7 and MDA-MB-231 cells exposed to a conditioned medium collected from the acidosis-exposed acADSCs compared to that of acADSCs grown in a standard medium (Figure 10). Overall, these findings suggest a potential pro-tumoral effect of acADSCs exposed to an acidic microenvironment to promote the CAF differentiation of standard fibroblasts and support tumor cell proliferation, invasion, and drug resistance. ## 4. Discussion The TME is a heterogeneous ecosystem composed of multiple cell types including the cells of the immune system, vasculature, and stromal cells, all reprogrammed in favor of tumor sustainment. Extracellular matrix and soluble factors produced by both tumor and host stromal cells are also important elements of the TME, as well as hypoxia and acidosis, which are able to alter and reprogram almost every cell and even non-cellular components of the TME in a pro-tumoral way [32]. Focusing on the biochemical aspects of the TME, to date, much evidence has been provided about the crucial role exerted by hypoxia and acidosis in tumor progression, which made increasingly evident the intense alliance between them and cancer cells, with the common goal of tumor advancement and disease progression [8]. Extracellular acidosis has recently been included in the hallmarks of cancer. Indeed, in the last two decades, the scientific community realized that it represents a peculiar trait of most solid tumors fostering aggressive features of cancer cells and disease progression. This microenvironmental condition arises from the combination of typical features of the tumor tissues, i.e., the boosted cancer glycolytic metabolism with the subsequent release of lactic acid in the extracellular milieu, the impaired drainage by the lymphatic system, and the high interstitial pressure characterizing cancer tissues [2,4]. Extensive literature demonstrates how extracellular lactic acidosis sustains tumor progression by acting either directly on cancer cells, or indirectly, by reprogramming the stromal component within the tumor mass to guarantee cancer cell proliferation and survival even under prohibitive and hostile conditions. Briefly, the acidic TME promotes, on the one hand, the acquisition by cancer cells of high plasticity that renders them extremely adaptable to various scenarios they may encounter [10]; on the other one, extracellular acidosis is able to induce a pro-tumoral reprogramming in the stromal cells, for instance, by altering the capacity of natural and adaptive immune cells to face with the tumor [5,7,17], or by remodeling the vasculature rendering vessels more permeant and subsequently permissive to cancer cell intra/extravasation [33], or further, by inducing the generation of CAFs with all the pro-tumoral activity they are endowed with [12]. What is still not known, is the effect that the acidic TME may exert on the adipocyte compartment. Here, we demonstrated for the first time that the acidic TME induces the AMT process in adipocytes, fueling the generation of CAF-like cells sustaining the aggressiveness of breast cancer cells. Briefly, we demonstrated that acADSCs grown under lactic acidosis conditions lose the adipocyte differentiation markers FABP4, C/EBPα, adiponectin, and perilipin-1 while acquiring the myofibroblast markers α-SMA, COL1A1, and COL1A2. Such a phenotypic switch driven by lactic acidosis was accompanied by a modest but significant increase in the mRNA expression of the pro-inflammatory cytokines IL-6, IL-1β, and TGFβ1, which in turn could account at least in part for the ability of acid-adapted acADSCs to induce the activation of normal fibroblasts. As recently reviewed [34,35], lactate can serve as a pro- or anti-inflammatory mediator, inducing pleiotropic effects on the inflammatory process. Indeed, lactate could exert differential effects depending not only on the pathological process studied, but also on the cellular metabolic status, and the cell type analyzed. For instance, lactate is pro-inflammatory in endothelial cells and fibroblasts, while it exerts both pro- and anti-inflammatory activity on immune cells depending on the immunophenotypes analyzed. Moreover, the complexity of this scenario increases considering that lactate and H+ (i.e., the single components of lactic acidosis) may also exert differential effects in the inflammatory microenvironment, as reviewed by Certo and colleagues [36]. Contextualizing our data in this setting, we did not wonder that lactic acidosis is able to stimulate the production by adipocytes of pro-inflammatory molecules that could account for the fibroblast activation and generation of CAFs with pro-tumoral activity. On the other hand, by reducing the anti-cancer immune response and inducing the pro-tumoral/anti-inflammatory M2 macrophage polarization, lactic acidosis seems to strongly favor cancer progression [36]. It is noteworthy that the conditioned medium of acADSCs exposed to the acidic condition was also able to induce breast cancer cell proliferation, migration, invasion, and drug resistance. Therefore, the identification of the main players of such a pro-tumoral secretome would be of extreme interest in order to develop new anti-cancer strategies. The peritumoral adipocytes undergo profound reprogramming toward CAAs, especially in those tumors that are closely related to adipose tissue, such as breast, colorectal, and ovarian cancers [37]. Particular attention is dedicated to breast cancer, where the adipose tissue represents the main component. CAAs, in contrast to mature adipocytes, are characterized by a loss of lipid droplets accompanied by a decreased expression of adipocyte differentiation markers [27,38]. Moreover, CAAs may mediate multiple paracrine, juxtacrine, and endocrine signaling, sustaining tumor growth, metastasis, and drug resistance. Then, their pro-tumoral functions may pass through their ability to reprogram cancer metabolism. The disaggregation process of the lipid droplets occurring in adipocyte-to-CAA transformation frees and renders exogenous free fatty acids and high-energy metabolites available to cancer cells, providing them with a precious energy source for their expansion [28,39]. In turn, tumor cells induce lipolysis in adipocytes and promote their transformation toward CAAs [40]. In line with such considerations, our data enable the speculation that the acidic TME may induce the AMT process in which adipocytes are deprived of lipid droplets and, consequently, free fatty acids are freed and made available to either cancer or stromal cells as an important energy source to sustain the overall tumor progression. Interestingly, even pro-tumoral M2 macrophages, induced by a lactic acid-enriched microenvironment, mainly rely on the uptake and subsequent oxidation of free fatty acids freed in the surrounding milieu [41,42]. Evidence already reported the existence of lipid transferring from peritumoral adipocytes to cancer cells, which in turn exploit this high-energetic metabolic source to sustain not only their growth but also to remodel the stromal components to be more permissive to cancer dissemination [43]. The secretome released by CAAs also sustains disease advancement. For instance, it was demonstrated that CCL2/monocyte chemoattractant protein-1, CCL5, IL-1β, IL-6, tumor necrosis factor α, vascular endothelial growth factor, and leptin released by CAAs in the TME promote the proliferation, and dissemination of breast cancer cells [29,44,45]. Furthermore, CAAs induce resistance to chemotherapy, radiation therapy, immunotherapy, and hormone therapy in breast cancer [27], and affect tumor extracellular matrix remodeling and adipose tissue vascularization, leading to the generation of hypoxic and fibrotic conditions that in turn mediate the epithelial-to-mesenchymal transition process in breast cancer cells [46]. Notably, recent evidence suggested that CAAs at the invasive front of the tumor mass undergo the AMT process consisting of the phenotypic switch of adipocytes toward myofibroblast/CAF-like cells endowed with all the pro-tumoral features typical of CAFs [31,47]. Thereby, the AMT, by fueling the CAF compartment, becomes a pro-tumoral process itself. ## 5. Conclusions Overall, this study highlights that extracellular acidosis may reprogram adipocytes toward pro-tumoral CAF-like cells, which are also able to recruit CAF from quiescent fibroblasts. As a result, the acidic TME could fuel the CAF compartment within the tumor mass, thereby promoting disease advancement. A limiting aspect of this study is that lactic acidosis has only been evaluated in vitro as a unique entity, without determining the single contributions that lactate and acidosis per se could exert in the AMT process. In particular, it would be interesting to further investigate the possible involvement in this phenomenon of lactate transporters (e.g., monocarboxylate transporters MCT1-4 and sodium-dependent transporters SMCT1 and SMCT2) and receptors (e.g., G protein-coupled receptors GPR81 and GPR132) together with the intracellular signaling triggered upon lactate binding and cellular uptake. Another important aspect that needs to be further investigated is whether the lactic acidosis-induced AMT could either be partially or fully reverted when basal pH conditions are restored. These points deserve a dedicated mechanistic study in which the molecular mechanisms leading to AMT need to be identified, to eventually reveal the potential therapeutic targets to counteract tumor progression. ## References 1. Reshetnyak Y.K., Yao L., Zheng S., Kuznetsov S., Engelman D.M., Andreev O.A.. Measuring Tumor Aggressiveness and Targeting Metastatic Lesions with Fluorescent PHLIP. Mol. Imaging Biol. (2011) 13 1146-1156. DOI: 10.1007/s11307-010-0457-z 2. Vaupel P., Multhoff G.. Revisiting the Warburg Effect: Historical Dogma. J. Physiol. (2021) 599 1745-1757. DOI: 10.1113/JP278810 3. Liu C., Jin Y., Fan Z.. The Mechanism of Warburg Effect-Induced Chemoresistance in Cancer. Front. Oncol. (2021) 11 698023. DOI: 10.3389/fonc.2021.698023 4. Vaupel P.. Tumor Microenvironmental Physiology and Its Implications for Radiation Oncology. Semin. Radiat. Oncol. (2004) 14 198-206. DOI: 10.1016/j.semradonc.2004.04.008 5. Corbet C., Feron O.. Tumour Acidosis: From the Passenger to the Driver’s Seat. Nat. Rev. Cancer (2017) 17 577-593. DOI: 10.1038/nrc.2017.77 6. De la Cruz-López K.G., Castro-Muñoz L.J., Reyes-Hernández D.O., García-Carrancá A., Manzo-Merino J.. Lactate in the Regulation of Tumor Microenvironment and Therapeutic Approaches. Front. Oncol. (2019) 9 1143. DOI: 10.3389/fonc.2019.01143 7. Fischbeck A.J., Ruehland S., Ettinger A., Paetzold K., Masouris I., Noessner E., Mendler A.N.. Tumor Lactic Acidosis: Protecting Tumor by Inhibiting Cytotoxic Activity Through Motility Arrest and Bioenergetic Silencing. Front. Oncol. (2020) 10 589434. DOI: 10.3389/fonc.2020.589434 8. Andreucci E., Peppicelli S., Ruzzolini J., Bianchini F., Calorini L.. Physicochemical Aspects of the Tumour Microenvironment as Drivers of Vasculogenic Mimicry. Cancer Metastasis Rev. (2022) 41 935-951. DOI: 10.1007/s10555-022-10067-x 9. Webb B.A., Chimenti M., Jacobson M.P., Barber D.L.. Dysregulated PH: A Perfect Storm for Cancer Progression. Nat. Rev. Cancer (2011) 11 671-677. DOI: 10.1038/nrc3110 10. Andreucci E., Peppicelli S., Ruzzolini J., Bianchini F., Biagioni A., Papucci L., Magnelli L., Mazzanti B., Stecca B., Calorini L.. The Acidic Tumor Microenvironment Drives a Stem-like Phenotype in Melanoma Cells. J. Mol. Med. (2020) 98 1431-1446. DOI: 10.1007/s00109-020-01959-y 11. Noren Hooten N., Yáñez-Mó M., DeRita R., Russell A., Quesenberry P., Ramratnam B., Robbins P.D., Di Vizio D., Wen S., Witwer K.W.. Hitting the Bullseye: Are Extracellular Vesicles on Target?. J. Extracell. Vesicles (2020) 10 e12032. DOI: 10.1002/jev2.12032 12. Peppicelli S., Bianchini F., Toti A., Laurenzana A., Fibbi G., Calorini L.. Extracellular Acidity Strengthens Mesenchymal Stem Cells to Promote Melanoma Progression. Cell Cycle (2015) 14 3088-3100. DOI: 10.1080/15384101.2015.1078032 13. Pavlides S., Whitaker-Menezes D., Castello-Cros R., Flomenberg N., Witkiewicz A.K., Frank P.G., Casimiro M.C., Wang C., Fortina P., Addya S.. The Reverse Warburg Effect: Aerobic Glycolysis in Cancer Associated Fibroblasts and the Tumor Stroma. Cell Cycle (2009) 8 3984-4001. DOI: 10.4161/cc.8.23.10238 14. Bonuccelli G., Avnet S., Grisendi G., Salerno M., Granchi D., Dominici M., Kusuzaki K., Baldini N.. Role of Mesenchymal Stem Cells in Osteosarcoma and Metabolic Reprogramming of Tumor Cells. Oncotarget (2014) 5 7575-7588. DOI: 10.18632/oncotarget.2243 15. Andreucci E., Margheri F., Peppicelli S., Bianchini F., Ruzzolini J., Laurenzana A., Fibbi G., Bruni C., Bellando-Randone S., Guiducci S.. Glycolysis-Derived Acidic Microenvironment as a Driver of Endothelial Dysfunction in Systemic Sclerosis. Rheumatology (2021) 60 4508-4519. DOI: 10.1093/rheumatology/keab022 16. García-Caballero M., Sokol L., Cuypers A., Carmeliet P.. Metabolic Reprogramming in Tumor Endothelial Cells. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms231911052 17. Erra Díaz F., Dantas E., Geffner J.. Unravelling the Interplay between Extracellular Acidosis and Immune Cells. Mediat. Inflamm. (2018) 2018 1218297. DOI: 10.1155/2018/1218297 18. Wu H., Estrella V., Beatty M., Abrahams D., El-Kenawi A., Russell S., Ibrahim-Hashim A., Longo D.L., Reshetnyak Y.K., Moshnikova A.. T-Cells Produce Acidic Niches in Lymph Nodes to Suppress Their Own Effector Functions. Nat. Commun. (2020) 11 4113. DOI: 10.1038/s41467-020-17756-7 19. Calcinotto A., Filipazzi P., Grioni M., Iero M., De Milito A., Ricupito A., Cova A., Canese R., Jachetti E., Rossetti M.. Modulation of Microenvironment Acidity Reverses Anergy in Human and Murine Tumor-Infiltrating T Lymphocytes. Cancer Res. (2012) 72 2746-2756. DOI: 10.1158/0008-5472.CAN-11-1272 20. Gottfried E., Kunz-Schughart L.A., Ebner S., Mueller-Klieser W., Hoves S., Andreesen R., Mackensen A., Kreutz M.. Tumor-Derived Lactic Acid Modulates Dendritic Cell Activation and Antigen Expression. Blood (2006) 107 2013-2021. DOI: 10.1182/blood-2005-05-1795 21. Onuora S.. Adipocyte–Myofibroblast Transition: Linking Intradermal Fat Loss to Skin Fibrosis in SSc. Nat. Rev. Rheumatol. (2015) 11 63. DOI: 10.1038/nrrheum.2014.223 22. Marangoni R.G., Korman B.D., Wei J., Wood T.A., Graham L.V., Whitfield M.L., Scherer P.E., Tourtellotte W.G., Varga J.. Myofibroblasts in Murine Cutaneous Fibrosis Originate From Adiponectin-Positive Intradermal Progenitors: Adipocyte-Myofibroblast Transition in Skin Fibrosis. Arthritis Rheumatol. (2015) 67 1062-1073. DOI: 10.1002/art.38990 23. Marangoni R.G., Korman B., Varga J.. Adipocytic Progenitor Cells Give Rise to Pathogenic Myofibroblasts: Adipocyte-to-Mesenchymal Transition and Its Emerging Role in Fibrosis in Multiple Organs. Curr. Rheumatol. Rep. (2020) 22 79. DOI: 10.1007/s11926-020-00957-w 24. Rosa I., Romano E., Fioretto B.S., Matucci-Cerinic M., Manetti M.. Adipose-Derived Stem Cells: Pathophysiologic Implications. World J. Stem Cells (2021) 13 30-48. DOI: 10.4252/wjsc.v13.i1.30 25. Mukherjee A., Bilecz A.J., Lengyel E.. The Adipocyte Microenvironment and Cancer. Cancer Metastasis Rev. (2022) 41 575-587. DOI: 10.1007/s10555-022-10059-x 26. Rybinska I., Mangano N., Tagliabue E., Triulzi T.. Cancer-Associated Adipocytes in Breast Cancer: Causes and Consequences. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22073775 27. Zhao C., Wu M., Zeng N., Xiong M., Hu W., Lv W., Yi Y., Zhang Q., Wu Y.. Cancer-Associated Adipocytes: Emerging Supporters in Breast Cancer. J. Exp. Clin. Cancer Res. (2020) 39 156. DOI: 10.1186/s13046-020-01666-z 28. Wu Q., Li B., Li Z., Li J., Sun S., Sun S.. Cancer-Associated Adipocytes: Key Players in Breast Cancer Progression. J. Hematol. Oncol. (2019) 12 95. DOI: 10.1186/s13045-019-0778-6 29. Lapeire L., Hendrix A., Lambein K., Van Bockstal M., Braems G., Van Den Broecke R., Limame R., Mestdagh P., Vandesompele J., Vanhove C.. Cancer-Associated Adipose Tissue Promotes Breast Cancer Progression by Paracrine Oncostatin M and Jak/STAT3 Signaling. Cancer Res. (2014) 74 6806-6819. DOI: 10.1158/0008-5472.CAN-14-0160 30. D’Esposito V., Ambrosio M.R., Giuliano M., Cabaro S., Miele C., Beguinot F., Formisano P.. Mammary Adipose Tissue Control of Breast Cancer Progression: Impact of Obesity and Diabetes. Front. Oncol. (2020) 10 1554. DOI: 10.3389/fonc.2020.01554 31. Zhu Q., Zhu Y., Hepler C., Zhang Q., Park J., Gliniak C., Henry G.H., Crewe C., Bu D., Zhang Z.. Adipocyte Mesenchymal Transition Contributes to Mammary Tumor Progression. Cell Rep. (2022) 40 111362. DOI: 10.1016/j.celrep.2022.111362 32. Neophytou C.M., Panagi M., Stylianopoulos T., Papageorgis P.. The Role of Tumor Microenvironment in Cancer Metastasis: Molecular Mechanisms and Therapeutic Opportunities. Cancers (2021) 13. DOI: 10.3390/cancers13092053 33. Andreucci E., Ruzzolini J., Bianchini F., Versienti G., Biagioni A., Lulli M., Guasti D., Nardini P., Serratì S., Margheri F.. MiR-214-Enriched Extracellular Vesicles Released by Acid-Adapted Melanoma Cells Promote Inflammatory Macrophage-Dependent Tumor Trans-Endothelial Migration. Cancers (2022) 14. DOI: 10.3390/cancers14205090 34. Baltazar F., Afonso J., Costa M., Granja S.. Lactate Beyond a Waste Metabolite: Metabolic Affairs and Signaling in Malignancy. Front. Oncol. (2020) 10 231. DOI: 10.3389/fonc.2020.00231 35. Manosalva C., Quiroga J., Hidalgo A.I., Alarcón P., Anseoleaga N., Hidalgo M.A., Burgos R.A.. Role of Lactate in Inflammatory Processes: Friend or Foe. Front. Immunol. (2022) 12 808799. DOI: 10.3389/fimmu.2021.808799 36. Certo M., Llibre A., Lee W., Mauro C.. Understanding Lactate Sensing and Signalling. Trends Endocrinol. Metab. (2022) 33 722-735. DOI: 10.1016/j.tem.2022.07.004 37. Yao H., He S.. Multi-faceted Role of Cancer-associated Adipocytes in the Tumor Microenvironment (Review). Mol. Med. Rep. (2021) 24 866. DOI: 10.3892/mmr.2021.12506 38. Choi J., Cha Y.J., Koo J.S.. Adipocyte Biology in Breast Cancer: From Silent Bystander to Active Facilitator. Prog. Lipid Res. (2018) 69 11-20. DOI: 10.1016/j.plipres.2017.11.002 39. Pérez-Escuredo J., Van Hée V.F., Sboarina M., Falces J., Payen V.L., Pellerin L., Sonveaux P.. Monocarboxylate Transporters in the Brain and in Cancer. Biochim. Biophys. Acta (2016) 1863 2481-2497. DOI: 10.1016/j.bbamcr.2016.03.013 40. Yang E., Wang X., Gong Z., Yu M., Wu H., Zhang D.. Exosome-Mediated Metabolic Reprogramming: The Emerging Role in Tumor Microenvironment Remodeling and Its Influence on Cancer Progression. Sig. Transduct. Target. Ther. (2020) 5 242. DOI: 10.1038/s41392-020-00359-5 41. Kim J.. Regulation of Immune Cell Functions by Metabolic Reprogramming. J. Immunol. Res. (2018) 2018 8605471. DOI: 10.1155/2018/8605471 42. Zhou H., Yan X., Yu W., Liang X., Du X., Liu Z., Long J., Zhao G., Liu H.. Lactic Acid in Macrophage Polarization: The Significant Role in Inflammation and Cancer. Int. Rev. Immunol. (2022) 41 4-18. DOI: 10.1080/08830185.2021.1955876 43. Wagner M., Bjerkvig R., Wiig H., Melero-Martin J.M., Lin R.-Z., Klagsbrun M., Dudley A.C.. Inflamed Tumor-Associated Adipose Tissue Is a Depot for Macrophages That Stimulate Tumor Growth and Angiogenesis. Angiogenesis (2012) 15 481-495. DOI: 10.1007/s10456-012-9276-y 44. D’Esposito V., Liguoro D., Ambrosio M.R., Collina F., Cantile M., Spinelli R., Raciti G.A., Miele C., Valentino R., Campiglia P.. Adipose Microenvironment Promotes Triple Negative Breast Cancer Cell Invasiveness and Dissemination by Producing CCL5. Oncotarget (2016) 7 24495-24509. DOI: 10.18632/oncotarget.8336 45. De Palma M., Biziato D., Petrova T.V.. Microenvironmental Regulation of Tumour Angiogenesis. Nat. Rev. Cancer (2017) 17 457-474. DOI: 10.1038/nrc.2017.51 46. Ribeiro R.J.T., Monteiro C.P.D., Cunha V.F.P.M., Azevedo A.S.M., Oliveira M.J., Monteiro R., Fraga A.M., Príncipe P., Lobato C., Lobo F.. Tumor Cell-Educated Periprostatic Adipose Tissue Acquires an Aggressive Cancer-Promoting Secretory Profile. Cell. Physiol. Biochem. (2012) 29 233-240. DOI: 10.1159/000337604 47. Rybinska I., Agresti R., Trapani A., Tagliabue E., Triulzi T.. Adipocytes in Breast Cancer, the Thick and the Thin. Cells (2020) 9. DOI: 10.3390/cells9030560
--- title: 'Lace Up and Mindfulness: A Randomized Controlled Trial Intervention to Reduce Emotional Eating, Anxiety, and Sleep Disturbances in Latinx and Black Youth' authors: - Norma Olvera - Sascha Hein - Molly Matthews-Ewald - Rongfang Zhang - Rhonda Scherer journal: Children year: 2023 pmcid: PMC10046922 doi: 10.3390/children10030538 license: CC BY 4.0 --- # Lace Up and Mindfulness: A Randomized Controlled Trial Intervention to Reduce Emotional Eating, Anxiety, and Sleep Disturbances in Latinx and Black Youth ## Abstract This study assessed the effects of a 12-week afterschool mindfulness-based diet and exercise intervention on mental and physical health in Latinx and Black youth. One hundred forty-eight boys and girls (average age = 10.1 years, SD = 1.3 years; $52\%$ girls; $72.3\%$ Latinx) were randomized to either the experimental group ($$n = 80$$) or the control group ($$n = 68$$). The experimental group participants engaged in fitness yoga, kickboxing, and/or spinning sessions, and mindfulness practices (e.g., breathing, meditation, and mindful eating) twice per week for 12 weeks. The control group participants engaged in a recreational play session once per week for 12 weeks. All participants completed surveys (demographics, acculturation, anxiety, emotional eating, sleep, and food intake) and had their height, weight, and percent body fat measured pre- and post-intervention. Participants wore an accelerometer for 7 days pre- and post-intervention. Repeated measures analysis of covariance indicated that the experimental group participants reported lower scores in emotional eating, anxiety, and sleep latency post-intervention compared to the control group participants. Conversely, no significant differences were observed in physical activity between the experimental and control group participants post-intervention. These findings indicate that a mindfulness-based intervention has a positive effect on emotional eating, anxiety, and sleep latency among youth of color. ## 1. Introduction Childhood obesity remains a persistent health problem in the United States. According to national data, $19.3\%$ of American children (ages 6–11) and $20.9\%$ of adolescents (ages 12–19) are classified as obese [1]. Childhood obesity is disproportionately affecting children and adolescents of color compared to their White peers. Specifically, $28\%$ of Mexican American and $23\%$ of African American children between the ages of 6 and 11 have been found to have obesity compared to $16\%$ of White children. This discrepancy persists throughout childhood and into young adulthood. More specifically, even higher disproportional obesity rates are reported in adolescents and young adults (ages 12–19), with $31\%$ of Mexican American and $28\%$ of African American adolescents and young adults having obesity compared to $21\%$ of their White counterparts [1]. The physical and psychological consequences of childhood obesity are well documented. Children and adolescents with obesity are more likely to engage in maladaptive eating practices such as emotional eating [2] and overeating [3]; engage in lower levels of physical activity [4]; report higher rates of both anxiety and depression [5]; and sleep disturbances [6] compared to their counterparts of normal weight. Because of the physical and psychological consequences of obesity and high obesity prevalence among children and adolescents, particularly youth of color, the development of effective interventions to prevent and treat obesity among youth is crucial. In general, childhood and adolescent obesity prevention and treatment interventions have focused on diet only, physical activity only, or a combination of diet and physical activity, and have yielded unsatisfactory or mixed results. For example, Brown and colleagues [7] conducted a meta-analysis review of 153 randomized controlled trials (RCTs) of obesity interventions that included children and adolescents (ages 0–18 years; 85 studies focused on children ages 6–12 years). The results from this meta-analysis suggest that obesity interventions that incorporated only diet reduction or only increased physical activity were not effective in lowering standardized body mass index (BMI) z-scores. Similarly, other meta-analyses have shown no statistically significant BMI and BMI z-score reductions in children and adolescents after participation in nutrition and physical activity interventions [8,9]. Given the limited effectiveness of childhood obesity interventions relying primarily on improving the quality of dietary intake and physical activity, alternative approaches to developing obesity programs have been suggested. Based on Marks’s circle of discontent theory [10], obesity programs might focus on addressing psychological pathways linking obesity and negative affect. Furthermore, in developing obesity programs, Rosenbaum and White [11] call for a more complete biopsychosocial approach that includes the physiological implications of the negative affect that can be associated with obesity. Emotional eating has been identified as a way that some individuals mitigate negative affect and is strongly and positively associated with both weight gain [12,13,14], and a higher percentage of body fat in youth [15]. Therefore, focusing on decreasing emotional eating episodes may curb the consumption of energy-dense foods, thereby reducing the likelihood of excess weight gain. Emotional eating is defined as either eating in response to emotional cues (e.g., boredom, anxiety or fear, sadness, or depression) [16] or eating in an unhealthy manner to alleviate negative emotions [17]. Mindfulness has emerged as an approach that can improve emotional eating regulation. Mindfulness, according to Kabat-Zinn [18] (p. 4), is defined as ‘paying attention in a particular way: On purpose, in the present moment, and nonjudgmentally.’ Mindfulness is often further described as being self-aware, focusing on the present moment, and acknowledging feelings, thoughts, and body sensations in a non-judgmental manner [19]. Interventions including mindful eating practices (e.g., eating slowly) have been effective in decreasing engagement in emotional eating because they promote emotion regulation designed to reduce caloric intake [20] and sugar and fat consumption [21]. Moreover, mindful eating practices encourage individuals to pay close attention to body-related sensations in response to the foods they consume as well as thoughts they have about food [22]. Mindful eating has also been used to increase dietary self-control to suppress short-term impulses to eat with the aim of pursuing long-term weight goals [23]. Some studies have also shown that mindfulness meditation interventions reduce emotional eating in college students [24] and food cravings and overeating in women with overweight [25]. However, no studies have been conducted to assess the effects of mindfulness practices such as meditation and mindful eating on emotional eating among youth of color. Compared to their impact on emotional eating, the effects of mindfulness practices on reducing anxiety and sleep disturbances have been more extensively investigated [5,6]. For instance, Eberth and Sedlmeier [26] reviewed 49 studies involving different types of meditation practices. The results from this review indicated that there is a strong association between mindfulness practices and stress and anxiety reduction and lower negative emotions. Dunning and colleagues [27] reviewed 33 RCTs and found significant positive effects of mindfulness-based interventions on depression and anxiety among children and adolescents. Similarly, intentional breathing exercises have demonstrated improvement in sustained attention, affect, and cortisol levels [28]. In sum, mindful stress reduction practices including meditation and intentional breathing exercises have been found to reduce mental distress in youth after an intervention. In addition, integrating mindfulness practices with physical activities such as yoga has shown positive effects on mental health [29]. Studies indicate that children who engage in yoga may experience a reduction in stress or anxiety, enhancement in mood [30,31], and improved breathing efficiency [32]. Furthermore, school-based yoga-type programs have shown decreases in anxiety and depression scores among children immediately after they participated in these programs [33,34,35]. Kennedy and Resnick [29] also suggest that integrating mindfulness with exercise is one way to initiate exercise and improve self-efficacy. Mindfulness–exercise practices have also been recognized to be beneficial in lowering sleep disturbances (e.g., quality, duration, and latency) in adult women (e.g., via a mobile app) [36]. Limited research has been conducted on the effects of mindfulness–exercise-based interventions on youth’s sleep disturbances [6], and these studies appear to have had mixed results. Some research has shown that mindfulness–exercise-based interventions improved global sleep quality, sleep onset latency, and daytime sleepiness among adolescents [37]. In addition, Bei and colleagues [38] reported that sleep education and mindfulness training (e.g., meditation) had beneficial effects on the sleep quality of girls aged 13–15 years. By contrast, Sibinga et al. [ 39] found that although a school-based mindfulness intervention was associated with reduced anxiety and rumination, it did not have an impact on sleep quality among high school boys (with $95\%$ of the sample being African American). Taken together, these studies suggest that promoting mindfulness practices (e.g., meditation, breathing, and yoga) and exercise has positive effects on participants’ physical and mental health. Despite the positive effects of mindfulness, very few mindfulness interventions have included youth of color. Thus, the purpose of this study was to test the efficacy of a 12-week mindfulness diet- and exercise-based intervention named Lace Up and Mindfulness (LUAM) in reducing emotional eating, reducing anxiety, and improving sleep quality among Latinx and Black youth using an RCT approach. It was hypothesized that experimental group participants would exhibit lower anxiety and emotional eating scores, and increased sleep quality compared to the control group participants after completing a 12-week intervention. ## 2.1. Participants The sample consisted of 148 children ($52\%$ girls [$$n = 77$$]; mean age = 10.11 years, SD = 1.3 years; $72.3\%$ Latinx, $18.9\%$ Black, $8.8\%$ multiracial [ClinicalTrials.gov identifier is NCT03885115]). To participate in this study, children had to be: [1] between 9 and14 years old; [2] of Latinx or Black descent; [3] without a physical disability (e.g., inability to walk) or medical conditions (e.g., heart condition) that may interfere with their participation in the exercise part of the intervention; and [4] willing to participate in pre- and post-intervention measurement sessions. Children, regardless of their weight status, were encouraged to participate. Recruitment efforts were conducted at seven elementary schools with a >$80\%$ enrollment of Latinx or Black children from a major southwest city independent school district. With the permission of school administrators, research assistants employed several recruitment strategies: (a) sending study flyers home with children; (b) informing caregivers (mainly mothers) and children about the study at school events such as health fairs, parent-teacher organization meetings, and celebrations; and (c) referrals from school nurses and teachers to encourage eligible families to participate in this study. The university’s Institutional Review Board approved the study protocol. ## 2.2. Procedures The research team invited interested caregivers and their children to attend a study orientation session at the child’s school grounds. During the study orientation, research assistants provided the caregiver and child with more detailed information about the study description, expectations, time commitment, randomization procedures, and eligibility requirements. At this time, potential participants also had an opportunity to ask questions. At the end of each orientation, if potential participants were still interested in participating in this study, the caregiver signed an informed consent form for their participation and a permission form for their child. Children also signed an assent form if they agreed to participate. Subsequently, research assistants scheduled parents and children for two [2] 60 min baseline measurement sessions conducted at the child’s school. For this study, only child data collection and analyses are presented since we explored the impact of an intervention on children’s mental and physical health. During the first baseline measurement session, after a brief instruction section, children completed a series of surveys and had their height, weight, and percent body fat recorded privately. At the end of this first baseline measurement session, each child received an accelerometer (ActiGraph wGT3X-BT model), with instructions on how to wear it for 7 days. Within one [1] week, research assistants scheduled a second baseline measurement session where participants finished completing the surveys, returned the accelerometers, and received information on the next steps of the study. Research assistants followed similar measurement procedures during the post-intervention measurement session (12 weeks after the pre-intervention assessment). After the completion of baseline data collection, the assignment to either the experimental group (EG) or control group (CG) was conducted based on a simple (i.e., unstratified) random sampling procedure. A study researcher used a randomization table to randomly assign the child ID number to one of the study conditions and concealed this process from all study researchers, project coordinators, research assistants, and instructors. Then, the researcher informed the project coordinator of the participants’ assigned study condition, so they could subsequently notify families of their assigned study condition. In the case of a family that had two children eligible to participate in this study, both children were assigned to the same treatment condition to circumvent treatment diffusion across EG and CG participants. ## 2.3.1. Experimental Group (EG) The EG participants took part in the LUAM intervention aimed to reduce emotional eating, anxiety, and sleep disturbances through mindfulness practices. Specifically, the LUAM intervention consisted of a 12-week after-school program that included two group sessions per week of either one [1] 60-min kickboxing or spinning session and one [1] 60-min fitness yoga session with the inclusion of mindfulness practices at each exercise session. Certified fitness instructors led each exercise session, consisting of a 5-min warm-up period, 40-min kickboxing/spinning or fitness yoga session, 10-min mindfulness practices and cool-down period, and a 5-min bathroom and water break. In this study, kickboxing included repetitive rapid movement with hands (e.g., throwing punches) and feet (e.g., kicking) with increasing intensity. Kickboxing was chosen as one type of exercise as it provides an aerobic workout, which is beneficial for burning calories, and improving cardiovascular fitness. Kickboxing incorporates martial art techniques and boxing skills accompanied by popular music and is aimed at increasing energy expenditure while strengthening core muscles [40]. During the spinning sessions, the EG participants used a stationary bicycle with a weighted flywheel to increase or decrease resistance while pedaling. The spinning focused on increasing endurance, strength, and intensity. Spinning has been shown to increase cardiovascular endurance [41]. The fitness yoga sessions incorporated both strengthening and stretching activities in a fast-paced environment, increasing energy expenditure while promoting breathing control and meditation. Fitness yoga is reported to benefit children’s physical and mental health [42]. Additionally, the EG participants engaged in intentional mindfulness practices during and after each exercise session. Mindfulness practices focused on guided intentional focus or meditation, sustained attention, breathing practices, body awareness, and mindful eating [43]. For example, children engaged in guided meditation and breathing practices during yoga and kickboxing/spinning. Further, children engaged in breathing activities such as butterfly breath and hot chocolate breath. In the butterfly breath activity, the child used either their hands, arms, or legs to move or flap as they would breathe in and out. For the hot chocolate breath activity, the child would pretend to be freezing and have a hot chocolate in their hands. The child would blow on hot chocolate to cool it down. Furthermore, children were asked to practice breathing and meditation any time they felt afraid, anxious, or upset. The EG participants also engaged in several mindful eating activities while consuming a healthy snack and water. For instance, children were encouraged to eat food items slowly and with their eyes closed to become more aware of food traits through other senses, such as smell, touch, and taste, while noting thoughts, feelings, and physical responses to that food. Another example of a mindful eating activity involved children focusing on body awareness with an emphasis on recognizing eating cues for hunger or emotional factors. Those in the EG also participated in food demonstrations of healthy snacks (e.g., parfaits and smoothies) that they could make on their own at home. The combined exercise and eating snack sessions lasted 1.5 h per week. The EG participants received a LUAM t-shirt at the end of the intervention. ## 2.3.2. Control Group (CG) CG participants attended weekly 1.5 h recreational play sessions for 12 weeks. These sessions included several recreational games, such as “Tic Tac Toe” (relay game) and “Pac-Man” (tagging), and playing sports such as soccer or basketball. Each week, research assistants provided CG participants with specific options for games or sports to play. Children selected which game or sport to play for the first 30 min of the session and then switched activities for another 30 min if they wished to change. If they did not wish to change, they could play the same game or sport for another 30 min. CG participants received a healthy snack and water at the end of each session and a t-shirt at the end of the program. ## 2.3.3. Setting and Safety Both the EG and CG group sessions were implemented for different cohorts of participants at the children’s school facilities such as a school gym, an empty classroom dedicated to our study, or school green areas. The EG and CG group sessions were typically held between 4:00 and 5:30 p.m. or 4:30–6:00 p.m. (depending on the principal’s approval) during 2017–2019 before the COVID-19 pandemic. Participants’ safety was paramount in this study. Several safety guidelines were followed to minimize participants’ risk of injury. For instance, before each exercise session (experimental and control), the instructor reminded participants to check the laces of their tennis shoes to make sure they were tied. During the EG sessions, participants were encouraged to breathe and exercise at a safe and effective pace that they could maintain. Modifications were provided to ensure safe engagement in the activities and to accommodate varied fitness levels. Instructors were required to complete the American Red Cross CPR/First Aid/AED training and have a readily available first aid kit. ## 2.4. Measures Except for demographic and acculturation data collected only at baseline, data were collected at two time points: [1] baseline (T1) assessment before group assignment randomization, and [2] post-intervention (T2) assessment immediately after participation in the 12-week intervention. ## 2.4.1. Demographic Characteristics and Acculturation The demographic questionnaire included 12 items that assessed age, gender, grade, family structure (i.e., number of family members living in the same household), birthplace (e.g., the United States, Mexico, Central America, etc.), and access to fitness equipment (e.g., bike). All participants completed the 12-item Short Acculturation Scale for Hispanic Youth [44]. This instrument assessed acculturation to mainstream culture in terms of language use/proficiency (9 items, e.g., “What language do you usually speak at home?”), and social relations (3 items, e.g., “Your close friends are…”). All participants answered these acculturation questions on a 5-point Likert-type scale ranging from 1 (only Spanish) to 5 (only English) and from 1 (all Hispanic) to 5 (all non-Hispanic), respectively. A sum score greater than or equal to 30 indicated high acculturation and a sum score of less than 30 designated low acculturation. The internal consistency of participants’ acculturation scale scores was adequate for Latinx participants (Cronbach’s α = 0.88) and for Black participants (Cronbach’s α = 0.79). ## 2.4.2. Weight Status and Percent Body Fat Trained research assistants measured each participant’s body height using a Seca 213 stadiometer (Hamburg, Germany) and rounded to the nearest 0.1 cm, and body weight using a scale (Tanita SC-331S, Tokyo, Japan), rounded to the nearest 0.1 kg. Participants were weighed without shoes or heavy clothing. Based on body height and weight, research assistants computed the body mass index using Quetelet’s index [body weight (kg)/height (m2)]. Obesity status was calculated using BMI values for the age and sex-specific percentiles according to the Centers for Disease Control and Prevention (CDC) guidelines [45]. Using CDC guidelines, participants were classified as being underweight with a BMI < 5th percentile; healthy weight with a BMI in the 5th–84th percentile; overweight with a BMI in the 85th–94th percentile; and obese with BMI ≥ 95th percentile. The Tanita scale was also used to assess percent body fat (%BF). ## 2.4.3. Emotional Eating Emotional eating was assessed using the McKnight Risk Factor Survey IV [46] emotional eating subscale. The emotional eating score was computed by averaging responses to six items. Three out of the six items had questions regarding eating less such as, “In the past month, how often did you eat less than usual [1] when you were bored? [ 2] when you try to feel better about yourself?; and [3] when you were upset?” The other three items included questions involving eating more such as, “In the past month, how often did you eat more than usual when you were bored, trying to feel better about yourself, and when you were upset?” Items from this subscale were rated using a 5-point Likert scale (1 = never to 5 = always). In this study, the internal consistency (Cronbach’s α) of the emotional eating subscale for the pre- and post-assessments was good (0.86 and 0.87, respectively). ## 2.4.4. Multidimensional Anxiety Scale for Children, 2nd Edition (MASC-2™) The MASC-2™ is a comprehensive measure of anxiety-related symptoms in youth aged 8 to 19 years [47]. This 50-item measure assesses a broad range of emotional, physical, cognitive, and behavioral symptoms of anxiety. Participants answered items on a 4-point Likert scale, ranging from 0 (never true about me) to 3 (often true about me). The questionnaire yields raw scores and standardized T-scores of the overall degree of self-reported total anxiety, six anxiety scales, and four subscales. These anxiety scales assess separation anxiety/phobias, generalized anxiety disorder, social anxiety (humiliation/rejection and performance fears subscales), obsessions and compulsions, physical symptoms (panic and tense/restless subscales), and harm avoidance. Following the MASC-2™ scoring guidelines, T-scores were used for analyses in the current study, as they are standard scores using the MASC-2™ profile for boys and girls. The MASC-2™ is recognized as a valid and reliable anxiety scale, with excellent internal consistency (Cronbach’s α = 0.92 for the total score) [48]. In the current sample, the internal consistency of the MASC-2™ total score was excellent for both pre- and post-intervention assessments (Cronbach’s α = 0.94). ## 2.4.5. Pittsburg Sleep Quality Index (PSQI) The PSQI is a 19-item self-report questionnaire composed of seven subscales: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication, and daytime drowsiness [49]. The present study focused on three sleep quality subscales (sleep duration, sleep efficiency, and sleep latency) as suggested by previous research [50]. Duration is a measure of total sleep hours at night per day. Efficiency refers to the percentage of total time in bed spent in sleep. Latency is a computed Likert rating (scores of 0 to 3) based on two indicators of how quickly a participant falls asleep (where higher scores indicate more difficulty in falling asleep). A global PSQI score ranged from 0 to 21. Global scores lower than 5 indicate healthy sleep quality, whereas scores greater than or equal to 6 indicate worse sleep quality. The internal consistency of the PSQI was reported by Ranit et al. [ 51] as acceptable (Cronbach’s α = 0.73). In this study, the internal consistency of the PSQI was acceptable for both the pre-intervention assessment (Cronbach’s α = 0.74) and post-intervention assessment (Cronbach’s α = 0.69). ## 2.4.6. Physical Activity Pre-to-post-intervention changes in physical activity (i.e., duration, frequency, and intensity) were assessed using an accelerometer (ActiGraph wGT3X-BT model). Research assistants provided participants with instructions (both written and oral) on how to wear an accelerometer fastened to a belt and placed on the right hip for 7 days (including sleep time). A valid assessment was determined by a minimum of 8 h of wear time for at least 4 days (2 weekdays and 2 weekend days). A one-minute epoch was the amount of time over which activity counts were integrated and recorded. The algorithm utilized to identify non-wear time in the ActiGraph device included intervals of at least 60 consecutive minutes of zero activity counts between 1 and 100 counts. Following the pre- and post-data collection, the accelerometer data were downloaded and analyzed using the ActiLife software to determine participants’ physical activity intensity: sedentary, light, moderate, moderate-to-vigorous, and vigorous physical activity. The intensity of physical activity was established using the Evenson and colleagues’ [52] children algorithm cut-off points in the ActiLife software. ## 2.4.7. Food Frequency Questionnaire (FFQ) Participants completed a food frequency questionnaire developed by Matt and colleagues [53] to assess baseline differences in food intake between the EG and CG participants. Using this instrument, participants indicated how often they consumed food from the following food groups: fruits, vegetables, sweets, and sweetened beverages. Participants answered questions such as “How often did you eat/drink (food/beverage item)?” on a Likert-type scale: [1] never, [2] less than once a month, [3] once a month, [4] 1–2 per week, [5] 3–4 per week, [6] 5–6 per week, [7] 1 per day, and [8] 2 or more per day. This instrument was selected because of its improved intake recall and reporting in ethnically diverse populations versus prior instruments such as the Fred Hutchinson FFQ [53]. In this study, the FFQ internal consistency for the pre- and post-assessments was excellent (Cronbach’s α = 0.93 and 0.92, respectively). ## 2.5. Data Analysis Statistical analyses for this study were conducted in four phases. First, the missing data pattern and statistical assumptions were analyzed using SPSS 23.0. The variables showed slight deviations from normality, with skewness ranging from 2.29 to 2.85 and kurtosis ranging from 11.02 to 0.97. According to the rules of thumb for normality from Hair et al. [ 54] and Byrne [55], skewness values should range between 2 and −2, and kurtosis between 7 and −7. Fortunately, repeated measures analysis of covariance (ANCOVA) is robust to violations of normality [56]. Second, we analyzed baseline descriptive characteristics for the retained and withdrawn children to assess potential bias based on study attrition. Then, we analyzed baseline descriptive characteristics for the EG and CG participants to determine baseline equivalence. Third, pre- and post-intervention changes in primary outcomes were analyzed using repeated measures ANCOVA with three covariates (gender, age, and BMI). ANCOVA assumptions (e.g., sphericity) were tested and no violations were detected. Fourth, the differential effects of the intervention based on the EG participant attendance (% of sessions attended) were explored using repeated measures ANCOVA with three covariates (gender, age, and BMI). Complete-case analysis was preferred over other approaches (e.g., multiple imputations of missing post-intervention scores) given the considerable number of outcome variables in relation to the modest sample size. ## 3.1. Sample Size Initial recruitment efforts included reaching out to 640 Latinx and Black youth and their caregivers who received information about the study (Figure 1). Out of the contacted 640 parent–child dyads, 231 ($36.1\%$) pairs expressed an interest in participating in the study orientation. At baseline, 17 participants dropped out and had incomplete data, and 5 participants were excluded for not meeting ethnicity requirements, thereby reducing the sample to 209 children. The remaining 209 children were randomly assigned to either the EG ($$n = 117$$) or CG ($$n = 92$$). After randomization, the final analytic sample was reduced to 148 participants due to 61 children (nEG = 37, $31.6\%$ of EG participants; nCG = 24, $26.1\%$ of CG participants) being excluded as a result of missing key post-intervention measures (nEG = 13, nCG = 9), or for dropping out of the study (nEG = 24, nCG = 15). Reasons for dropping out included children having time conflicts with competing academic and sports demands, school truancy, losing interest, and having unreliable transportation. Results from a sensitivity GPower (F test, repeated measures ANOVA, within-between interaction) with the present sample size of $$n = 148$$, alpha error probability of 0.05, power of 0.8, and average correlation among repeated measures of 0.5 showed that a required effect size f of 0.12 is considered statistically significant. Demographic and key variables of interest at baseline were compared between the remaining participants ($$n = 148$$; EG = 80, CG = 68) and those excluded from the analysis ($$n = 61$$). No significant group differences in demographic characteristics and study outcomes were found, except for sleep duration (Supplemental Tables S1 and S2). On average, total hours of sleep at night were significantly higher among the included participants ($M = 8.61$, SD = 2.10) than among the withdrawn children ($M = 7.83$, SD = 1.76; t [175] = 2.18, $$p \leq 0.03$$, and Cohen’s $d = 0.39$) [57]. ## 3.2.1. Demographic and Acculturation Characteristics As shown in Table 1, most of the children were born in the United States ($$n = 134$$; $90.5\%$), self-identified as Latinx ($$n = 107$$; $72.3\%$), and reported a high acculturation level ($$n = 78$$; $74.3\%$). When comparing the EG and CG participants’ demographic characteristics, only ethnicity was significantly different between the groups (χ2(2, $$n = 107$$) = 8.83, $$p \leq 0.01$$, Cramer’s $V = 0.24$). That is, a larger percentage of Latinx participants were randomly assigned to the EG ($78.8\%$) compared to CG ($64.7\%$). No significant group differences at baseline were observed in participants’ age, gender, place of birth, and acculturation level, confirming the robustness of the randomization procedure. ## 3.2.2. Weight Status and Percent Body Fat As shown in Table 1, $52\%$ of the participants were overweight or obese, $39.2\%$ were of normal weight, $3.4\%$ were underweight, and $5.4\%$ missing. On average, participants had a percent body fat (%BF) of $27.47\%$ (SD = $10.35\%$, values exceeding 75th percentile cutoffs of %BF for 10-year-old boys ($24.5\%$) and 10-year-old girls ($26.4\%$), respectively) [58]. We used 10-year-olds as a reference because the mean age of the sample was 10 years. There were no significant differences in weight status and %BF at baseline between the EG and CG participants. ## 3.2.3. Emotional Eating, Anxiety, and Sleep Scores As shown in Table 2, for emotional eating at baseline, participants had a mean composite emotional eating score of 1.69 (SD = 0.82), indicating some engagement in emotional eating. Participants’ T-scores on the MASC-2 averaged 51.43 (SD = 12.89), which falls in the average anxiety score range. Participants reported a mean global score of 3.86 (SD = 3.38) indicating, on average, good sleep quality. Specifically, participants reported a mean sleep duration of 8.61 h per night (SD = 2.10 h), had on average $88.24\%$ sleep efficiency (SD = 14.28), and reported good sleep latency (scaled from 0—good to 3—bad; $M = 0.71$, SD = 0.85). No significant differences between the EG and CG participants were detected at baseline in emotional eating, anxiety, and some sleep variables such as global sleep quality, sleep duration, and sleep latency. However, sleep efficiency among the EG participants ($M = 85.20$, SD = 15.75) was significantly lower compared to that among the CG participants ($M = 91.61$, SD = 11.73; t [94] = −2.31, $$p \leq 0.02$$, Cohen’s $d = 0.46$). ## 3.2.4. Physical Activity and Dietary Intake At baseline, participants ($$n = 73$$) engaged in an average of 26.34 min (SD = 21.87 min) of daily moderate-to-vigorous physical activity (MVPA). In terms of their dietary intake, only $33\%$ of participants at baseline reported consuming fruit and $12\%$ consuming vegetables at least 5 times per day (data not tabled). Participants on average drank one sweetened beverage every day and ate sweets twice per day. There were no significant differences between the EG and CG participants regarding physical activity and dietary intake at baseline. ## 3.3.1. Emotional Eating, Anxiety, and Sleep Scores Significant pre- and post-intervention effects on emotional eating and anxiety scores were observed among the EG participants, but not in the CG participants (Table 3). Specifically, there was a significant pre-to-post-intervention decrease in the EG participants’ emotional eating score (F [1,121] = 5.41, $$p \leq 0.02$$, partial η2 = 0.04), compared to that of CG participants. Likewise, there was a significant decrease in the EG participants’ anxiety total T-scores (F [1,118] = 5.09, $$p \leq 0.03$$, partial η2 = 0.04), separation anxiety/phobia subscale T-scores (F [1,117] = 12.53, $p \leq 0.001$, partial η2 = 0.10), and humiliation/rejection subscale T-scores (F [1,118] = 5.83, $$p \leq 0.02$$, partial η2 = 0.05) from T1 to T2. Regarding sleep quality, the EG and CG participants differed significantly in sleep latency (time taken to fall asleep) from pre- to post-intervention assessments (F [1,118] = 4.98, $$p \leq 0.03$$, partial η2 = 0.04, Table 3). On average, the EG participants improved their reported sleep latency pre- and post-intervention. By contrast, CG participants reported longer time needed to fall asleep both pre- and post-intervention. No significant differences in the rates of change were observed in sleep duration and sleep efficiency between the two treatment groups. It is noteworthy that both the EG and CG participants reported a good quality in terms of sleep duration, sleep efficiency, and sleep latency pre- and post-intervention. ## 3.3.2. Physical Activity and Dietary Intake We also assessed pre- and post-intervention differences in MVPA in the EG or CG participants (Table 3). On average, the EG participants ($$n = 31$$) recorded about 22 min per day of MVPA both before ($M = 22.67$, SD = 16.03) and after ($M = 22.41$, SD = 13.89) the intervention. CG participants ($$n = 33$$) decreased from an average of 29.44 min per day (SD = 26.30) at baseline to 21.49 min per day (SD = 21.49) after the intervention. In addition, no significant changes in dietary intake were observed pre- and post-intervention between the EG and CG participants (Table 3). ## 3.4. Intervention Dosage-Response Effects Finally, we examined whether the changes in outcomes pre- and post-intervention for the EG participants were different depending on their intervention attendance by assessing a time-by-attendance interaction. As presented in Table 4, the results showed that after controlling for participants’ gender, age, and BMI, only the change in sleep efficiency across both time points differed significantly as a function of attendance level (F [1,18] = 2.61, $$p \leq 0.03$$, partial η2 = 0.67). This interaction indicates a greater increase in sleep efficiency among the EG participants with higher attendance rates. No other changes in outcomes were significantly different among the EG participants as a function of attendance. However, this finding might be explained by low variability in the EG participants’ attendance rate, as attendance was high among the EG participants; $97.6\%$ of the EG participants attended at least $50\%$ of the sessions (no tabulations are presented). ## 4. Discussion This study tested the efficacy of a 12-week LUAM intervention in reducing emotional eating, reducing anxiety, and improving sleep quality in a sample of Latinx and Black youth using an RCT approach. To the authors’ knowledge, the LUAM intervention is one of the few RCT mindfulness studies assessing its impact on Latinx and Black early adolescents’ mental and physical health. Consistent with the study’s hypotheses, the findings of this research indicate that the LUAM intervention lowered pre- and post-intervention emotional eating in the EG participants when compared to those of CG participants. Our study’s effect of decreasing emotional eating scores in youth is consistent with previous research [59] and expands the extant understanding of the effect of mindfulness practices (e.g., mindful eating and breathing) on emotional eating in youth of color with overweight/obesity. The reduction in emotional eating through mindfulness practices is significant because mindfulness practices promote emotion regulation with the aim of reducing caloric intake [20] and sugar and fat consumption [21]. Mindful eating practices encourage youth to pay close attention to body-related sensations in response to the foods they consume as well as thoughts they have about food [22]. The results from the LUAM intervention also showed positive effects on lowering anxiety and sleep disturbance scores in Latinx and Black youth after participation in the EG compared to the CG. Our findings are congruent with several studies that have shown mindfulness practices (e.g., meditation and breathing) as being helpful in reducing anxiety and sleep disturbances among children and adolescents [5,6,26,27,28]. In addition, the use of a mindfulness–exercise-based approach (e.g., yoga and spinning/kickboxing) supports research suggesting that such approaches are negatively associated with both anxiety [60,61,62] and sleep disturbances [37,38]. Overall, given the connection between sleep problems and the development of psychological distress including anxiety and depression [62,63,64], our findings also provide a promising approach for lowering anxiety and sleep disturbances. It is noteworthy to mention that the LUAM intervention did not show significant changes in pre- and post-intervention daily minutes of MVPA between the EG and CG participants. It would seem plausible to suggest that the exercise dosage in the current study was not sufficient to see a marked increase in daily time spent in MVPA. Although results among youth tend to be mixed, there is some evidence to suggest that there may be physical activity compensation among youth. That is, youth who engage in increased physical activity on a particular day may compensate for this activity by decreasing their activity the following day. Indeed, among a group of 8–11-year-olds, Ridgers et al. [ 65] found that when participants engaged in higher levels of light physical activity (LPA), they engaged in less LPA and MVPA the following day. Likewise, increased MVPA time was associated with less LPA and MVPA time the following day. Thus, it could be that youth in the current study may have been compensating for the activity received as part of the intervention during non-intervention times. However, more research is needed to test this hypothesis. Another potential reason that there were no statistically significant differences in pre-to-post-intervention daily minutes of MVPA between the EG and CG participants is the relatively low amount of complete data in each group. It is possible that those participants in both groups who wore the accelerometer had greater awareness of their levels of physical activity and thus may have engaged in physical activity outside of the intervention at relatively similar levels. ## Limitations This study has several limitations. First, because of the small and unequal sample size of Latinx and Black participants, we were unable to examine the impact of this intervention on participants by ethnicity. Second, because neither the Latinx nor Black populations are monolithic, nor are their social environments and family contexts, our results might not be generalizable to a wider population of Latinx and Black youth. Participants in this study included mostly Latinx youth born in the United States, with high acculturation levels. Future studies should include a larger representative sample of Latinx and Black youth to be able to disaggregate the sample and to examine biopsychosocial factors that might differentially impact the intervention effects across the two ethnic groups. Some of these biopsychosocial factors could include parent stressors related to safety in the community, which have been positively associated with obesity among adolescents [66]. Third, although the PSQI is frequently used as a measure of sleep quality, the relatively low internal consistency for the current study should be noted. Alternative sleep quality measures should be considered in the future. Fourth, participants in both the EG and CG were recruited from the same set of schools. Despite the random assignment, it is possible that participants who were randomly assigned to the EG discussed the activities in which they engaged as part of the intervention which may have inadvertently impacted the CG participants (e.g., the CG participants may have become more aware of their physical activity, eating, and sleeping habits because of learning about the EG intervention). However, differences between the EG and CG participants were still identified, indicating that the mindfulness–exercise-based programs may have positively impacted the youth. Fifth, although randomization minimizes the potential for bias as it enables comparisons to be made between the experimental and control groups, it is possible that other biases, such as performance bias, may have been present among the EG participants. The performance bias might be due to differences between groups in the treatments that they receive, or in exposure to factors other than the interventions of interest. Sixth, this study examined the immediate impact of the intervention on youth. It is unknown whether these changes are sustained long-term. For example, it is unclear whether youth in the EG continued to engage in mindfulness practices once the 12-week program was complete. It is also possible that the positive effects of the intervention might be short-lived and inconsequential in the long term if some family factors such as parental stressors are not addressed. Future studies should examine the sustained, longer-term impact of the intervention and how this long-term impact could be hindered by family factors. ## 5. Conclusions Despite these limitations, this study provides preliminary evidence, through an RCT approach, that there are benefits to utilizing mindfulness practices (e.g., mindful eating and exercise-based approaches) in reducing emotional eating and anxiety and increasing sleep quality among youth of color. Given that there have been relatively few studies examining the impact of mindfulness interventions, especially among marginalized youth who are at an increased risk for obesity, this study begins to fill an important gap in the literature addressing the psychological factors associated with obesity. Further, apart from spinning, the activities and practices included in the LUAM intervention (i.e., kickboxing and mindful eating) can be completed without equipment, at youths’ homes. This has important implications for the feasibility of such an intervention in a real-world setting. ## References 1. Ogden C.L., Fryar C.D., Martin C.B., Freedman D.S., Carroll M.D., Gu Q., Hales C.M.. Trends in obesity prevalence by race and Hispanic origin—1999–2000 to 2017–2018. JAMA* (2020.0) 324 1208-1210. DOI: 10.1001/jama.2020.14590 2. Jalo E., Konttinen H., Vepsäläinen H., Chaput J.P., Hu G., Maher C., Maia J., Sarmiento O.L., Standage M., Tudor-Locke C.. Emotional eating, healthy behaviours, and obesity in children: A 12-country cross-sectional study. Nutrients (2019.0) 11. DOI: 10.3390/nu11020351 3. Egbert A.H., Wilfley D.E., Eddy K.T., Boutelle K.N., Zucker N., Peterson C.B., Doyle A.C., Grange D.L., Goldschmidt A.B.. Attention-deficit/hyperactivity disorder symptoms are associated with overeating with and without loss of control in youth with overweight/obesity. Child. Obes. (2018.0) 14 50-57. DOI: 10.1089/chi.2017.0114 4. Sahoo K., Sahoo B., Choudhury A.K., Sofi N.Y., Kumar R., Bhadoria A.S.. Childhood obesity: Causes and consequences. J. Fam. Med. Prim. Care (2015.0) 4 187-192. DOI: 10.4103/2249-4863.154628 5. Hussein R., Mohammed R.A.E., Ahmed I.H.. Psychological impact of obesity in children. Sci. J. Al-Azhar Med. Fac. Girls (2020.0) 4 17-21. DOI: 10.4103/sjamf.sjamf_98_19 6. Gohil A., Hannon T.S.. Poor sleep and obesity: Concurrent epidemics in adolescent youth. Front. Endocrinol. (2018.0) 9 364. DOI: 10.3389/fendo.2018.00364 7. Brown A.G., Esposito L.E., Fisher R.A., Nicastro H.L., Tabor D.C., Walker J.R.. Food insecurity and obesity: Research gaps, opportunities, and challenges. Transl. Behav. Med. (2019.0) 9 980-987. DOI: 10.1093/tbm/ibz117 8. Kornet-van der Aa D.A., Altenburg T.M., van Randeraad-van der Zee C.H., Chinapaw M.J.M.. The effectiveness and promising strategies of obesity prevention and treatment programmes among adolescents from disadvantaged backgrounds: A systematic review. Obes. Rev. (2017.0) 18 581-593. DOI: 10.1111/obr.12519 9. Mead E., Brown T., Rees K., Azevedo L.B., Whittaker V., Jones D., Olajide J., Mainardi G.M., Corpeleijn E., O’Malley C.. Diet, physical activity and behavioural interventions for the treatment of overweight or obese children from the age of 6 to 11 years. Cochrane Database Syst. Rev. (2017.0) 6 CD012651. DOI: 10.1002/14651858.CD012651 10. Marks D.F.. Homeostatic theory of obesity. Health Psychol. Open (2015.0) 2 1-30. DOI: 10.1177/2055102915590692 11. Rosenbaum D.L., White K.S.. Understanding the complexity of biopsychosocial factors in the public health epidemic of overweight and obesity. Health Psychol. (2016.0) 3 1-4. DOI: 10.1177/2055102916634364 12. Aparicio E., Canals J., Arija V., De Henauw S., Michels N.. The role of emotion regulation in childhood obesity: Implications for prevention and treatment. Nutr. Res. Rev. (2016.0) 29 17-29. DOI: 10.1017/S0954422415000153 13. Boutelle K.N., Braden A., Knatz-Peck S., Anderson L.K., Rhee K.E.. An open trial targeting emotional eating among adolescents with overweight or obesity. Eat. Disord. (2018.0) 26 79-91. DOI: 10.1080/10640266.2018.1418252 14. Harrist A.W., Hubbs-Tait L., Topham G.L., Shriver L.H., Page M.C.. Emotion regulation is related to children’s emotional and external eating. J. Dev. Behav. Pediatr. (2013.0) 34 557-565. DOI: 10.1097/DBP.0b013e3182a5095f 15. Olvera N., Dempsey A., Gonzalez E., Abrahamson C.. Weight-related teasing, emotional eating, and weight control behaviors in Hispanic and African American girls. Eat. Behav. (2013.0) 14 513-517. DOI: 10.1016/j.eatbeh.2013.06.012 16. Tak S.R., Hendrieckx C., Nefs G., Nyklíček I., Speight J., Pouwer F.. The association between types of eating behaviour and dispositional mindfulness in adults with diabetes. Results from Diabetes MILES. The Netherlands. Appetite (2015.0) 87 288-295. DOI: 10.1016/j.appet.2015.01.006 17. Evers C., Stok F.M., de Ridder D.T.D.. Feeding your feelings: Emotion regulation strategies and emotional eating. Pers. Soc. Psychol. B (2010.0) 36 792-804. DOI: 10.1177/0146167210371383 18. Kabat-Zinn J.. Wherever You Go, There You Are (1994.0) 19. Chiesi F., Dellagiulia A., Lionetti F., Bianchi G., Primi C.. Using item response theory to explore the psychometric properties of the Italian version of the Child and Adolescent Mindfulness Measure (CAMM). Mindfulness (2017.0) 8 351-360. DOI: 10.1007/s12671-016-0604-y 20. Creswell J.D.. Mindfulness interventions. Annu. Rev. Psychol. (2017.0) 68 491-516. DOI: 10.1146/annurev-psych-042716-051139 21. Mantzios M., Egan H., Hussain M., Keyte R., Bahia H.. Mindfulness, self-compassion, and mindful eating in relation to fat and sugar consumption: An exploratory investigation. Eat. Weight Disord. (2018.0) 23 833-840. DOI: 10.1007/s40519-018-0548-4 22. Alberts H.J.E.M., Thewissen R., Raes L.. Dealing with problematic eating behaviour. The effects of a mindfulness-based intervention on eating behaviour, food cravings, dichotomous thinking and body image concern. Appetite (2012.0) 58 847-851. DOI: 10.1016/j.appet.2012.01.009 23. Johnson F., Pratt M., Wardle J.. Dietary restraint and self-regulation in eating behavior. Int. J. Obes. (2012.0) 36 665-674. DOI: 10.1038/ijo.2011.156 24. Katterman S.N., Kleinman B.M., Hood M.M., Nackers L.M., Corsica J.A.. Mindfulness meditation as an intervention for binge eating, emotional eating, and weight loss: A systematic review. Eat. Behav. (2014.0) 15 197-204. DOI: 10.1016/j.eatbeh.2014.01.005 25. Mason A.E., Jhaveri K., Cohn M., Brewer J.A.. Testing a mobile mindful eating intervention targeting craving-related eating: Feasibility and proof of concept. J. Behav. Med. (2018.0) 41 160-173. DOI: 10.1007/s10865-017-9884-5 26. Eberth J., Sedlmeier P.. The effects of mindfulness meditation: A meta-analysis. Mindfulness (2012.0) 3 174-189. DOI: 10.1007/s12671-012-0101-x 27. Dunning D.L., Griffiths K., Kuyken W., Crane C., Foulkes L., Parker J., Dalgleish T.. Research review: The effects of mindfulness-based interventions on cognition and mental health in children and adolescents—A meta-analysis of randomized controlled trials. J. Child Psychol. Psychiatry (2018.0) 60 244-258. DOI: 10.1111/jcpp.12980 28. Ma X., Yue Z.Q., Gong Z.Q., Zhang H., Duan N.Y., Shi Y.T., Wei G.X., Li Y.F.. The effect of diaphragmatic breathing on attention, negative affect and stress in healthy adults. Front. Psychol. (2017.0) 8 874. DOI: 10.3389/fpsyg.2017.00874 29. Kennedy A.B., Resnick P.B.. Mindfulness and physical activity. Am. J. Lifestyle Med. (2015.0) 9 221-223. DOI: 10.1177/1559827614564546 30. Eggleston B.. The benefits of yoga for children in schools. Int. J. Health Well Soc. (2015.0) 5 1-7. DOI: 10.18848/2156-8960/CGP/v05i03/41125 31. Nanthakumar C.. The benefits of yoga in children. J. Integr. Med. (2018.0) 16 14-19. DOI: 10.1016/j.joim.2017.12.008 32. Telles S., Gupta R.K., Gandharva K., Vishwakarma B., Kala N., Balkrishna A.. Immediate effect of a yoga breathing practice on attention and anxiety in pre-teen children. Children (2019.0) 6. DOI: 10.3390/children6070084 33. James-Palmer A., Anderson E.Z., Zucker L., Kofman Y., Daneault J.F.. Yoga as an intervention for the reduction of symptoms of anxiety and depression in children and adolescents: A systematic review. Front. Pediat. (2020.0) 8 78. DOI: 10.3389/fped.2020.00078 34. Hall A., Ofei-Tenkorang N.A., Machan J.T., Gordon C.M.. Use of yoga in outpatient eating disorder treatment: A pilot study. J. Eat. Disord. (2016.0) 4 38. DOI: 10.1186/s40337-016-0130-2 35. Velásquez A.M., López M.A., Quiñonez N., Paba D.P.. Yoga for the prevention of depression, anxiety, and aggression and the promotion of socio-emotional competencies in school-aged children. Educ. Res. Eval. (2015.0) 21 407-421. DOI: 10.1080/13803611.2015.1111804 36. Rung A.L., Oral E., Berghammer L., Peters E.S.. Feasibility and acceptability of a mobile mindfulness meditation intervention among women: Intervention study. JMIR mHealth uHealth (2020.0) 8 e15943. DOI: 10.2196/15943 37. Blake M., Waloszek J.M., Schwartz O., Raniti M., Simmons J.G., Blake L., Murray G., Dahl R.E., Bootzin R., Dudgeon P.. The SENSE study: Post intervention effects of a randomized controlled trial of a cognitive–behavioral and mindfulness-based group sleep improvement intervention among at-risk adolescents. J. Consult. Clin. Psychol. (2016.0) 84 1039-1051. DOI: 10.1037/ccp0000142 38. Bei B., Byrne M.L., Ivens C., Waloszek J., Woods M.J., Dudgeon P., Murray G., Nicholas C.L., Trinder J., Allen N.B.. Pilot study of a mindfulness-based, multi-component, in-school group sleep intervention in adolescent girls. Early Interv. Psychiatry (2013.0) 7 213-220. DOI: 10.1111/j.1751-7893.2012.00382.x 39. Sibinga E.M.A., Perry-Parrish C., Chung S., Johnson S.B., Smith M., Ellen J.E.. School-based mindfulness instruction for urban male youth: A small randomized controlled trial. Prev. Med. (2013.0) 57 799-801. DOI: 10.1016/j.ypmed.2013.08.027 40. Chu Z.. Effects of kickboxing on body composition of adolescents with obesity. Exerc. Biochem. Rev. (2018.0) 1 PO-001-PO-100. DOI: 10.14428/ebr.v1i3.8993 41. Chavarrias M., Carlos-Vivas J., Collado-Mateo D., Pérez-Gómez J.. Health benefits of indoor cycling: A systematic review. Medicina (2019.0) 55. DOI: 10.3390/medicina55080452 42. Nayek B., Chatterjee K.. Effect of 6-weeks yoga and aerobic exercise on muscular strength and flexibility of pre-adolescence students. Int. J. Yogic Hum. Mov. Sport. Sci. (2016.0) 1 12-16 43. Mindfulness for Teens 44. Barona A., Miller J.A.. Short acculturation scale for Hispanic youth (SASH-Y): A preliminary report. Hisp. J. Behav. Sci. (1994.0) 16 155-162. DOI: 10.1177/07399863940162005 45. BMI Percentile Calculator for Child and Teen 46. Risk factors for the onset of eating disorders in adolescent girls: Results of the McKnight longitudinal risk factor study. Am. J. Psychiatry (2003.0) 160 248-254. DOI: 10.1176/ajp.160.2.248 47. March J.S.. The Multidimensional Anxiety Scale for Children (MASC2) (2013.0) 48. Fraccaro R.L., Stelnicki A.M., Nordstokke D.W.. Test review: Multidimensional anxiety scale for children by J. S. March. Can. J. Sch. Psychol. (2015.0) 30 70-77. DOI: 10.1177/0829573514542924 49. Buysse D.J., Reynolds III C.F., Monk T.H., Berman S.R., Kupfer D.J.. The Pittsburgh Sleep Quality Index: A new instrument for psychiatric practice and research. Psychiatry Res. (1989.0) 28 193-213. DOI: 10.1016/0165-1781(89)90047-4 50. Kredlow M.A., Capozzoli M.C., Hearon B.A., Calkins A.W., Otto M.W.. The effects of physical activity on sleep: A meta-analytic review. J. Behav. Med. (2015.0) 38 427-449. DOI: 10.1007/s10865-015-9617-6 51. Ranit M.B., Waloszek J.M., Schwartz O., Allen N.B., Trinder J.. Factor structure and psychometric properties of the Pittsburgh Sleep Quality Index in community-based adolescents. Sleep (2018.0) 41 zsy066. DOI: 10.1093/sleep/zsy066 52. Evenson K., Catellier D., Gill K., Ondrak K., McMurray R.. Calibration of two objective measures of physical activity for children. J. Sport Sci. (2008.0) 26 1557-1565. DOI: 10.1080/02640410802334196 53. Matt G.E., Rock C.L., Johnson-Kozlow M.. Using recall cues to improve measurement of dietary intakes with a food frequency questionnaire in an ethnically diverse population: An exploratory study. J. Am. Diet. Assoc. (2006.0) 106 1209-1217. DOI: 10.1016/j.jada.2006.05.009 54. Hair J.F., Black W.C., Babin B.J., Anderson R.E.. Multivariate Data Analysis (2018.0) 55. Byrne B.M.. Structural Equation Modeling with AMOS: Basic Concepts, Applications, and Programming (2010.0) 56. Stevens J.. Applied Multivariate Statistics for the Social Sciences (2002.0) 57. Cohen J.. Statistical Power Analysis for the Behavioural Sciences (1969.0) 58. Laurson K.R., Eisenmann J.C., Welk G.J.. Body fat percentile curves for US children and adolescents. Am. J. Prev. Med. (2011.0) 41 87-92. DOI: 10.1016/j.amepre.2011.06.044 59. Gouveia M.J., Canavarro M.C., Moreira H.. Associations between mindfulness, self-compassion, difficulties in emotion regulation, and emotional eating among adolescents with overweight/obesity. J. Child. Fam. Stud. (2018.0) 28 273-285. DOI: 10.1007/s10826-018-1239-5 60. De Mello M.T., de Aquino Lemos V., Antunes H.K., Bittencourt L., Santos-Silva R., Tufik S.. Relationship between physical activity and depression and anxiety symptoms: A population study. J. Affect. Disord. (2013.0) 149 241-246. DOI: 10.1016/j.jad.2013.01.035 61. Hegberg N.J., Tone E.B.. Physical activity and stress resilience: Considering those at-risk for developing mental health problems. Ment. Health Phys. Act. (2015.0) 8 1-7. DOI: 10.1016/j.mhpa.2014.10.001 62. Wang H., Fu J., Lu Q., Tao F., Hao J.. Physical activity, body mass index and mental health in Chinese adolescents: A population based study. J. Sports Med. Phys. Fit. (2014.0) 54 518-525 63. Lovato N., Gradisar M.. A meta-analysis and model of the relationship between sleep and depression in adolescents: Recommendations for future research and clinical practice. Sleep Med. Rev. (2014.0) 18 521-529. DOI: 10.1016/j.smrv.2014.03.006 64. McMakin D.L., Alfano C.A.. Sleep and anxiety in late childhood and early adolescence. Curr. Opin. Psychiatry (2015.0) 28 483. DOI: 10.1097/YCO.0000000000000204 65. Ridgers N.D., Timperio A., Cerin E., Salmon J.. Compensation of physical activity and sedentary time in primary school children. Med. Sci. Sport. Exerc. (2014.0) 46 1564-1569. DOI: 10.1249/MSS.0000000000000275 66. Allport L., Song M., Leung C.W., McGlumphy K.C., Hasson R.E.. Influence of parent stressors on adolescent obesity in African American youth. J. Obes. (2019.0) 2019 1316765. DOI: 10.1155/2019/1316765
--- title: Incidence of Concomitant Neoplastic Diseases, Tumor Characteristics, and the Survival of Patients with Lung Adenocarcinoma or Squamous Cell Lung Carcinoma in Tobacco Smokers and Non-Smokers—10-Year Retrospective Single-Centre Cohort Study authors: - Błażej Ochman - Paweł Kiczmer - Paweł Ziora - Mateusz Rydel - Maciej Borowiecki - Damian Czyżewski - Bogna Drozdzowska journal: Cancers year: 2023 pmcid: PMC10046928 doi: 10.3390/cancers15061896 license: CC BY 4.0 --- # Incidence of Concomitant Neoplastic Diseases, Tumor Characteristics, and the Survival of Patients with Lung Adenocarcinoma or Squamous Cell Lung Carcinoma in Tobacco Smokers and Non-Smokers—10-Year Retrospective Single-Centre Cohort Study ## Abstract ### Simple Summary Lung cancer is one of the biggest public health issues due to its high prevalence and mortality. Currently, increasing consideration is given to the incidence of lung cancer in the individuals with no lifetime history of tobacco smoking. However, up-to-date data on the characteristics of a group of non-smokers with lung cancer are limited. The current literature also contains gaps in the differences in the clinical course of lung cancer in smokers and non-smokers. This study aimed to investigate the differences in tumor characteristics, survival rates, and comorbidities between a group of smokers and a group of non-smokers with lung cancer. The presented results may be used in clinical practice and in shaping future lung cancer prevention programs. ### Abstract Changes in smoking trends and changes in lifestyle, together with worldwide data regarding the incidence of lung cancer in the group of patients with no previous history of smoking, leads to consideration of the differences in the course of the disease, the time of cancer diagnosis, the survival rate, and the occurrence of comorbidities in this group of patients. This study aimed to determine the occurrence of non-smokers among patients undergoing anatomical resection of the lung tissue due to lung carcinoma and to investigate the differences between the course of lung cancer, survival, and the comorbidities in the groups of patients with lung cancer depending on the history of tobacco smoking. The study included a cohort of 923 patients who underwent radical anatomical resection of the lung tissue with lung primary adenocarcinoma or squamous cell carcinoma. The Chi2 Pearson’s test, the t-test, the Mann–Whitney U test, the Kaplan–Meier method, the Log-rank test with Mantel correction, and the Cox proportional hazard model were used for data analysis. We observed a significantly higher mean age of smoking patients compared to the mean age of non-smoking patients. The coexistence of former neoplastic diseases was significantly more frequent in the group of non-smokers compared to the group of smoking patients. We did not observe differences depending on smoking status in the tumor stage, grade, vascular and pleural involvement status in the diagnostic reports. We did not observe differences in the survival between smokers vs. non-smokers, however, we revealed better survival in the non-smoker women group compared to the non-smoker men group. In conclusion, $22.11\%$ of the patients undergoing radical anatomical resection of the lung tissue due to lung cancers were non-smokers. More research on survival depending on genetic differences and postoperative treatment between smokers and non-smokers is necessary. ## 1. Introduction For decades, lung cancer has maintained its position as one of the cancers with the highest incidence rate, furthermore, it is the leading cause of cancer mortality globally. According to GLOBOCAN 2020 data, the incidence of lung cancer was $11.4\%$ with a mortality rate of $18\%$ of all cancer deaths for both sexes worldwide. Differences in incidences and mortality have been observed between women and men: incidence rates were $8.4\%$ and $14.3\%$, and cancer-related mortality was $13.7\%$ and $21.5\%$ for women and men, respectively [1]. Continuous exposure to tobacco smoke is largely to blame as the primary and predominant cause of lung cancer [2,3]. However, among lung cancer patients there is also a group of patients who have never smoked tobacco, amounting to about 10–$25\%$ depending on the examined population, geographic distribution, and methodology of the studies. Due to the high incidence of lung cancer, even a small proportion of patients with no history of smoking may represent a large group of patients [4,5]. Considering the generally decreasing prevalence of smoking worldwide [6,7,8], lung cancer risk factors other than smoking itself and lung cancer non-smoking patients’ characteristics should be under the spotlight, both now and in the future. The relationship between the increased incidence of lung cancer in non-smokers with a family history of cancer, as observed in many studies, seems to support the considered significant role of the influence of genetic factors on the incidence of lung cancer in non-smokers [9,10]. Independent risk factors for the development of lung cancer in non-smokers, in addition to environmental air pollution [11,12] and passive smoking [13], also include previous respiratory diseases, such as asthma, pneumonia, tuberculosis, chronic bronchitis, and viral infections [14,15]. Besides considered genetic factors, obesity and diet may also be especially important in the development of lung adenocarcinoma [16,17]. That may seem even more important due to the higher incidence of adenocarcinoma in comparison to the other histological subtypes of lung cancer in the group of never-smokers [18]. In our study, we analyzed the data of patients operated on for lung cancer in 2012–2021. The analyzed data included the smoking history, gender, concomitant neoplastic diseases other than lung cancer, comorbidities, family cancer history, tumor stage, grade, histologic type, lymphovascular, and pleural invasion, and the patients’ survival data. Currently, the characteristics of non-smokers with lung cancer, including the survival, course differences, tumor stage, and grade at diagnosis are both heterogeneous and limited. Due to the lack of data on smoking in most cancer registries, it is maintained that the incidence of lung cancer in never-smokers has not yet been thoroughly and comprehensively investigated [19]. This study aimed to determine the percentage of non-smokers among patients undergoing anatomical resection of the lung tissue due to lung carcinoma and to investigate the differences between the course of lung cancer, survival, and the comorbidities in the groups of patients with lung cancer depending on the history of tobacco smoking. Additionally, the study compared risk factors and outcomes between smokers and non-smokers. The term non-smokers refers to persons who have smoked fewer than 100 cigarettes in their lifetime, including lifetime non-smokers [20]. ## 2.1. Study Design A study was performed on a cohort of 923 patients who underwent radical anatomical resection of the lung tissue (segmentectomy, lobectomy, bilobectomy or pneumonectomy) due to lung cancer between May 2012 and December 2021. The medical records of all patients operated on due to lung cancer in our center were analyzed in detail and collected in a dedicated database. The database collected data on the previous medical history, exposure to harmful environmental factors and stimulants, family cancer history, precise assessment of the stage of cancer along with the exact diagnosis of the type of lung cancer determined on the basis of the postoperative histopathological examination and the results of perioperative care. In addition, each of the operated patients remained under the supervision of the outpatient clinic, which enabled the assessment of long-term treatment results. The day of the surgical procedure was the starting point of the observation. The observation of patient survival was conducted up to five years after the surgery. Data about the patients’ survival were collected up to 1 May 2022. All further outcomes were considered incomplete. The inclusion criteria were: histopathologically confirmed primary adenocarcinoma or squamous cell carcinoma. The exclusion criteria were: age < 18 years old, histopathologically confirmed adenosquamous carcinoma, secondary lung neoplasm confirmed histopathologically, occurrence of more than one histologically different tumor in post-operative material. The detailed study design is presented in Figure 1 below. The mean age of patients in the cohort was 66.8 +/− 7.54 years (range 38–81 years). There were 403 women ($43.66\%$) and 520 men ($56.33\%$). In total, 204 of the patients were non-smokers ($22.11\%$, $95\%$ CI: 19.43–$24.79\%$). Further information about the group is presented in the results section. The study was approved by Bioethics Committee of Medical University of Silesia in Katowice ## 2.2. Statistical Analysis The data was presented as the number of cases with percent value for categorical variables and mean +/− SD for quantitative variables. The normality assumption was tested for each quantitative variable based on a graphical interpretation of the Q–Q plot and histogram. The odds ratios with a $95\%$ confidence interval were calculated for the categorical variables. To determine the differences between categorical variables, the Chi2 Pearson’s test was performed; for qualitative variables we used the t-test for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables. The Kaplan–Meier method was used to determine the survival probability among groups. Comparison of survival was performed using the Log-rank test with Mantel correction in case of comparing more than 2 groups. To assess the influence of more than one variable on patients’ survival, Cox proportional hazard model was performed. p values lower than 0.05 were considered significant. Analysis was performed using the R language in Rstudio software. ## 3.1. Age of Patients and Concomitant Neoplastic Diseases The mean age of smoking patients was significantly higher than non-smokers (Table 1). No difference in the occurrence of neoplasms among family members was found. We found that non-active smoking individuals were suffering from other neoplastic diseases compared to active smokers (Table 2). ## 3.2. Tumor Characteristics We found no significant differences in the tumor stage, grade, vessel and pleural involvement status, nor tumor size (Table 3 and Table 4). However, adenocarcinoma was statistically a more frequent tumor among non-smokers compared to smokers. ## 3.3. Survival Analysis For purposes of survival analysis, only patients with R0 resection status were included. No significant differences in survival rate were found among the groups of non-smokers vs. smokers (Table 5, Figure 2). No significant differences in survival between cancer histologic subtypes were found, however we found significantly better survival among non-smoking women compared to non-smoking men (Figure 3; Table 6). The Cox proportional hazard model including the age, gender, smoking status and histologic type showed a weak influence of age on survival rate (HR = 1.01); male gender had 1.29 HR at $$p \leq 0.09.$$ The model consisting of only gender and smoking status showed a significant influence of male gender on the risk of death (Table 7). ## 4. Discussion In our single-center study of 204 non-smokers and 719 smokers, we investigated the characteristics of lung cancer among non-smokers (LCNS) and their comparison to lung cancer among smokers (LCS) with adenocarcinoma and squamous cell carcinoma. In the studied groups, we observed differences in the age of diagnosis, the histological subtype, and the incidence of other neoplastic diseases. Our analysis revealed significantly better survival among non-smoking women compared to non-smoking men. Moreover, it should be noted that we did not observe significant differences in the survival time of the patients with LCNS compared to LCS at the specified time intervals and between the TNM scale and tumor sizes, the other examined pathomorphological parameters. In a study of two American Cancer Society Cancer Prevention Study, cohorts of 940,000 patients who died of lung cancer with no history of smoking in 1959–2000, men showed higher mortality than women in the study group, 17.1 and 14.7 per 100,000 people per year in cohort two, respectively. In both studied cohorts, there was a significantly higher proportion of women in the group of non-smokers with lung cancer than men (382,854 vs. 94,041 in cohort I (1959–1972) and 341,643 vs. 122,563 in cohort II (1982–2000), respectively [21]). The incidence of lung cancer in non-smoking men and women was from 4.8 to 13.7 cases per 100,000 people per year and from 14.4 to 20.8 per 100,000 people per year, respectively, in the analysis of six cohorts mainly from Sweden and the USA in 1971–2002. Therefore, the obtained results supported the increased incidence of lung cancer in non-smoking women compared to non-smoking men observed in other studies [22]. Non-smoking individuals were $22.11\%$, $95\%$ (19.43–$24.79\%$) of all patients undergoing lobectomy due to primary adenocarcinoma or squamous cell carcinoma. Bade and Cruz reported in their study that $25\%$ [1] of worldwide lung cancer cases occur among non-smokers [20]. These estimations are similar to our results. Studies report that in certain parts of the world, like Asia, non-smoking lung cancer tends to occur more frequently among women compared to men. Our study did not show any difference in the gender distribution among groups, both consisted of slightly fewer women than men. Our analysis revealed the increased incidence and risk of neoplastic diseases in the LCNS group (OR: 1.68, $95\%$ CI: 1.13, $$p \leq 0.013$$) compared to the LCS group. Factors that could play a key role in the formation of lung cancer in non-smokers are similar to the factors considered to be associated with the higher prevalence of the other cancers. An analysis of 183,248 patients showed a significant impact of the metabolic syndrome (MS) on the formation of various cancers and the increased risk of lung cancer (OR: 1.11, $95\%$ CI: 1.05–1.16) and both pre-menopausal endometrial cancer and post-menopausal endometrial cancer (respectively: OR: 2.14, $95\%$ CI: 1.74–2.65, and OR: 2.46, $95\%$ CI: 2.20–2.74) [23]. To our knowledge, there is currently no relevant data on the comorbidity of neoplastic diseases in that group of patients. However, the occurrence of mutations and polymorphisms of genes playing an essential role in the process of carcinogenesis in the lungs and other sites, as shown in the current literature, such as EGFR, ALK, TP53, BRCA$\frac{1}{2}$, YAP1 [24,25,26], seems to make the coexistence of the other cancers in this group of patients less unexpected. In our study, lung adenocarcinoma (LUAD) was more common than lung squamous cell carcinoma (LSCC) in non-smokers ($62.25\%$ vs. $37.75\%$). This is consistent with the results of most studies, where lung adenocarcinoma (LUAD) was the most common lung cancer among non-smokers [18,21,27,28,29,30]. The reason for the incidence of LUAD as the most common histological subtype in non-smokers may be the gene mutations observed more often in non-smokers with lung cancer. These mutations drive the development of this histological subtype [24,31], as well as the increased occurrence and intensity of predisposing factors to the development of adenocarcinoma in this group of patients, such as passive smoking [32,33] and obesity, which were more often observed in the group of non-smokers with LUAD compared to smokers, although the significant impact of the considered factors on the development of adenocarcinoma remains ambiguous [34,35,36]. Obtained results differ in the incidence rates of adenocarcinoma and squamous cell carcinoma within the study groups of non-smokers and smokers, which is probably due to the differences in group sizes and ethnic races between the other studies, as well as other histological cancer subtypes included in the studies [21,27,28,29,30]. The effect of lifetime smoking on lung cancer survival is ambiguous. A great number of studies have shown a significantly longer survival time from the moment of lung cancer diagnosis in non-smokers compared to smokers. A large, 13-cohort study spanning lung cancer patients from 1960 to 2004 revealed among European descent patients a significantly higher mortality rate in the group of smokers compared to the group of non-smokers. Th lung cancer death rates of lifetime smoker men were 21.9 times higher than never-smoker men, and for lifetime smoker women cancer death rates were 13.7 times higher compared to never-smoker women [37]. A study conducted in the Czech Republic in 2021 among 2439 lung cancer patients found that non-smoking patients were diagnosed with lung cancer at a later stage, but non-smokers had better survival rates than smokers [38]. Another analysis of 3380 smokers and 334 never-smokers diagnosed with lung cancer in 2003–2016 showed significantly higher overall survival among never-smokers compared to smokers; the 5-year survival rate was higher in never-smokers compared to smokers ($57.9\%$ vs. $42.6\%$, $$p \leq 0.05$$) [39]. However, there are also studies in which, similarly to the results of our study, no significant differences in the survival rates between the groups of smokers and non-smokers were observed [40,41]. No meaningful differences were observed in the overall survival in the 1-, 2-, and 5-year observation times in our study between the smokers and the never-smokers. However, our study groups differed from the groups in other studies due to the criteria for the inclusion and exclusion of patients from the study we had established. The differences in the observed survival times within groups of patients depending on the history of smoking in various studies and within the results of our study may be due to the differences in the number of non-smokers in the study group, the incidence, and the histological subtypes of lung cancer, the different proportions in staging in both groups of patients, and the various treatments after diagnosis. In our analysis, the percentage of non-smokers was $22.11\%$. We did not observe significant differences in staging and grading between the smokers vs. the non-smokers groups. In the group of non-smokers, a large proportion of the patients ($58.83\%$) had staging I (IA2, IA3 and IB), similar to the group of smoking patients ($56.33\%$). That makes our study different from Subramanian et al. which also showed no differences in the survival of patients from both groups, but in the study group as many as $62.5\%$ of patients had stage III or IV [41]. Similar to the study by Nemesure et al., stage III and IV in the group of smokers and non-smokers amounted to $65.5\%$ and $61.4\%$, respectively, which showed significant differences in the survival of patients with lung cancer depending on the smoking status [39]. Our study, which differed from other studies with the participation of patients in individual groups with different staging, allowed us to conclude the differences in survival of patients with a lower stage because the group we examined contained a smaller percentage of patients with staging III and IV compared to the cited studies, being $17.15\%$ and $15.58\%$, respectively, in the group non-smokers and smokers. This resulted directly from the characteristics of the study group, which consisted of patients qualified for the surgical treatment of lung cancer, i.e., patients at stage IA1–IIIA, and cases of higher stages of the disease were incidental. Even though the study group contained an equal distribution of staging, the lack of differences we observed in terms of the survival of non-smokers and smokers seemed to provide new information in light of previous studies, which was particularly important in the group of patients with low staging, who were diagnosed with adenocarcinoma and squamous cell carcinoma. It seems that non-smokers should have a longer survival time due to the expected lower number of comorbidities that increase the risk of postoperative complications and worsen the prognosis. In our study, we found improved survival among the non-smoking women compared to the men; no differences in survival between the smokers and the non-smokers were found. Apart from the survival time itself, we did not have reliable information on the patients’ further post-operative status regarding smoking status after the procedure. An important limitation of our study in terms of survival results was also the lack of information about the percentage of smokers who continued to use tobacco after the procedure. We also had limited information on passive smoking, exposure to mutagenic risk factors, and mutations in the genes involved in lung cancer in non-smokers. The lack of reliable information on the pathogenetic background of the studied subtypes of lung cancer in non-smokers, which may have an impact on survival, may also be a limitation of the results presented in the study, allowing the impact of possible heterogeneity of lung cancer patients in the group of non-smokers to effect the obtained results in terms of survival. A large amount of data regarding lung cancer in non-smokers remain unclear, although the results obtained by various centers around the world provide a lot of valuable information. In the future, this information may contribute to a more accurate characterization of the LCNS, creating the basis for shaping future lung cancer prevention programs including a group of patients with precisely defined risk factors, and also increasing sensitivity to this group of patients in clinical practice. To our knowledge, the results of our study may contribute to future more detailed research on the observed increased incidence of comorbidity of neoplastic diseases other than lung cancer in the LCNS group patients, further indicating no significant differences in the detection of lung cancer in this group of patients compared to a group of smoking patients. The observed lack of difference in the survival rates may further illuminate the importance of the problem now posed by LCNS, ranked seventh as a stand-alone malignant disease in the incidence of malignant tumors in general, only improving in importance due to the decrease in smoking trends and the increase in the importance of factors potentially predisposing to the development of LCNS. The selected patients with adenocarcinoma and squamous cell carcinoma seemed to reflect well the characteristics of the entire group of patients with NSCLC, which is the most common group of lung cancer tumors not only in the group of non-smokers but also smokers. A significant share of non-smokers in the study group (over $20\%$) compared to other current studies, with a comparable median age in both groups of patients, was undoubtedly a strength of our study. Unfortunately, we did not include in our analysis some important risk factors for the development of LCNS, such as passive smoking or obesity, due to the lack of reliable information in this regard. It remains a topic that requires deeper research. As in the case of information on the impact of environmental pollution and occupational exposure on factors predisposing to the occurrence of lung cancer, these factors are difficult to measure objectively and reliably. In our study, in regard to the environmental factors, our patients were diagnosed and operated on in one center, living in an area with similar air pollution. The study group, including 923 patients with adenocarcinoma and squamous cell carcinoma, in the light of the selected 10-year period, seemed to be well selected in terms of time, due to the changing standards of oncological diagnostics and treatment over the years. The homogeneity of the study group in terms of ethnicity seemed to provide a better reference of the results to this ethnic group in which modifiable risk factors for lung cancer in non-smokers seemed to increase. The limitation of our study was the lack of information about the further treatment of the patients. Additionally, no relevant information about passive smoking was obtained from the patients. ## 5. Conclusions In conclusion, $22.11\%$ of the patients undergoing radical anatomical resection of the lung tissue due to lung cancers were non-smokers. There were no significant differences in the stage of disease nor between the tumor grade among groups. In non-smokers with lung cancer, other cancers were more common. Adenocarcinoma tended to be more frequent among non-smokers than squamous cell carcinoma. Non-smoking women had a significantly better survival rate compared to non-smoking men. No differences in survival rate were found between the smokers group compared to the non-smokers group. The differences between lung cancer patients with and without a smoking history should be further explored to better establish future lung cancer screening and prevention programs for non-smokers. ## References 1. Sung H., Ferlay J., Siegel R.L., Laversanne M., Soerjomataram I., Jemal A., Bray F.. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. (2021) 71 209-249. DOI: 10.3322/caac.21660 2. Gandini S., Botteri E., Iodice S., Boniol M., Lowenfels A.B., Maisonneuve P., Boyle P.. Tobacco smoking and cancer: A meta-analysis. Int. J. Cancer (2008) 122 155-164. DOI: 10.1002/ijc.23033 3. Akhtar N., Bansal J.G.. Risk factors of Lung Cancer in nonsmoker. Curr. Probl. Cancer (2017) 41 328-339. DOI: 10.1016/j.currproblcancer.2017.07.002 4. Parkin D.M., Bray F., Ferlay J., Pisani P.. Global Cancer Statistics, 2002. CA Cancer J. Clin. (2005) 55 74-108. DOI: 10.3322/canjclin.55.2.74 5. Sun S., Schiller J.H., Gazdar A.F.. Lung cancer in never smokers—A different disease. Nat. Rev. Cancer (2007) 7 778-790. DOI: 10.1038/nrc2190 6. Vaičiūnas T., Žemaitaitytė M., Lange S., Štelemėkas M., Oja L., Petkevičienė J., Kowalewska A., Pudule I., Piksööt J., Šmigelskas K.. Trends in Adolescent Substance Use: Analysis of HBSC Data for Four Eastern European Countries, 1994–2018. Int. J. Environ. Res. Public Health (2022) 19. DOI: 10.3390/ijerph192315457 7. Creedon T.B., Wayne G.F., Progovac A.M., Levy D.E., Cook B.L.. Trends in cigarette use and health insurance coverage among US adults with mental health and substance use disorders. Addiction (2022) 118 353-364. DOI: 10.1111/add.16052 8. Yang H., Ma C., Zhao M., Magnussen C.G., Xi B.. Prevalence and trend of smokeless tobacco use and its associated factors among adolescents aged 12–16 years in 138 countries/territories, 1999–2019. BMC Med. (2022) 20. DOI: 10.1186/s12916-022-02662-0 9. Coté M.L., Liu M., Bonassi S., Neri M., Schwartz A.G., Christiani D.C., Spitz M.R., Muscat J.E., Rennert G., Aben K.K.. Increased risk of lung cancer in individuals with a family history of the disease: A pooled analysis from the International Lung Cancer Consortium. Eur. J. Cancer (2012) 48 1957-1968. DOI: 10.1016/j.ejca.2012.01.038 10. Gu J., Zhao H.. Systematic review of the relationship between family history of malignant tumor and lung cancer risk. Chin. J. Prev. Control. Chronic Dis. (2010) 18 358-361 11. Hosgood H.D., Boffetta P., Greenland S., Lee Y.-C.A., McLaughlin J., Seow A., Duell E.J., Andrew A.S., Zaridze D., Szeszenia-Dabrowska N.. In-Home Coal and Wood Use and Lung Cancer Risk: A Pooled Analysis of the International Lung Cancer Consortium. Environ. Health Perspect. (2010) 118 1743-1747. DOI: 10.1289/ehp.1002217 12. Loomis D., Grosse Y., Lauby-Secretan B., El Ghissassi F., Bouvard V., Benbrahim-Tallaa L., Guha N., Baan R., Mattock H., Straif K.. The carcinogenicity of outdoor air pollution. Lancet Oncol. (2013) 14 1262-1263. DOI: 10.1016/S1470-2045(13)70487-X 13. Hackshaw A.K.. Lung cancer and passive smoking. Stat. Methods Med. Res. (1998) 7 119-136. DOI: 10.1177/096228029800700203 14. Brenner D.R., Boffetta P., Duell E.J., Bickeböller H., Rosenberger A., McCormack V., Muscat J.E., Yang P., Wichmann H.-E., Brueske-Hohlfeld I.. Previous Lung Diseases and Lung Cancer Risk: A Pooled Analysis from the International Lung Cancer Consortium. Am. J. Epidemiol. (2012) 176 573-585. DOI: 10.1093/aje/kws151 15. Robinson L.A., Jaing C.J., Pierce Campbell C., Magliocco A., Xiong Y., Magliocco G., Thissen J.B., Antonia S.. Molecular evidence of viral DNA in non-small cell lung cancer and non-neoplastic lung. Br. J. Cancer (2016) 115 497-504. DOI: 10.1038/bjc.2016.213 16. Pan S.Y., Johnson K.C., Ugnat A.-M., Wen S.W., Mao Y.. Association of Obesity and Cancer Risk in Canada. Am. J. Epidemiol. (2004) 159 259-268. DOI: 10.1093/aje/kwh041 17. Hosseini M., Naghan P.A., Jafari A.M., Yousefifard M., Taslimi S., Khodadad K., Mohammadi F., Sadr M., Rezaei M., Mortaz E.. Nutrition and lung cancer: A case control study in Iran. BMC Cancer (2014) 14. DOI: 10.1186/1471-2407-14-860 18. Li D., Shi J., Dong X., Liang D., Jin J., He Y.. Epidemiological characteristics and risk factors of lung adenocarcinoma: A retrospective observational study from North China. Front. Oncol. (2022) 12 892571. DOI: 10.3389/fonc.2022.892571 19. Rivera G.A., Wakelee H.. Lung Cancer in Never Smokers. Adv. Exp. Med. Biol. (2015) 893 43-57. DOI: 10.1007/978-3-319-24223-1_3 20. Bade B.C., Cruz C.S.D.. Lung Cancer 2020: Epidemiology, Etiology, and Prevention. Clin. Chest Med. (2020) 41 1-24. DOI: 10.1016/j.ccm.2019.10.001 21. Thun M.J., Henley S.J., Burns D., Jemal A., Shanks T.G., Calle E.E.. Lung Cancer Death Rates in Lifelong Nonsmokers. JNCI J. Natl. Cancer Inst. (2006) 98 691-699. DOI: 10.1093/jnci/djj187 22. Wakelee H.A., Chang E.T., Gomez S.L., Keegan T.H., Feskanich D., Clarke C.A., Holmberg L., Yong L.C., Kolonel L.N., Gould M.K.. Lung Cancer Incidence in Never Smokers. J. Clin. Oncol. (2007) 25 472-478. DOI: 10.1200/JCO.2006.07.2983 23. López-Jiménez T., Duarte-Salles T., Plana-Ripoll O., Recalde M., Xavier-Cos F., Puente D.. Association between metabolic syndrome and 13 types of cancer in Catalonia: A matched case-control study. PLoS ONE (2022) 17. DOI: 10.1371/journal.pone.0264634 24. Tessema M., Rossi M.R., Picchi M.A., Yingling C.M., Lin Y., Ramalingam S.S., Belinsky S.A.. Common cancer-driver mutations and their association with abnormally methylated genes in lung adenocarcinoma from never-smokers. Lung Cancer (2018) 123 99-106. DOI: 10.1016/j.lungcan.2018.07.011 25. de Alencar V.T.L., Figueiredo A.B., Corassa M., Gollob K.J., de Lima V.C.C.. Lung cancer in never smokers: Tumor immunology and challenges for immunotherapy. Front. Immunol. (2022) 13 984349. DOI: 10.3389/fimmu.2022.984349 26. Hu X., Yang D., Li Y., Li L., Wang Y., Chen P., Xu S., Pu X., Zhu W., Deng P.. Prevalence and clinical significance of pathogenic germline BRCA1/2 mutations in Chinese non-small cell lung cancer patients. Cancer Biol. Med. (2019) 16 556-564. DOI: 10.20892/j.issn.2095-3941.2018.0506 27. Wu Z., Tan F., Yang Z., Wang F., Cao W., Qin C., Dong X., Zheng Y., Luo Z., Zhao L.. Sex disparity of lung cancer risk in non-smokers: A multicenter population-based prospective study based on China National Lung Cancer Screening Program. Chin. Med. J. (2022) 135 1331-1339. DOI: 10.1097/CM9.0000000000002161 28. Kawaguchi T., Takada M., Kubo A., Matsumura A., Fukai S., Tamura A., Saito R., Kawahara M., Maruyama Y.. Gender, Histology, and Time of Diagnosis Are Important Factors for Prognosis: Analysis of 1499 Never-Smokers with Advanced Non-small Cell Lung Cancer in Japan. J. Thorac. Oncol. (2010) 5 1011-1017. DOI: 10.1097/JTO.0b013e3181dc213e 29. Pinheiro P.S., Callahan K.E., Medina H.N., Koru-Sengul T., Kobetz E.N., Gomez S.L., Lopes G.D.L.. Lung cancer in never smokers: Distinct population-based patterns by age, sex, and race/ethnicity. Lung Cancer (2022) 174 50-56. DOI: 10.1016/j.lungcan.2022.10.009 30. Krishnamurthy A., Das A., Ramshankar V., Sagar T., Swaminathan R.. The increasing challenge of never smokers with adenocarcinoma lung: Need to look beyond tobacco exposure. Indian J. Cancer (2017) 54 172. DOI: 10.4103/ijc.IJC_33_17 31. Zhang Y., Sun Y., Pan Y., Li C., Shen L., Li Y., Luo X., Ye T., Wang R., Hu H.. Frequency of Driver Mutations in Lung Adenocarcinoma from Female Never-Smokers Varies with Histologic Subtypes and Age at Diagnosis. Clin. Cancer Res. (2012) 18 1947-1953. DOI: 10.1158/1078-0432.CCR-11-2511 32. Samet J.M., Avila-Tang E., Boffetta P., Hannan L.M., Olivo-Marston S., Thun M.J., Rudin C.M.. Lung Cancer in Never Smokers: Clinical Epidemiology and Environmental Risk Factors. Clin. Cancer Res. (2009) 15 5626-5645. DOI: 10.1158/1078-0432.CCR-09-0376 33. Spitz M.R., Gorlov I.P., Amos C.I., Dong Q., Chen W., Etzel C.J., Gorlova O.Y., Chang D.W., Pu X., Zhang D.. Variants in Inflammation Genes Are Implicated in Risk of Lung Cancer in Never Smokers Exposed to Second-hand Smoke. Cancer Discov. (2011) 1 420-429. DOI: 10.1158/2159-8290.CD-11-0080 34. Murphy R.A., Darvishian M., Qi J., Chen Y., Chu Q., Vena J., Dummer T.J.B., Le N., Sweeney E., DeClercq V.. Lifestyle factors and lung cancer risk among never smokers in the Canadian Partnership for Tomorrow’s Health (CanPath). Cancer Causes Control. (2022) 33 913-918. DOI: 10.1007/s10552-022-01566-x 35. Yang Y., Dong J., Sun K., Zhao L., Zhao F., Wang L., Jiao Y.. Obesity and incidence of lung cancer: A meta-analysis. Int. J. Cancer (2013) 132 1162-1169. DOI: 10.1002/ijc.27719 36. Zhu C.-Y., Qu J.-C., Cao H.-X., Chen G.-Y., Shi Y.-H., Fan J.-G.. Obesity and nonalcoholic fatty liver disease associated with adenocarcinoma in patients with lung cancer. Medicine (2019) 98 e17098. DOI: 10.1097/MD.0000000000017098 37. Thun M.J., Hannan L.M., Adams-Campbell L.L., Boffetta P., Buring J.E., Feskanich D., Flanders W.D., Jee S.H., Katanoda K., Kolonel L.N.. Lung Cancer Occurrence in Never-Smokers: An Analysis of 13 Cohorts and 22 Cancer Registry Studies. PLoS Med. (2008) 5. DOI: 10.1371/journal.pmed.0050185 38. Venclicek O., Skrickova J., Brat K., Fischer O., Havel L., Hrnciarik M., Marel M., Opalka P., Krakorova G., Rozsivalova D.. Lung Cancer in Non-smokers in Czech Republic: Data from LUCAS Lung Cancer Clinical Registry. Anticancer Res. (2021) 41 5549-5556. DOI: 10.21873/anticanres.15368 39. Nemesure B., Albano D., Nemesure A.. Short- and long-term survival outcomes among never smokers who developed lung cancer. Cancer Epidemiol. (2021) 75 102042. DOI: 10.1016/j.canep.2021.102042 40. Toh C.-K., Wong E.-H., Lim W.-T., Leong S.-S., Fong K.-W., Wee J., Tan E.-H.. The Impact of Smoking Status on the Behavior and Survival Outcome of Patients with Advanced Non-small Cell Lung Cancer: A Retrospective Analysis. Chest (2004) 126 1750-1756. DOI: 10.1378/chest.126.6.1750 41. Subramanian J., Velcheti V., Gao F., Govindan R.. Presentation and Stage-Specific Outcomes of Lifelong Never-smokers with Non-small Cell Lung Cancer (NSCLC). J. Thorac. Oncol. (2007) 2 827-830. DOI: 10.1097/JTO.0b013e318145af79
--- title: Maximal Exercise Improves the Levels of Endothelial Progenitor Cells in Heart Failure Patients authors: - Suiane Cavalcante - Sofia Viamonte - Rui S. Cadilha - Ilda P. Ribeiro - Ana Cristina Gonçalves - João Sousa-Venâncio - Marisol Gouveia - Manuel Teixeira - Mário Santos - José Oliveira - Fernando Ribeiro journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10046939 doi: 10.3390/cimb45030125 license: CC BY 4.0 --- # Maximal Exercise Improves the Levels of Endothelial Progenitor Cells in Heart Failure Patients ## Abstract The impact of exercise on the levels of endothelial progenitor cells (EPCs), a marker of endothelial repair and angiogenesis, and circulating endothelial cells (CECs), an indicator of endothelial damage, in heart failure patients is largely unknown. This study aims to evaluate the effects of a single exercise bout on the circulating levels of EPCs and CECs in heart failure patients. Thirteen patients with heart failure underwent a symptom-limited maximal cardiopulmonary exercise test to assess exercise capacity. Before and after exercise testing, blood samples were collected to quantify EPCs and CECs by flow cytometry. The circulating levels of both cells were also compared to the resting levels of 13 volunteers (age-matched group). The maximal exercise bout increased the levels of EPCs by $0.5\%$ [$95\%$ Confidence Interval, 0.07 to $0.93\%$], from 4.2 × 10−3 ± 1.5 × 10−$3\%$ to 4.7 × 10−3 ± 1.8 × 10−$3\%$ ($$p \leq 0.02$$). No changes were observed in the levels of CECs. At baseline, HF patients presented reduced levels of EPCs compared to the age-matched group ($$p \leq 0.03$$), but the exercise bout enhanced circulating EPCs to a level comparable to the age-matched group (4.7 × 10−3 ± 1.8 × 10−$3\%$ vs. 5.4 × 10−3 ± 1.7 × 10−$3\%$, respectively, $$p \leq 0.14$$). An acute bout of exercise improves the potential of endothelial repair and angiogenesis capacity by increasing the circulating levels of EPCs in patients with heart failure. ## 1. Introduction Heart failure (HF) is a clinical syndrome characterized by structural and functional alterations in the cardiovascular system [1]. Patients with HF present abnormal hemodynamic alterations, such as increased intracardiac pressures and/or depressed cardiac output, which can be evidenced during physical efforts and/or at rest [1]. The endothelium plays a paramount role in hemodynamic control and vascular function [2,3]. Endothelial dysfunction is related to the progression of HF [4], being an independent predictor for HF [5] and a potential treatment target [6]. In a healthy endothelium, there is a delicate balance between endothelial injury (e.g., assessed by the levels of endothelial damage biomarkers, such as circulating endothelial cells (CECs)) and the endogen repair capacity (e.g., assessed by the circulating levels of pro-angiogenic and endothelial repair/maintenance factors, such as endothelial progenitor cells (EPCs)) [7]. The EPCs are bone-marrow-derived cells with the potential to migrate and differentiate in mature endothelial cells. EPCs are attracted to sites of endothelial damage, contributing to endothelial repair, maintenance, and angiogenesis [8]. Indeed, the EPCs can also be described through their biological properties and the time of growth in vivo culture, namely as early EPCs and late EPCs. Early EPCs are known to participate in the formation of vessels through paracrine mechanisms, while late EPCs participate in endothelial tubulogenesis [9]. The number of EPCs is positively related to endothelial function [10] and has been used for the assessment of endothelial dysfunction [4]. Functionality and levels of EPCs are reduced in HF patients [11,12] and tend to decline with aging [13]. Indeed, reduced circulating levels of EPCs are related to increasing hazard ratios for all-cause mortality and cardiovascular death, independently of N-terminal pro B-type natriuretic peptide (NT-proBNP) levels [14], and circulating levels of EPCs are considered a strong and independent predictor of mortality in HF [15,16]. The CECs are mature endothelial cells that have detached from the endothelial layer of the vessel after endothelial injury [17], being the most direct cellular marker of endothelial damage [18,19,20]. CECs are present at low levels in healthy subjects [21], whereas high levels are present in cardiovascular diseases [22], including HF [21,22]. Recently, CECs were proposed as a diagnostic biomarker for HF with preserved ejection fraction (HFpEF) [23], which highlights the potential of the CEC count for clinical settings. Physical exercise can improve exercise capacity and decrease HF hospitalizations, being strongly recommended in the treatment of patients with HF [1]. One of the benefits of exercise training among patients with HF seems to be the increased mobilization of EPCs from bone marrow [24,25,26]. A growing number of studies have investigated the influence of isolated exercise sessions, namely maximal exercise, in the mobilization of EPCs in HF patients [27,28,29,30,31,32]. However, none considered the additional analysis of the effects of a maximal exercise bout on endothelial damage indicators. The optimization of strategies that could increase the endogenous mobilization of EPCs without inducing endothelial damage is of undeniable clinical importance. Thus, the present study aims to evaluate the impact of a maximal exercise bout on the levels of EPCs and CECs in HF patients. We hypothesized that a maximal exercise bout increases the levels of EPCs to those of age-matched individuals. Therefore, we assessed the circulating levels of EPCs and CECs in HF patients before and after a cardiopulmonary exercise test (CPET) and compared it with the resting circulating levels of age-matched adults free from cardiovascular disease. ## 2.1. Participants and Design of the Study Thirteen patients with chronic HF were recruited from the cardiac rehabilitation program of the North Rehabilitation Center, Vila Nova de Gaia, Portugal. The eligibility was assessed by a clinician based on the following inclusion criteria: patients with HF according to the European Society of Cardiology (ESC) guidelines [1], age ≥ 45 years old; New York Heart Association (NYHA) class I–III; clinically stable. All patients were referred to the cardiac rehabilitation program after an HF diagnosis. In brief, the diagnosis was established after the assessment of clinical history, physical examination, brain natriuretic peptide evaluation, electrocardiogram, and echocardiography to confirm structural and/or functional alterations of the heart [1]. Participants with lung disease, peripheral artery disease, heart transplantation and/or exercise-limiting orthopedic disabilities were excluded. Additionally, a group of 13 age-matched adults was recruited to serve as a reference group for baseline EPC and CEC levels. The reference group was composed of adults with cardiovascular risk factors referred by their physicians for participation in a primary prevention program. The inclusion criteria were the presence of cardiovascular risk factors but free from cardiovascular disease; age ≥ 45 years old. The exclusion criteria were the same as for the HF group. The clinical and sociodemographic data were retrieved from the clinical files and confirmed with the participants. Body weight, height, and the level of adherence to the Mediterranean Diet were assessed before the CPET. Adherence to the Mediterranean Diet was assessed with the 14-Item Mediterranean Diet Adherence Screener (MEDAS); the result of the MEDAS was obtained by summing the score (0 or 1) assigned to each question (0–14). A total score ≤ 5 indicates “weak adherence”, 6 to 9 “moderate to fair adherence,” and ≥10 “good or very good adherence” to the Mediterranean diet [33]. Blood collection was performed before and 30 min after the completion of the CPET to assess the number of EPCs and CECs. All the participants provided written informed consent. The study procedures were in accordance with the Declaration of Helsinki and the study was approved by the Ethics Committee for Social and Health fields of the Santa Casa da Misericórdia do Porto (the entity that at that time was running the North Rehabilitation Center) (Ata No. 28, 28 March 2017). ## 2.2. Cardiopulmonary Exercise Test Patients underwent a symptom-limited maximal CPET in the afternoon (between 2 and 4 pm), using a calibrated electronic treadmill (Bruce protocol). Patients were instructed to take all medications on the test day and were stimulated to exercise until exhaustion. Briefly, the cardiac rhythm was assessed by a 12-lead electrocardiogram throughout the CPET. Blood pressure was recorded at rest and during the CPET. Ventilation (VE), oxygen uptake (VO2), and carbon dioxide output (VCO2) were measured breath-by-breath using the CS-200 gas analyzer Ergo Spiro (Schiller, Baar, Switzerland). The VO2 peak (mL/kg/min) was considered the highest value reached at the end of the test, and the respiratory exchange ratio was registered for the evaluation of the level of effort. The VE/VCO2 slope was calculated by automatic linear regression fitting the relationship between VE and VCO2. The duration of the exercise test (min) was recorded. All the included HF patients completed the symptom-limited maximal CPET. ## 2.3. Quantification of Circulating Number of EPCs and CECs Blood samples (3 mL) for cytometry analysis were collected into ethylenediaminetetraacetic acid (EDTA) tubes and treated, according to the manufacturer’s instructions, with TransFix (Cytomark, Caltag Medsystems Ltd., Buckingham, UK) at a 1:5 ratio immediately after collection. Transfix can stabilize cell populations and permits blood analysis for up to seven days after blood collection [34]. Blood samples were collected in the afternoon between 2 and 4 pm (patients were not fasting), before and 30 min after the CPET, and stored in the dark at room temperature. The flow cytometry analysis (FACS-Calibur flow cytometer, Becton Dickinson, San Jose, CA, USA) was performed two to three days after the blood collection. Staining and analysis were performed using a protocol adapted from Ahmed, et al. [ 35], as previously reported [11]. For the quantitative assessment of circulating EPCs and CECs by flow cytometry, whole blood samples were incubated for 10 min with FcR-blocking reagent to block unwanted binding of antibodies to human Fc receptor-expressing cells. All staining procedures were executed at room temperature. Samples were incubated with BV410 CD34 (BD Horizon, BD, Franklin Lakes, NJ, USA), PE CD309 (VEGFR-2/KDR; BD Pharmingen, BD, Franklin Lakes, NJ, USA), FITC CD144 (BD Pharmingen, BD, Franklin Lakes, NJ, USA), BV510 CD45 (BD Horizon, BD, Franklin Lakes, NJ, USA), and APC CD$\frac{133}{1}$ (Miltenyi Biotec, Cologne, Germany), according to manufacturer’s instructions. After erythrocyte lysis, at least 500,000 CD45+ and a minimum of 100 CD34+ cells were acquired on a BD FACS Canto II™ system using BD FACSDiva™ version 6.1.3 software (BD Biosciences, Franklin Lakes, NJ, USA). All samples were analyzed in duplicate. Data were analyzed using InfinicytTM (Cytognos, Salamanca, Spain). The EPCs were defined as CD45low/CD34+/CD309+/CD133+/CD144− cells [36] and the CECs as CD45low/CD34+/CD309+/CD133−/CD144+ cells [7,23,37]. EPC and CEC counts were expressed as % leukocytes (CD45+ cells). The intra-assay variation was <$5\%$. ## 2.4. Statistical Analysis Exploratory analysis and Shapiro-Wilk tests were performed to determine the normality of the data distribution. Variables are expressed as mean ± standard deviation, mean differences with their 2-sided $95\%$ CIs or absolute number and percentage. Paired-sample T-tests were performed for the within-group comparisons (EPCs and CECs) from baseline to 30 min after the CPET. Between-group comparisons were performed using the Student’s independent t-test (EPCs, CECs, and clinical characteristics), Chi-squared (χ2) test, or Fisher’s exact test (clinical characteristics). Also, the HF patients were divided into two groups according to the severity of exercise intolerance (based on the median VO2 peak: 17.0 mL/kg/min); then, a student’s independent t-test was performed to assess whether those with lower cardiorespiratory fitness show a similar EPC response to those with higher fitness. The value of significance was set at a 1-sided p value ≤ 0.05. Data analyses were made using IBM SPSS Statistics 26 (IBM Corp., Armonk, NY, USA). ## 3.1. Participants’ Characteristics HF patients and age-matched adults free from cardiovascular disease were well balanced for age (67.8 ± 9.7 vs. 65.7 ± 7.1 years old, $$p \leq 0.52$$) and body mass index (27.7 ± 3.6 vs. 29.6 ± 5.6 kg/m2, $$p \leq 0.30$$) (Table 1). Some cardiovascular risk factors were more prevalent in HF patients, including diabetes mellitus ($53.8\%$) and hypertension ($84.6\%$). The HF patients showed a reduced left ventricular ejection fraction (LVEF) (37.2 ± $12.0\%$) and VO2 peak (16.8 ± 4.1 mL/kg/min; 76.2 ± $22.9\%$ of predicted). The VE/VCO2 slope was 34.6 ± 10.9 and the duration of the CPET was 9.0 ± 1.9 min. All patients completed the maximal exercise bout without signs or symptoms of ischemia. At baseline, HF patients presented a significantly lower number of circulating EPCs ($$p \leq 0.03$$) but similar levels of CECs ($$p \leq 0.08$$) when compared to the age-matched group (Table 1). ## 3.2. Effects of Maximal Exercise in the Levels of EPCs and CECs In HF patients, a single maximal exercise bout increased the levels of EPCs by 0.5 [$95\%$ Confidence Interval (CI), 0.07 to $0.93\%$], i.e., from 4.2 × 10−3 ± 1.5 × 10−$3\%$ to 4.7 × 10−3 ± 1.8 × 10−$3\%$ ($$p \leq 0.02$$) (Figure 1a). The circulating levels of CECs did not change in response to the exercise bout (3.8 × 10−3 ± 1.6 × 10−$3\%$ to 3.8 × 10−3 ± 1.3 × 10−$3\%$, $$p \leq 0.98$$) (Figure 1b). The EPC response to the exercise bout was similar among HF patients with lower (<17.0 mL/kg/min) versus higher VO2 peak (≥17.0 mL/kg/min) (0.41 ± 0.77 vs. 0.69 ± $0.71\%$, $$p \leq 0.53$$). When comparing the levels of EPCs of the HF patients after the maximal exercise bout with the resting value of the age-matched group, we observed that the circulating level of EPCs in HF patients was not significantly different from the level of the age-matched participants (4.7 × 10−3 ± 1.8 × 10−$3\%$ vs. 5.4 × 10−3 ± 1.7 × 10−$3\%$, respectively, $$p \leq 0.14$$) (Figure 1a). ## 4. Discussion The main findings of this study confirm our hypothesis that a maximal exercise bout increases the circulating levels of EPCs in HF patients, and that the number of EPCs in circulation after the exercise bout is similar to the baseline count of age-matched adults free from cardiovascular disease. Moreover, acute exercise improves EPC levels without increasing endothelial damage, assessed by measuring the levels of CECs. The effects of acute exercise on EPCs are still uncertain, with previous studies reporting increased levels of EPCs after a maximal exercise test [27,28,29] while others describing no differences [30,31,32,38,39]. For instance, Kourek, et al. [ 27] observed an increase in the levels of EPCs after a maximal exercise bout performed on a cycle ergometer. The same authors tested the influence of HF severity on the ability to mobilize EPCs after a maximal exercise and concluded that the ability to mobilize EPCs after a maximal exercise was not affected by HF severity, regardless of the criteria (VO2 peak, predicted VO2 peak or VE/VCO2) used to determine HF severity [28]. Another study found a nearly four-fold increase in the number of EPCs [40]. Our preliminary results tend to indicate that the EPC response to a maximal exercise bout is similar irrespective of patients’ VO2 peak. On the other hand, Van Craenenbroeck, et al. [ 31] did not find significant changes in EPCs at several time points after a maximal exercise test (i.e., 10 min, 30 min, 1 h, 2 h, 4 h, 8 h, 12 h, 24 h, 48 h) in a group of patients with chronic HF. Some authors suggested that the blunted response of EPCs to exercise may be related to inflammatory factors and diminished availability of nitric oxide (NO) in response to exercise [41,42]. It must be highlighted that the literature is not consistent in the antibodies used to define or select cells that express properties attributed to EPCs, for instance, EPCs have been identified as CD34+/CD133+/VEGFR2, CD34+/CD45−/CD133+/VEGFR2, and CD34+/CD45−/CD133+ [27,28,29] or CD34+/KDR+/CD3− [30,31,32] or CD34+/KDR+ [38], CD34+/KDR+/CD45dim [39], and CD133+/CD144+ (or AC133+/VE-Cadherin+) cells [40]. Additionally, the different units used to count circulating EPCs (e.g., cells per event, percentage of cells within the lymphocyte or the total mononuclear cell population, or cells per volume) make comparisons between studies difficult. The exercise characteristics (e.g., intensity and duration) must be considered when evaluating the physiological response of endothelial repair mechanisms to exercise. The intensity and type of exercise may have a bigger influence on EPC dynamics than the duration. Laufs and colleagues tested the acute effects of different durations of exercise and found that a 10-min aerobic exercise session at moderate intensity does not change the levels of EPCs, while a 30-min exercise session at moderate to vigorous intensity increases EPCs in healthy young men [43]. Moreover, a recent study tested chronic HF patients (EF ≤ $45\%$) and found that both 31 min of high-intensity interval exercise—4 min of aerobic exercise at $80\%$ VO2 peak intermitted by 3 min of active recovery at moderate intensity ($50\%$ VO2 peak)—and 53-min of continuous exercise at moderate intensity ($50\%$ VO2 peak) (both exercise sessions were similar in total work) increased the levels of EPCs immediately and 40 min after the end of the exercise sessions [44]. In this sense, considering that in the present study, the exercise bout lasted an average of 9 min, and CPET is characterized by an effort of short duration and high intensity [45], we may assume that the intensity of exercise may be more important to the mobilization of EPCs than the duration of exercise. Our results showed lower levels of EPCs in patients with HF at rest, which is in line with previous research [11,16,46]. Indeed, HF patients may present a down-regulation of endothelial nitric oxide synthase and increased oxidative stress that may affect the mobilization of EPCs [4,14]. These mechanisms participate in the progression of endothelial dysfunction and contribute to a lower capacity of endothelial repair and regeneration in HF [4,14]. The present study did not assess the mechanisms by which exercise would increase the number of EPCs in HF patients. However, a schematic diagram summarizing the potential pathways involved in EPCs dynamics is provided in Figure 2. Regarding the levels of CECs, we did not find significant differences in response to maximal exercise in HF patients. Previous studies showed that a maximal exercise session [51] or 30 min of high-intensity interval exercise [52] did not change the CEC count in healthy young men. However, a previous study reported an increase in CEC levels immediately after a CPET, which was followed by a significant fall 30 min after exercise among coronary artery disease patients [19]. This biphasic pattern was also observed in 13 patients with effort angina (i.e., patients diagnosed during the exercise, which triggered angina pectoris or ST segment depression during the CPET) since the CEC count increased after exercise testing, followed by a decrease 4 h after the exercise test [53]. An acute exercise session [54] can lead to a biphasic acute response in endothelial function, i.e., a decrease in endothelial function followed by a significant improvement until reaching or overcoming the resting levels of endothelial function [54,55]. As recently discussed [54], acute exercise at high intensities can lead to endothelial damage, which would be reflected in decreased endothelial function and increased circulating indicators of endothelial damage (e.g., CECs). Thus, the activation of vascular repair mechanisms (e.g., mobilization of EPCs) would increase the repair capacity of the endothelium. Therefore, the endothelial damage could be detectable only for a short period, i.e., until vascular repair mechanisms act to repair the endothelial damage [54]. Methodological differences, such as the lack of standardization of surface markers for the quantification of CECs, limit the comparisons between studies and may, at least partially, explain divergent results. Our study presented some limitations. Firstly, we only collected blood at one time point after the CPET. Secondly, the present study lacks additional analysis of other endothelial damage markers (e.g., endothelial-derived microparticles [7] and inflammatory cytokines [41], which could provide further information regarding the effects of maximal exercise on endothelial damage. Finally, age-matched subjects did not perform a CPET, which limits the comparison of the exercise-induced EPC response between them and HF patients. Despite these limitations, these preliminary results may provide new insights and perspectives into the effects of an acute exercise session on vascular health among HF patients. Future research with a larger sample and evaluating additional indicators of endothelial repair and vascular angiogenesis is necessary to confirm our findings. ## 5. Conclusions A maximal exercise bout seems to increase the circulating levels of EPCs in HF patients without increasing endothelial damage (CEC count) in HF patients. ## References 1. McDonagh T.A., Metra M., Adamo M., Gardner R.S., Baumbach A., Böhm M., Burri H., Butler J., Čelutkienė J., Chioncel O.. 2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: Developed by the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC) with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur. Heart J. (2021) 42 3599-3726. DOI: 10.1093/eurheartj/ehab368 2. Godo S., Shimokawa H.. Endothelial Functions. Arterioscler. Thromb. Vasc. Biol. (2017) 37 e108-e114. DOI: 10.1161/ATVBAHA.117.309813 3. Ribeiro F., Alves A.J., Teixeira M., Ribeiro V., Duarte J.A., Oliveira J.. Endothelial function and atherosclerosis: Circulatory markers with clinical usefulness. Rev. Port. Cardiol. (2009) 28 1121-1151. PMID: 20058778 4. Giannitsi S., Bougiakli M., Bechlioulis A., Naka K.. Endothelial dysfunction and heart failure: A review of the existing bibliography with emphasis on flow mediated dilation. JRSM Cardiovasc. Dis. (2019) 8 2048004019843047. DOI: 10.1177/2048004019843047 5. Akiyama E., Sugiyama S., Matsuzawa Y., Konishi M., Suzuki H., Nozaki T., Ohba K., Matsubara J., Maeda H., Horibata Y.. Incremental prognostic significance of peripheral endothelial dysfunction in patients with heart failure with normal left ventricular ejection fraction. J. Am. Coll. Cardiol. (2012) 60 1778-1786. DOI: 10.1016/j.jacc.2012.07.036 6. Ambrosino P., Papa A., Buonauro A., Mosella M., Calcaterra I., Spedicato G.A., Maniscalco M., Di Minno M.N.D.. Clinical assessment of endothelial function in heart failure with preserved ejection fraction: A meta-analysis with meta-regressions. Eur. J. Clin. Investig. (2021) 51 e13552. DOI: 10.1111/eci.13552 7. Sabatier F., Camoin-Jau L., Anfosso F., Sampol J., Dignat-George F.. Circulating endothelial cells, microparticles and progenitors: Key players towards the definition of vascular competence. J. Cell. Mol. Med. (2009) 13 454-471. DOI: 10.1111/j.1582-4934.2008.00639.x 8. Hristov M., Erl W., Weber P.C.. Endothelial progenitor cells: Mobilization, differentiation, and homing. Arterioscler. Thromb. Vasc. Biol. (2003) 23 1185-1189. DOI: 10.1161/01.ATV.0000073832.49290.B5 9. Cheng C.C., Chang S.J., Chueh Y.N., Huang T.S., Huang P.H., Cheng S.M., Tsai T.N., Chen J.W., Wang H.W.. Distinct angiogenesis roles and surface markers of early and late endothelial progenitor cells revealed by functional group analyses. BMC Genom. (2013) 14. DOI: 10.1186/1471-2164-14-182 10. Hill J.M., Zalos G., Halcox J.P., Schenke W.H., Waclawiw M.A., Quyyumi A.A., Finkel T.. Circulating endothelial progenitor cells, vascular function, and cardiovascular risk. N. Engl. J. Med. (2003) 348 593-600. DOI: 10.1056/NEJMoa022287 11. Lopes J., Teixeira M., Cavalcante S., Gouveia M., Duarte A., Ferreira M., Simões M.I., Conceição M., Ribeiro I.P., Gonçalves A.C.. Reduced Levels of Circulating Endothelial Cells and Endothelial Progenitor Cells in Patients with Heart Failure with Reduced Ejection Fraction. Arch. Med. Res. (2022) 53 289-295. DOI: 10.1016/j.arcmed.2022.02.001 12. Kakzanov Y., Sevilya Z., Veturi M., Goldman A., Lev E.I.. Circulating Endothelial Progenitor Cells in Patients with Heart Failure with Preserved versus Reduced Ejection Fraction. Isr. Med. Assoc. J. (2021) 23 364-368. PMID: 34155850 13. Ross M.D.. Endothelial Regenerative Capacity and Aging: Influence of Diet, Exercise and Obesity. Curr. Cardiol. Rev. (2018) 14 233-244. DOI: 10.2174/1573403X14666180726112303 14. Djohan A.H., Sia C.H., Lee P.S., Poh K.K.. Endothelial Progenitor Cells in Heart Failure: An Authentic Expectation for Potential Future Use and a Lack of Universal Definition. J. Cardiovasc. Transl. Res. (2018) 11 393-402. DOI: 10.1007/s12265-018-9810-4 15. Koller L., Hohensinner P., Sulzgruber P., Blum S., Maurer G., Wojta J., Hülsmann M., Niessner A.. Prognostic relevance of circulating endothelial progenitor cells in patients with chronic heart failure. Thromb. Haemost. (2016) 116 309-316. DOI: 10.1160/TH16-01-0051 16. Samman Tahhan A., Hammadah M., Sandesara P.B., Hayek S.S., Kalogeropoulos A.P., Alkhoder A., Mohamed Kelli H., Topel M., Ghasemzadeh N., Chivukula K.. Progenitor Cells and Clinical Outcomes in Patients with Heart Failure. Circ. Heart Fail. (2017) 10 e004106. DOI: 10.1161/CIRCHEARTFAILURE.117.004106 17. Dignat-George F., Sampol J.. Circulating endothelial cells in vascular disorders: New insights into an old concept. Eur. J. Haematol. (2000) 65 215-220. DOI: 10.1034/j.1600-0609.2000.065004215.x 18. Woywodt A., Blann A.D., Kirsch T., Erdbruegger U., Banzet N., Haubitz M., Dignat-George F.. Isolation and enumeration of circulating endothelial cells by immunomagnetic isolation: Proposal of a definition and a consensus protocol. J. Thromb. Haemost. (2006) 4 671-677. DOI: 10.1111/j.1538-7836.2006.01794.x 19. Boos C.J., Balakrishnan B., Lip G.Y.. The effects of exercise stress testing on soluble E-selectin, von Willebrand factor, and circulating endothelial cells as indices of endothelial damage/dysfunction. Ann. Med. (2008) 40 66-73. DOI: 10.1080/07853890701652833 20. Sapp R.M., Hagberg J.M.. CrossTalk opposing view: Acute exercise does not elicit damage to the endothelial layer of systemic blood vessels in healthy individuals. J. Physiol. (2018) 596 541-544. DOI: 10.1113/JP274751 21. Martínez-Sales V., Sánchez-Lázaro I., Vila V., Almenar L., Contreras T., Reganon E.. Circulating endothelial cells in patients with heart failure and left ventricular dysfunction. Dis. Markers (2011) 31 75-82. DOI: 10.1155/2011/757840 22. Erdbruegger U., Haubitz M., Woywodt A.. Circulating endothelial cells: A novel marker of endothelial damage. Clin. Chim. Acta (2006) 373 17-26. DOI: 10.1016/j.cca.2006.05.016 23. Farinacci M., Krahn T., Dinh W., Volk H.-D., Düngen H.-D., Wagner J., Konen T., von Ahsen O.. Circulating endothelial cells as biomarker for cardiovascular diseases. Res. Pract. Thromb. Haemost. (2018) 3 49-58. DOI: 10.1002/rth2.12158 24. Cavalcante S.L., Lopes S., Bohn L., Cavero-Redondo I., Álvarez-Bueno C., Viamonte S., Santos M., Oliveira J., Ribeiro F.. Effects of exercise on endothelial progenitor cells in patients with cardiovascular disease: A systematic review and meta-analysis of randomized controlled trials. Rev. Port. Cardiol. (2019) 38 817-827. DOI: 10.1016/j.repc.2019.02.016 25. Ribeiro F., Ribeiro I.P., Alves A.J., do Céu Monteiro M., Oliveira N.L., Oliveira J., Amado F., Remião F., Duarte J.A.. Effects of exercise training on endothelial progenitor cells in cardiovascular disease: A systematic review. Am. J. Phys. Med. Rehabil. (2013) 92 1020-1030. DOI: 10.1097/PHM.0b013e31829b4c4f 26. Recchioni R., Marcheselli F., Antonicelli R., Lazzarini R., Mensà E., Testa R., Procopio A.D., Olivieri F.. Physical activity and progenitor cell-mediated endothelial repair in chronic heart failure: Is there a role for epigenetics?. Mech. Ageing Dev. (2016) 159 71-80. DOI: 10.1016/j.mad.2016.03.008 27. Kourek C., Alshamari M., Mitsiou G., Psarra K., Delis D., Linardatou V., Pittaras T., Ntalianis A., Papadopoulos C., Panagopoulou N.. The acute and long-term effects of a cardiac rehabilitation program on endothelial progenitor cells in chronic heart failure patients: Comparing two different exercise training protocols. Int. J. Cardiol. Heart Vasc. (2021) 32 100702. DOI: 10.1016/j.ijcha.2020.100702 28. Kourek C., Karatzanos E., Psarra K., Georgiopoulos G., Delis D., Linardatou V., Gavrielatos G., Papadopoulos C., Nanas S., Dimopoulos S.. Endothelial progenitor cells mobilization after maximal exercise according to heart failure severity. World J. Cardiol. (2020) 12 526-539. DOI: 10.4330/wjc.v12.i11.526 29. Kourek C., Karatzanos E., Psarra K., Ntalianis A., Mitsiou G., Delis D., Linardatou V., Pittaras T., Vasileiadis I., Dimopoulos S.. Endothelial progenitor cells mobilization after maximal exercise in patients with chronic heart failure. Hell. J. Cardiol. (2021) 62 70-72. DOI: 10.1016/j.hjc.2020.03.007 30. Van Craenenbroeck E.M., Beckers P.J., Possemiers N.M., Wuyts K., Frederix G., Hoymans V.Y., Wuyts F., Paelinck B.P., Vrints C.J., Conraads V.M.. Exercise acutely reverses dysfunction of circulating angiogenic cells in chronic heart failure. Eur. Heart J. (2010) 31 1924-1934. DOI: 10.1093/eurheartj/ehq058 31. Van Craenenbroeck E.M., Bruyndonckx L., Van Berckelaer C., Hoymans V.Y., Vrints C.J., Conraads V.M.. The effect of acute exercise on endothelial progenitor cells is attenuated in chronic heart failure. Eur. J. Appl. Physiol. (2011) 111 2375-2379. DOI: 10.1007/s00421-011-1843-1 32. Van Craenenbroeck E.M., Hoymans V.Y., Beckers P.J., Possemiers N.M., Wuyts K., Paelinck B.P., Vrints C.J., Conraads V.M.. Exercise training improves function of circulating angiogenic cells in patients with chronic heart failure. Basic Res. Cardiol. (2010) 105 665-676. DOI: 10.1007/s00395-010-0105-4 33. García-Conesa M.-T., Philippou E., Pafilas C., Massaro M., Quarta S., Andrade V., Jorge R., Chervenkov M., Ivanova T., Dimitrova D.. Exploring the Validity of the 14-Item Mediterranean Diet Adherence Screener (MEDAS): A Cross-National Study in Seven European Countries around the Mediterranean Region. Nutrients (2020) 12. DOI: 10.3390/nu12102960 34. Hoymans V.Y., Van Craenenbroeck A.H., Bruyndonckx L., van Ierssel S.H., Vrints C.J., Conraads V.M., Van Craenenbroeck E.M.. TransFix. Microvasc. Res. (2012) 84 384-386. DOI: 10.1016/j.mvr.2012.08.007 35. Ahmed F.W., Rider R., Glanville M., Narayanan K., Razvi S., Weaver J.U.. Metformin improves circulating endothelial cells and endothelial progenitor cells in type 1 diabetes: MERIT study. Cardiovasc. Diabetol. (2016) 15 116. DOI: 10.1186/s12933-016-0413-6 36. Schmidt-Lucke C., Fichtlscherer S., Aicher A., Tschöpe C., Schultheiss H.P., Zeiher A.M., Dimmeler S.. Quantification of circulating endothelial progenitor cells using the modified ISHAGE protocol. PLoS ONE (2010) 5. DOI: 10.1371/journal.pone.0013790 37. Kraan J., Strijbos M.H., Sieuwerts A.M., Foekens J.A., den Bakker M.A., Verhoef C., Sleijfer S., Gratama J.W.. A new approach for rapid and reliable enumeration of circulating endothelial cells in patients. J. Thromb. Haemost. (2012) 10 931-939. DOI: 10.1111/j.1538-7836.2012.04681.x 38. Van Craenenbroeck E.M., Denollet J., Paelinck B.P., Beckers P., Possemiers N., Hoymans V.Y., Vrints C.J., Conraads V.M.. Circulating CD34+/KDR+ endothelial progenitor cells are reduced in chronic heart failure patients as a function of Type D personality. Clin. Sci. (2009) 117 165-172. DOI: 10.1042/CS20080564 39. Gevaert A.B., Beckers P.J., Van Craenenbroeck A.H., Lemmens K., Van De Heyning C.M., Heidbuchel H., Vrints C.J., Van Craenenbroeck E.M.. Endothelial dysfunction and cellular repair in heart failure with preserved ejection fraction: Response to a single maximal exercise bout. Eur. J. Heart Fail. (2019) 21 125-127. DOI: 10.1002/ejhf.1339 40. Rehman J., Li J.L., Parvathaneni L., Karlsson G., Panchal V.R., Temm C.J., Mahenthiran J., March K.L.. Exercise acutely increases circulating endothelial progenitor cells and monocyte-/macrophage-derived angiogenic cells. J. Am. Coll. Cardiol. (2004) 43 2314-2318. DOI: 10.1016/j.jacc.2004.02.049 41. West D.J., Campbell M.D., Gonzalez J.T., Walker M., Stevenson E.J., Ahmed F.W., Wijaya S., Shaw J.A., Weaver J.U.. The inflammation, vascular repair and injury responses to exercise in fit males with and without Type 1 diabetes: An observational study. Cardiovasc. Diabetol. (2015) 14 71. DOI: 10.1186/s12933-015-0235-y 42. Volaklis K.A., Tokmakidis S.P., Halle M.. Acute and chronic effects of exercise on circulating endothelial progenitor cells in healthy and diseased patients. Clin. Res. Cardiol. (2013) 102 249-257. DOI: 10.1007/s00392-012-0517-2 43. Laufs U., Urhausen A., Werner N., Scharhag J., Heitz A., Kissner G., Böhm M., Kindermann W., Nickenig G.. Running exercise of different duration and intensity: Effect on endothelial progenitor cells in healthy subjects. Eur. J. Cardiovasc. Prev. Rehabil. (2005) 12 407-414. DOI: 10.1097/01.hjr.0000174823.87269.2e 44. Mitsiou G., Karatzanos E., Smilios I., Psarra K., Patsaki I., Douda H.T., Ntalianis A., Nanas S., Tokmakidis S.P.. Exercise promotes endothelial progenitor cell mobilization in patients with chronic heart failure. Eur. J. Prev. Cardiol. (2022) 28 e24-e27. DOI: 10.1093/eurjpc/zwaa046 45. Ferentinos P., Tsakirides C., Swainson M., Davison A., Martyn-St James M., Ispoglou T.. The impact of different forms of exercise on circulating endothelial progenitor cells in cardiovascular and metabolic disease. Eur. J. Appl. Physiol. (2022) 122 815-860. DOI: 10.1007/s00421-021-04876-1 46. Chiang C.H., Huang P.H., Leu H.B., Hsu C.Y., Wang K.F., Chen J.W., Lin S.J.. Decreased circulating endothelial progenitor cell levels in patients with heart failure with preserved ejection fraction. Cardiology (2013) 126 191-201. DOI: 10.1159/000351973 47. Aicher A., Heeschen C., Mildner-Rihm C., Urbich C., Ihling C., Technau-Ihling K., Zeiher A.M., Dimmeler S.. Essential role of endothelial nitric oxide synthase for mobilization of stem and progenitor cells. Nat. Med. (2003) 9 1370-1376. DOI: 10.1038/nm948 48. Yang Z., Wang J.M., Chen L., Luo C.F., Tang A.L., Tao J.. Acute exercise-induced nitric oxide production contributes to upregulation of circulating endothelial progenitor cells in healthy subjects. J. Hum. Hypertens (2007) 21 452-460. DOI: 10.1038/sj.jhh.1002171 49. Heinisch P.P., Bello C., Emmert M.Y., Carrel T., Dreßen M., Hörer J., Winkler B., Luedi M.M.. Endothelial Progenitor Cells as Biomarkers of Cardiovascular Pathologies: A Narrative Review. Cells (2022) 11. DOI: 10.3390/cells11101678 50. Chang E., Paterno J., Duscher D., Maan Z.N., Chen J.S., Januszyk M., Rodrigues M., Rennert R.C., Bishop S., Whitmore A.J.. Exercise induces stromal cell-derived factor-1α-mediated release of endothelial progenitor cells with increased vasculogenic function. Plast. Reconstr. Surg. (2015) 135 340e-350e. DOI: 10.1097/PRS.0000000000000917 51. O’Sullivan S.E.. The effects of exercise training on markers of endothelial function in young healthy men. Int. J. Sports Med. (2003) 24 404-409. DOI: 10.1055/s-2003-41183 52. Sapp R.M., Evans W.S., Eagan L.E., Chesney C.A., Zietowski E.M., Prior S.J., Ranadive S.M., Hagberg J.M.. The effects of moderate and high-intensity exercise on circulating markers of endothelial integrity and activation in young, healthy men. J. Appl. Physiol. (1985) (2019) 127 1245-1256. DOI: 10.1152/japplphysiol.00477.2019 53. Mutin M., Canavy I., Blann A., Bory M., Sampol J., Dignat-George F.. Direct evidence of endothelial injury in acute myocardial infarction and unstable angina by demonstration of circulating endothelial cells. Blood (1999) 93 2951-2958. DOI: 10.1182/blood.V93.9.2951 54. Adams V.. CrossTalk proposal: Acute exercise elicits damage to the endothelial layer of systemic blood vessels in healthy individuals. J. Physiol. (2018) 596 537-539. DOI: 10.1113/JP274750 55. Dawson E.A., Green D.J., Cable N.T., Thijssen D.H.. Effects of acute exercise on flow-mediated dilatation in healthy humans. J. Appl. Physiol. (1985) (2013) 115 1589-1598. DOI: 10.1152/japplphysiol.00450.2013
--- title: Mechanistic Insights into the Role of OPN in Mediating Brain Damage via Triggering Lysosomal Damage in Microglia/Macrophage authors: - Chengcheng Gai - Yijing Zhao - Danqing Xin - Tingting Li - Yahong Cheng - Zige Jiang - Yan Song - Dexiang Liu - Zhen Wang journal: Cells year: 2023 pmcid: PMC10046941 doi: 10.3390/cells12060854 license: CC BY 4.0 --- # Mechanistic Insights into the Role of OPN in Mediating Brain Damage via Triggering Lysosomal Damage in Microglia/Macrophage ## Abstract We previously found that osteopontin (OPN) played a role in hypoxia–ischemia (HI) brain damage. However, its underlying mechanism is still unknown. Bioinformatics analysis revealed that the OPN protein was linked to the lysosomal cathepsin B (CTSB) and galectin-3 (GAL-3) proteins after HI exposure. In the present study, we tested the hypothesis that OPN was able to play a critical role in the lysosomal damage of microglia/macrophages following HI insult in neonatal mice. The results showed that OPN expression was enhanced, especially in microglia/macrophages, and colocalized with lysosomal-associated membrane protein 1 (LAMP1) and GAL-3; this was accompanied by increased LAMP1 and GAL-3 expression, CTSB leakage, as well as impairment of autophagic flux in the early stage of the HI process. In addition, the knockdown of OPN expression markedly restored lysosomal function with significant improvements in the autophagic flux after HI insult. Interestingly, cleavage of OPN was observed in the ipsilateral cortex following HI. The wild-type OPN and C-terminal OPN (Leu152-Asn294), rather than N-terminal OPN (Met1-Gly151), interacted with GAL-3 to induce lysosomal damage. Furthermore, the secreted OPN stimulated lysosomal damage by binding to CD44 in microglia in vitro. Collectively, this study demonstrated that upregulated OPN in microglia/macrophages and its cleavage product was able to interact with GAL-3, and secreted OPN combined with CD44, leading to lysosomal damage and exacerbating autophagosome accumulation after HI exposure. ## 1. Introduction Osteopontin (OPN), also called secreted phosphoprotein 1, is an extracellular matrix protein involved in many physiological and pathophysiological processes [1]. The basal OPN level in adult brains is weak [2], while OPN expression is notably upregulated during the inflammation associated with Alzheimer’s disease and other neurodegenerative conditions [3,4]. Several studies have suggested the immuno-modulatory role of OPN in brain pathologies [5,6]. Our previous studies have shown that hypoxia–ischemia (HI) brain damage increased OPN expression in microglia/macrophages of neonatal mice, which contributed to neuroinflammation [7,8]. Autophagy is an evolutionarily conserved transport pathway crucial to maintaining cellular homeostasis through the sequestration, delivery, and degradation of unwanted proteins, macromolecular complexes, and organelles into lysosomes. Lysosomes are cytoplasmic membrane-enclosed organelles containing hydrolytic enzymes that participate in the control of the intracellular turnover of macromolecules [9]. A study has shown that lysosomal destabilization by oxidative stress and other apoptotic signals causes the leakage of cathepsins and other hydrolases into the cytosol, which is potentially harmful to cell survival [10]. For example, the leakage of cathepsin B (CTSB) from the lysosomes to the cytoplasm could trigger the activation of the NOD-like receptor thermal protein domain-associated protein 3 (NLRP3) inflammasome in microglia and additionally contribute to NLRP3-mediated pyroptosis [11,12]. Given the essential role of lysosomes in the maturation/degradation stage of autophagy, it is possible that progressive dysfunction in the lysosomal apparatus is deleterious to cells. Based on our previous studies, the aim of this study was to evaluate the effect of OPN on impairing the lysosomal function and autophagic flux following HI insult. We found that intracellular OPN interacted with galectin-3 (GAL-3) to induce lysosomal damage and further identified the binding region of OPN to GAL-3 as the C-terminal fragment. It was also found that the secreted OPN regulated lysosomal function by binding to its receptor CD44. The study provides novel mechanistic insights into the pathophysiology of lysosomal dysfunction after HI exposure. ## 2.1. Ethical Statement Ethics approval statements for animal work and procedures were provided by the Animal Ethical and Welfare Committee of Shandong University (approval No. ECSBMSSDU2020-2-067). ## 2.2. Animals and Treatment C57BL/6J mice (female, 3–5 g) provided by the Laboratory Animal Center of Shandong University (Jinan, China) were used in all experiments. A well-characterized model of neonatal HI was followed as previously described [13]. C57BL/6 mouse pups were anesthetized with $2\%$ isoflurane on postnatal day 7 (P7) and the right pulsating common carotid artery was carefully separated and double-ligated. The skin incision was sutured with a surgical suture. After surgery, the pups were returned to their dam for 30 min. The pups were then placed in a hypoxic chamber for 1 h. The chamber was maintained at 37 °C and contained $9.5\%$ O2 + $90.5\%$ N2. Mice in HI+3-methyladenine (3-MA, Cat# M9281, Sigma-Aldrich, St. Louis, MO, USA) and HI+chloroquine (CQ, Cat# C6628, Sigma-Aldrich, St. Louis, MO, USA) groups were administered 3-MA (10 mg/kg) or CQ (10 mg/kg) intraperitoneally at 1 h, 1 day, 2 days, and 3 days after HI insult [14,15,16], respectively. Finally, the mice were sacrificed to obtain the injured cortex for experimentation. ## 2.3. Sample Preparation for Tandem Mass Tag (TMT) Analysis This study randomly selected three right cerebral cortices from the Sham and HI samples, respectively, for TMT-based quantitative proteomics analysis. The samples were sent to Novogene Bioinformatics Technology Co. (Beijing, China) for TMT quantitative proteomics using a C18 column (Waters BEH C18 4.6 mm × 250 mm, 5 µm) on a Rigol L3000 HPLC. Statistical analysis for the proteomic data was performed using Proteome Discoverer software (2.2; Thermo Fisher Scientific, Waltham, MA, USA). ## 2.4. Protein–Protein Interaction (PPI) Network The PPI network was analyzed with the Cytoscape software (3.7, San Diego, CA, USA) by submitting sequences of differentially expressed proteins (DEPs), and an interaction with a combined score > 0.4 was considered as statistically significant. ## 2.5. Kyoto Encyclopedia of Genes and Genomes (KEGG) Analyses To identify the biological functions of the selected DEPs in the Sham and HI groups, the determined protein sequences were mapped using KEGG pathway enrichment analyses. ## 2.6. Lentiviral and Plasmid Transfection Lentiviral OPN, GAL-3 shRNA (si-OPN, si-GAL-3), lentiviral shRNA negative control (si-NC), and LC3-GFP plasmid were packaged in Genechem (Shanghai, China), and mouse CD44 small interfering RNA (si-CD44) and the negative control (si-NC) were obtained from Genechem (Shanghai, China). The GAL-3-mCherry, OPN-EGFP, wild-type OPN-Flag (OPN-WT-Flag), the N-terminal fragment of OPN (Met1-Gly151, OPN-N-Flag), and C-terminal fragment of OPN (Leu152-Asn294, OPN-C-Flag) plasmids were purchased from Miaoling Plasmidbio (Wuhan, China). Transfection of the lentivirus, plasmid, and RNA interference into cells was carried out using Lipofectamine 2000 transfection reagent (Cat# 11668019, Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. ## 2.7. Oxygen–Glucose Deprivation (OGD) BV-2 cells were obtained from Cellcook (Guangzhou Cellcook Biotech Co., Ltd., Guangzhou, China). BV-2 was used to imitate primary microglia for the observation of pathophysiological processes in microglia [17]. BV-2 cells seeded in 6-well plates were incubated overnight and treated with OGD. The culture medium was replaced with glucose-free Dulbecco’s modified eagle medium, and the culture dish was placed in a hypoxia incubator with a $1\%$ O2 and $5\%$ CO2 mixture for 3 h. After OGD, the cells were incubated for 24 h in normal oxygen. The cells were then collected at the time points indicated for use in future experiments. ## 2.8. Lyso-Tracker Red and Mito-Tracker Red CMXRos Staining Lyso-Tracker Red (Cat# C1046, Beyotime, Shanghai, China) staining was used to assess lysosomal activity. Lyso-Tracker Red was dissolved in phosphate-buffered saline (PBS) and stored at 4 °C. Prior to imaging, Lyso-Tracker Red (50 mol/L) was preheated at 37 °C and incubated with BV-2 cells or HEK293T cells for 30 min at 37 °C. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) at room temperature for 10 min. Images were taken on a fluorescence microscope (Axio Vert. A1, Carl Zeiss, Oberkochen, Baden-Württemberg, Germany). Mito-Tracker Red CMXRos purchased from Beyotime (Cat# C1035). Mito-Tracker Red CMXRos (50 mol/L) was incubated with HEK293T cells for 30 min at 37 °C. The cell nuclei were stained with DAPI at room temperature for 10 min. Images were taken on a Carl Zeiss (Axio Vert. A1) fluorescence microscope. Data analysis was performed using ImageJ software. ## 2.9. Analysis of Lysosome Membrane Permeability Acridine orange (AO) (Cat# YS175111, Solarbio, Beijing, China) staining was used to analyze the lysosomal membrane permeability as previously described [18]. AO is a lysosomotropic metachromatic fluorochrome and emits red fluorescence in intact lysosomes and green fluorescence in the cytoplasm [10]. BV-2 cells were cultured in 24-well chamber slides and treated without or with OGD. The cells were then incubated with 1 μg/mL AO for 30 min at 37 °C in the dark. The cells were visualized with a Carl Zeiss (Axio Vert. A1) fluorescence microscope. Data analysis was performed using ImageJ software. ## 2.10. Infarct Volume Measurement At 3 days post-HI, the brains were obtained and sectioned into 4 slices, and then immersed in $2\%$ 2, 3, 5-triphenyltetrazolium chloride monohydrate (TTC, Cat# T8877, Sigma-Aldrich, St. Louis, MO, USA) solution at 37 °C for 20 min. The infarct volume was traced and analyzed using ImageJ software (1.53c, NIH, Bethesda, MD, USA). Infract volume quantification was then conducted following procedures described previously [19]. ## 2.11. Lateral Cerebral Ventricle Injections P4 mouse pups were fixed in a prone position under isoflurane anesthesia. The injection site was positioned at the midpoint of the lambda and bregma sutures, 0.8–1 mm laterally from the sagittal suture [20]. Then, 3 × 107 TU/3 μL of si-OPN, si-GAL-3, and si-NC; 3 μL si-CD44 or si-NC at a concentration of 20 nmol/L were injected into the lateral ventricle (3 mm deep from the skull). The injection rate was 0.5 μL/min and the needle remained at the site for about 5 min after injection. The injection needle was then raised carefully at a speed of 1 mm per minute. ## 2.12. Immunofluorescence Staining Four-micrometer paraffin sections were baked, deparaffinized, and rehydrated in xylene and graded alcohols; antigen retrieval was then performed using citrate buffer (Cat# C1032, Solarbio, Beijing, China) for 20 min at 100 °C within a pressure chamber. Slides were blocked using normal $10\%$ goat serum or donkey serum for 1 h at room temperature, and subsequently incubated with indicated primary antibodies (Dilution: 1:200) at 4 °C overnight. Sections were then incubated for 1 h at 37 °C with the fluorescein-conjugated secondary antibodies, and the sections were then counterstained with DAPI. The negative control consisted of sections processed in the same way as the tests with the omission of the primary antibody incubation step. Images were examined through Carl Zeiss (Axio Vert. A1) fluorescence microscopy. All experiments were performed three times. There were four mice in each experimental group, and two sections per mouse were randomly selected for imaging. Three fields of the image were randomly selected for each section for counting. Counting was performed in a blinded manner. Data analysis was performed using ImageJ software. All antibodies used in the study are shown in Table 1. ## 2.13. Terminal Deoxynucleotidyltransferase-Mediated dUTP-Biotin Nick End Labeling (TUNEL) Assay The FITC-labeled TUNEL apoptosis assay (Cat# G1501, Servicebio, Wuhan, China) was performed to detect apoptosis in frozen slides of brain tissue following the manufacturer’s protocol. The frozen slides were restored to room temperature and dried in air, and then incubated for 20 min at 37 °C with a proteinase K working solution. The TUNEL reaction mixture (recombinant TdT enzyme: FITC-12-dUTP; labeling mix: equilibration buffer = 1 µL:5 µL:50 µL) was added to the samples. The samples were capped and incubated for 60 min at 37 °C in a humidified atmosphere in the dark. The slides were rinsed 3 times with PBS and counterstained with DAPI staining for 10 min. For the negative controls, the TdT enzyme was not added to the samples. Immunofluorescence images were obtained using a fluorescence microscope (Olympus BX53, Tokyo, Japan). There were four mice in each experimental group, and two sections per mouse were randomly selected for imaging. Three fields of the image were randomly selected for each section for counting. Counting was performed in a blinded manner. Data analysis was performed using ImageJ software. ## 2.14. Co-Immunoprecipitation (IP) Assay For co-IP in cells, the whole-cell lysates were incubated with protein A + G agarose beads conjugated with antibodies against OPN with rocking at 4 °C overnight. Immunoglobulin G (IgG) from the same animal species was used as the negative control. The beads were eluted five times with IP buffer and neutralized with 40 μL of 5 × SDS-PAGE loading buffer to heat at 100 °C for 10 min. IP and input samples were subjected to Western blot analyses. For the in vitro co-IP assay, OPN and GAL-3 proteins were generated in vitro through translation using TnT® T7 Coupled Reticulocyte Lysate Systems (Cat# L4600, Madison, Promega, WI, USA) according to the manufacturer’s protocol. The same analysis as that described above was performed after the protein products were combined. ## 2.15.1. Geotaxis Reflex Mice were placed with their heads facing downward on the center of a 30 cm grid inclined at an angle of 45 The time required for a full angle of 90 upward rotation within a maximum time of 30 s was recorded. Each mouse was tested 4 times and the average value from the 4 tests was calculated. ## 2.15.2. Grip Test The forelimbs of pups were placed on a 30 cm thin iron wire to assess their grip strength. The latent time to fall was recorded with a maximum time allowance of 60 s. Each mouse was tested 4 times and the average value from the 4 tests was calculated. ## 2.15.3. T-Maze T-maze alternation was used to determine cognitive impairment. For the first forced trial, the mouse was placed in the start arm; once it entered the goal arm, the goal arm was immediately blocked with a plastic guillotine door. The mouse was blocked in the chosen arm for 30 s. The mouse eventually returned to the start arm and the next trial was started. In the second run, with all of the arms unblocked, the mouse had to choose between the previously entered goal arm (no alternation) and a new arm (alternation) [21]. The test procedures were repeated 4 times per day for 3 consecutive days. Alternation (%)=The number of correct choicesTotal sets performed×$100\%$ Side preference (%)=The number of preferred side that the mouse has chosenTotal runs performed×$100\%$ ## 2.16. Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) The total RNA of tissue samples was extracted using an Ultrapure RNA Kit (Cat# CW0581M, CWBIO, Beijing, China) following the manufacturer’s protocol. The purity of the RNA was measured using a spectrophotometer at A260 nm and A280 nm, and samples with an A260/A280 ratio of ≥1.8 were used. A Revert Aid First Strand cDNA Synthesis Kit (Cat# FSQ-101, TOYOBO, Osaka, Japan) was used to synthesize first-strand cDNA with 1 μg of total RNA and primers. The qRT-PCR was completed with a SYBR Green PCR master mix (Cat# PC3301, Aidlab Biotechnologies, Beijing, China) using a Bio-Rad IQ5 Real-Time PCR System (Bio-Rad, Hercules, CA, USA). The reaction annealing temperature was 60 °C and amplification was performed for 40 cycles. The reverse transcription negative control (without reverse transcriptase) was included to confirm the absence of genomic DNA in the reactions in qRT-PCR analysis. The primer sequences used are described in Table 2. ## 2.17. Western Blot Analysis The ipsilateral cortexes from mice ($$n = 4$$ in each group) were lysed in RIPA buffer (Cat# P0013B, Beyotime) containing protease inhibitors and phosphatase inhibitor. A lysosome isolation kit (Cat# Lysiso1, Sigma-Aldrich, St. Louis, MO, USA) was used to obtain the lysosomes and the remaining cytoplasmic components in the brain tissue, and the cytoplasmic components were added to RIPA buffer for lysing to obtain cytoplasmic protein. The protein concentrations were determined using a BCA Protein Assay Kit (Cat# CW0014, CWBIO, Beijing, China). The protein sample lysates were electrophoretically separated on an SDS-PAGE ($10\%$ or $12\%$), then transferred from gels to 0.45 µm polyvinylidene fluoride membranes (Cat# IPVH00010, Millipore, Boston, MA, USA). The membranes with proteins were blocked for 1 h with $5\%$ non-fat milk in TBST at room temperature and then incubated overnight at 4 °C with the following primary antibodies (Dilution: 1:1000). An enhanced chemiluminescence (Cat# WBKLS0500, Millipore, Boston, MA, USA) system was used to monitor the blot signals. All experiments were performed three times. Protein quantitative analysis was conducted using ImageJ software. All antibodies used in the method are shown in Table 1. ## 2.18. Statistical Analysis The data from all experiments were analyzed using SPSS software. All of the experiments were carried out in a blinded manner. The data were presented as the mean ± SD of 4 or more biologically independent replicates. All experiments were performed three times. Normality between group samples was assessed using the Shapiro–Wilk test. The statistical significance of two independent groups was evaluated using Student’s t-test. One-way analysis of variance (ANOVA) followed by the Dunnett corrections or the Bonferroni corrections were used for comparisons of three or more groups. Statistical significance was defined as * $p \leq 0.05$; $p \leq 0.01$; * $p \leq 0.001.$ ## 3.1. Up-Regulated OPN Colocalized with LAMP1 and GAL-3 in Microglia/Macrophages after HI Insult We previously found that OPN protein expression was largely increased in the ipsilateral cortex and located in the microglia post-HI [8]. To explore the function of OPN, the Cytoscape software was used to predict OPN-interacting proteins and the KEGG pathway. The TMT-based quantitative proteomic analysis and PPI network of OPN revealed that a total of 23 DEPs exhibited a potential relationship with OPN, including cathepsins B, L, S, and Z, GAL-3 (GAL-3 is normally a cytosolic protein recruited to damaged lysosomes and triggers selective autophagy [22]), Galectin-1, and CD44 (Figure 1A). Cathepsins B, L, S, and Z are a class of lysosomal cysteine proteases. Moreover, Figure 1B reveals that significantly enriched pathways in DEPs were identified using KEGG pathway analysis in the comparison between the HI and Sham groups, and KEGG pathway analysis suggested that the pathway activated by HI injury was associated with lysosomes. For this reason, we hypothesized that OPN might be connected to lysosomal function. First, we speculated whether OPN expression was upregulated by HI insult in neonatal mice and affected injury. Consistent with our previous finding [7], OPN expression was undetectable in the naïve neonatal brain and in the hemisphere contralateral to the lesion, but was gradually upregulated in the ipsilateral cortex at 3 days after HI (Figure 1C). Immunofluorescence staining showed that increased OPN expression was mainly found in the microglia/macrophages (intracellular OPN) of the ipsilateral cortex at 3 days following HI insult. A few granular OPNs (secreted OPN) were scattered among the amoeboid microglia/macrophages in the lesion core (Figure S1A). Moreover, OPN puncta in the microglia/macrophages were colocalized with the lysosomal marker LAMP1 and GAL-3 (Figure 1D,E). To verify the findings in vivo, we also transfected HEK293T cells, which are easy to transfect and contain abundant lysosomes, with OPN-EGFP plasmid to overexpress OPN. The lysosome was stained with Lyso-Tracker Red and the mitochondria were stained with Mito-Tracker Red CMXRos. Similarly, the OPN-EGFP was colocalized with Lyso-Tracker Red (Figure 1F), not Mito-Tracker (Figure S1B). These data suggest that upregulated OPN may play an important role in lysosome function after HI insult. ## 3.2. HI Insult Led to Impairment of Lysosomal Function and Autophagic Flux Given that the lysosomal system plays an important role in homeostasis and neuronal integrity, we used different but complementary methods to evaluate lysosomal function. First, the effect of neonatal HI on the expression and distribution of some lysosomal proteins, such as LAMP1 (marker of lysosomes [23]) and GAL-3 was studied. The levels of LAMP1 mRNA and protein in the ipsilateral cortex were increased after HI exposure (Figure S2A and Figure 2A). The levels of GAL-3 mRNA and protein in the ipsilateral cortex were increased at 2 days and 3 days after HI exposure (Figure S2B and Figure 2A). Moreover, increased expression of GAL-3 was colocalized with LAMP1 (Figure S2C,D), similar to the previous report [24]. Lysosomes play an important role in autophagic degradation. Next, we examined the changes in the autophagic flux following HI insult. Consistent with our previous finding [25], the protein expression of microtubule-associated protein 1 light chain 3 (LC3) II in the ipsilateral cortex at 1 day, 2 days, 3 days, and 4 days post-HI increased significantly compared with the Sham group (Figure 2B). The protein expression of SQSTM1 (p62, a ubiquitin-binding protein delivered to lysosomes for degradation) gradually increased at 2 days, 3 days, and 4 days post-HI (Figure 2B), which suggested that the autophagic flux was significantly inhibited post-HI. We also found that 3-MA (inhibiting the formation of autophagosome precursors) effectively decreased the LC3-II and p62 levels at 3 days post-HI, indicating that the initiation of autophagy was affected by HI (Figure 2C). However, CQ (an inhibitor of autophagosome–lysosome fusion) could not elevate the LC3-II and p62 levels at 3 days post-HI (Figure 2C), indicating that its formation in the autolysosome was significantly blocked in HI brains and had remarkable ceiling effects. Previous findings have shown that microglial autophagy plays an important pro-inflammatory role in cytokine production and the neuroinflammatory response after ischemic stroke. Hypoxia/ischemia could lead to the excessive activation of autophagy in microglia, which exacerbates neuroinflammatory damage [26]. Considering that OPN is mainly found in the microglia and is involved in inflammation [7], in this study, we focused on the role of genetic OPN in lysosomal damage and the autophagy of microglia/macrophages after neonatal stroke. Evidence of lysosomal damage was detected by using Lyso-Tracker Red and AO staining. In control BV-2 cells, lysosomes were uniformly distributed, while they accumulated in a perinuclear location in OGD-exposed BV-2 cells (Figure 2D,E). AO yielded red fluorescence when accumulated within the lysosome and green fluorescence when released from ruptured lysosomes and diffused into the cytosol and nuclei [10]. OGD exposure weakened the red fluorescence in BV-2 cells (Figure 2F,G). Consistent with the in vitro results, these data suggested that HI insult led to lysosomal dysfunction and a subsequent increase in autophagosome accumulation and autophagic degradation blockage. ## 3.3. OPN Deficiency Reduced HI-Induced Lysosomes Damage and Autophagosome Accumulation, Associating with Improving Behavior Deficit To determine if OPN plays an important role in lysosome function, OPN was knocked down by lentiviral transfection in HI-injury mice (Figure S3A). Consistent with our previous finding [7], the genetic suppression of OPN significantly attenuated brain damage (Figure S3B–F). Upregulated LAMP1 and GAL-3 expressions in the ipsilateral cortex were reversed by the inhibition of OPN expression at 3 days following HI exposure (Figure 3A); meanwhile, the genetic suppression of OPN reduced HI-induced GAL-3 and LAMP1 mRNA expression (Figure 3B), demonstrating the attenuation in HI-induced damaged lysosome by OPN deficiency. In addition, blocking OPN expression decreased the LC3-II and p62 levels in the ipsilateral cortex at 3 days following HI exposure (Figure 3C). To assess whether OPN also follows a similar lysosomal function in vitro, the genetic suppression of OPN reversed OGD exposure-displaying a perinuclear location of Lyso-Tracker Red fluorescent signal (Figure 3D). The overexpression of OPN attenuated the red fluorescence of AO in BV-2 cells (Figure S3G). OGD exposure increased the LC3-GFP levels in the cytoplasm (Figure S3H,I), indicating that autophagosomes could not fully transform into autolysosomes, which was reversed by the knockdown of OPN with immunofluorescence staining (Figure 3E). The potential effects of OPN silencing in neurological outcomes after HI insult were evaluated by geotaxis reflexes and grip test. Compared with HI+si-NC mice, the times for the geotaxis reflexes in the HI+si-OPN group were significantly shortened at 1 day and 3 days after HI (Figure 3F). In the assessment of the grip test, OPN deficiency significantly restored the grip strength at 1 day, 3 days, and 7 days post-injury (Figure 3G). These data suggested that OPN deficiency provided short-term protection against neonatal HI brain injury. We then sought to determine whether this beneficial effect was long-lasting by performing a T-maze test 28 days after HI injury. The results showed that OPN down-regulation increased the alternation rate compared with the HI+si-NC group (Figure 3H,I). However, the side preference rate in each group had no significant difference (Figure 3I). ## 3.4. OPN Deficiency Reduced CTSB Release into Cytoplasm and NLRP3 Inflammasome Activation after HI Insult We next investigated whether the lysosome dysfunction induced by HI resulted in the release of CTSB into the cytosol. The expressions of CTSB mRNA and protein in the cytosol were increased following HI insult (Figure S4A and Figure 4A). The increased expression of CTSB was primarily in the Iba1+ microglia/macrophages, NeuN+ neurons, and GFAP+ astrocytes in the ipsilateral cortex at 3 days following HI (Figure 4B,C and Figure S4B–D). Moreover, the genetic suppression of OPN attenuated the HI-upregulated CTSB protein in the cytoplasm and mRNA of the ipsilateral cortex at 3 days following HI exposure (Figure 4D and Figure S4E), indicating that HI-induced damaged lysosomes and release of CTSB were reduced after OPN deficiency. The leakage of CTSB from the lysosomes to the cytoplasm can trigger the activation of the NLRP3 inflammasome in microglia/macrophages [12]. Next, we examined whether the up-regulation of OPN in microglia/macrophages might trigger the activation of the NLRP3 inflammasome and promote the production of IL-1β. The levels of NLRP3, Caspase-1, apoptosis-associated speck-like protein containing a CARD (ASC), and IL-1β mRNAs and proteins were up-regulated in the ipsilateral cortex after HI exposure (Figure S5A and Figure 4E,F). Consistent with the importance of OPN in the regulation of CTSB expression, the genetic suppression of OPN concurrently reduced the levels of NLRP3, Caspase-1, ASC, and IL-1β mRNA and protein in the ipsilateral cortex following HI exposure (Figure S5B and Figure 4G,H), demonstrating that OPN plays a critical role in NLRP3 inflammasome activation in microglia/macrophages. ## 3.5. OPN Interacts with GAL-3 to Induced Lysosomal Damage following HI Exposure As mentioned above, GAL-3, β-galactoside binding cytosolic lectin, has been shown to play a critical role in autophagy responses to lysosomal damage [27], and the genetic suppression of OPN reduced HI-induced GAL-3 expression in the ipsilateral cortex (Figure 3B). Next, we examined whether the up-regulation of OPN in the microglia/macrophages might interact with GAL-3 to induce lysosomal damage following HI insult. Blocking GAL-3 expression could alleviate HI-induced brain injury (Figure 5A–C) and HI-induced OPN and CTSB mRNA expression in the ipsilateral cortex (Figure 5D). The co-IP experiments were performed to determine the interaction between OPN and GAL-3. IP with OPN antibody using cell lysates from BV-2 cells treated with OGD found that endogenous OPN and GAL-3 formed a complex, indicating that there was a direct or indirect interaction between OPN and GAL-3 (Figure 5E). In order to observe this binding effect more intuitively, we co-transfected GAL-3-mcherry and OPN-EGFP plasmids in HEK293T cells, and the fluorescence results showed that GAL-3 was colocalized with OPN (Figure 5F). Then, in order to further explore whether GAL-3 and OPN were directly bound, we obtained GAL-3 and OPN protein translation using TnT® T7 Coupled Reticulocyte Lysate Systems in vitro and conducted co-IP experiments. The results found that there was a direct interaction between GAL-3 and OPN in vitro (Figure 5G). The above results indicated that OPN and GAL-3 directly interacted to induce lysosomal damage in microglia/macrophages. ## 3.6. Cleaved OPN Promoted Lysosomal Damage by Interacting with GAL-3 The biological activity of OPN can be modulated by proteolytic cleavage in the microenvironment and has been shown to be a substrate for thrombin and matrix metalloproteinases (MMP) [28,29]. First, we found that the levels of MMP-9 mRNA were greatly elevated at 3 days post-HI (Figure S6), confirming earlier results [30]. We found a significantly increased intensity of the full-length OPN band at ∼70 kDa, as well as three significantly increased fragments migrating at ∼20 kDa, ∼30 kDa, and ∼42 kDa detected by murine OPN antibody (Figure 6A). Next, we investigated whether cleaved OPN and intact OPN could bind GAL-3 and promote lysosome damage. Two OPN truncations, including the N-terminal fragment (Met1-Gly151) and the C-terminal (Leu152-Asn294) fragment, were constructed and expressed in vitro (Figure 6B). Plasmids of GAL-3, OPN-WT-Flag, OPN-N-Flag, and OPN-C-Flag were constructed and used to infect HEK293T cells. As shown in Figure 6C, OPN-WT and OPN-C, but not OPN-N, were bound to GAL-3. Further, we investigated the roles of OPN cleavage products in neuroinflammation by using OPN-N and OPN-C. The results showed that OPN-WT or OPN-C with GAL-3 overexpression up-regulated NLRP3, IL-1β, and CTSB mRNA expression in BV-2 cells. However, OPN-N had no effect (Figure 6D). These data indicated that WT-OPN and cleaved OPN-C bound to GAL-3, leading to lysosomal damage. ## 3.7. Role of CD44 in OPN-Induced Lysosomal Damage following HI Insult CD44, αv, and β3 are the most well-characterized integrin receptors for OPN. The induction of CD44 was restricted to activated microglia/macrophages within sites of intense neural damage in the ischemic brain [31]. In addition, αv and β3 integrin subunits were strongly induced in reactive astrocytes [32]. We found that the levels of CD44 mRNA and protein in the ipsilateral cortex were increased after HI exposure (Figure S7A and Figure 7A). Double labeling analysis indicated that CD44 immunoreactivity was mainly localized in large, round amoeboid-like brain microglia/macrophages (Figure 7B,C). In the ipsilateral cortex, a few NeuN+ neuron cells and GFAP+ astrocytes expressed CD44 (Figure S8). In addition, the mRNA level of αv integrin subunits in the ipsilateral cortex increased at 2 days post-HI, while the mRNA level of β3 integrin subunits in the ipsilateral cortex decreased at 3 days after HI (Figure S7B). As αv and β3 integrin subunits were strongly induced in reactive astrocytes after ischemia, CD44 was supposed to be the main OPN receptor in microglia/macrophages following HI in neonatal mice. To determine whether CD44 mediated OPN-induced lysosomal dysfunction following HI exposure, mice were pre-treated with si-CD44 to suppress CD44 expression. The CD44 deficiency inhibited HI-induced brain damage (Figure 7D,E and Figure S9) and the mRNA levels of LAMP1, NLRP3, and CTSB (Figure 7F). The genetic suppression of CD44 also reversed OGD exposure-displaying a perinuclear location of Lyso-Tracker Red fluorescent signal in BV-2 cells (Figure 7G,H). Moreover, blocking CD44 with antibody also attenuated HI-induced edema and infarct volume (Figure S10A–C). The inhibition of CD44 expression reduced CTSB and NLRP3 mRNA levels (Figure S10D,E). These data demonstrated that CD44 played a critical role in lysosomal dysfunction and CTSB release in the microglia. ## 3.8. Secreted OPN Stimulated Pro-Inflammatory Cytokines and CTSB Release through Binding with Cell Surface Receptors CD44 Next, we determined whether secreted OPN played a role in lysosomal dysfunction through binding with CD44. BV-2 cells were co-incubated with recombinant mouse OPN (rmOPN), and the qRT-PCR results showed that the expression of OPN mRNA was significantly up-regulated at the concentrations of 100 and 400 ng/mL rmOPN (Figure 8A). At the same time, the levels of CTSB, NLRP3, and IL-1β mRNA were up-regulated at the concentration of 400 ng/mL rmOPN in BV-2 cells (Figure 8B–D). The total cellular RNA was extracted from BV-2 cells with antibodies against CD44 or IgG and treated with/without rmOPN (No decrease in CD44 mRNA levels was observed following transfection with si-CD44 in BV-2 cells in the preliminary experiment, probably because of the low levels of CD44 in resting microglia. Herein we used CD44 antibody to block CD44 function). The Western blot results showed that anti-CD44 reversed the significant rmOPN-induced increases in the NLRP3, CTSB, and IL-1β levels in BV-2 cells (Figure 8E,F). These results suggested that secreted OPN bound to cell surface receptors CD44 and stimulated CTSB release and inflammation in BV-2 cells. ## 4. Discussion The present study demonstrated the vital role of OPN in HI-mediated lysosomal damage and autophagosome accumulation. We found that, at the early stage of ischemia, OPN expression was enhanced, especially in microglia, and colocalized with LAMP1 and GAL-3, which was accompanied by lysosomal damage, CTSB release, NLRP3 inflammasome activation, and autophagosomes accumulation after HI insult. Importantly, the knockdown of OPN expression significantly rescued the lysosomal damage with significant improvements in the autophagic flux following HI insult in neonatal mice, thus alleviating the HI brain damage. We identified, for the first time, the presence of OPN cleavage activity in the brain following HI insult. OPN and its cleavage product interacted with GAL-3, and secreted OPN combined with CD44 led to lysosomal damage and exacerbated autophagosome accumulation after HI exposure. Lysosomal dysfunction, as reflected by cytosolic acidification and rupture/permeabilization, was detected after ischemic insult [33]. Cathepsins play an important role in a range of cellular activities, such as protein degradation, antigen presentation, and cell death [34]. Several lines of evidence suggest that the leakage of CTSB from the lysosomes into the cytoplasm could activate the NLRP3 inflammasome and affect the processing and secretion of pro-inflammatory cytokines [35,36,37]. In the present study, we found that acidic compartments accumulated and the expression of lysosomal markers (LAMP1, GAL-3, and CTSB) was higher in the ipsilateral cortex of HI mice, suggesting the existence of a disruption of the lysosomal compartment in HI animals. Autophagy terminates with the degradation of the autophagosome contents in the lysosomes [38,39]. Our results demonstrated that HI insult led to a significant increase in the ratio of LC3-II to LC3-I over time. Simultaneously, the up-regulation of p62 was accompanied by an increase in autophagosomes, indicating that autophagic flux was inhibited post-HI. The increase in LC3-II could have been due to either the excessive initiation of autophagy or poor lysosomal clearance. We observed defects in lysosomal function after HI injury; thus, it is likely that the accumulation of LC3II in our study was due to lysosomal dysfunction. Therefore, the improvement of lysosomal dysfunction and autophagic flux are the current targets for the development of HI brain-damage drugs. Here, the PPI network showed that lysosomal cysteine proteases (cathepsins B, L, S, and Z, galectin-1, and GAL-3) exhibited potential relationships with OPN. Moreover, LAMP1, CTSB, GAL-3 proteins, and respective mRNAs were significantly increased and expressed in the lysosomes of microglia/macrophages in the brain after HI. OPN deficiency abrogated the processing and release of CTSB and NLRP3 inflammasome activation in HI-exposed microglia, as well as recovered impaired autophagic flux, suggesting that an increase in OPN in the microglia could cause defects in lysosomal function and abnormal autophagosome accumulation. Within the brain, GAL-3 is expressed by microglia and some astrocytes, and weakly in some cortical neurons [40]. The extracellular or intracellular GAL-3 levels are elevated in a variety of pathologies, potentially due to neuroinflammation [41,42]. GAL-3 modulates the inflammatory response of the nervous system and has been implicated in the pathogenesis of diverse neurological diseases, such as models of stroke [43,44] and neonatal HI brain injury [45,46]. Extracellular GAL-3 could activate microglia by directly activating TLR4 [47]. The genetic deletion of GAL-3 protected against HI injury, particularly in the hippocampus and striatum in neonatal mice [48]. In adult Huntington’s disease mice, the up-regulation of GAL-3 formed puncta in damaged lysosomes in primary microglia and contributed to inflammation through NF-κB and NLRP3 inflammasome-dependent pathways [24]. Consistent with these previous findings, we found, in the present study, that the down-regulation of GAL-3 in the HI-injured mice not only attenuated brain damage, but also reduced HI-induced OPN and CTSB expression. Moreover, the co-IP result demonstrated a direct protein–protein interaction between OPN and GAL-3. The suppression of OPN was able to reduce HI-induced GAL-3 expression. Collectively, these observations suggest that intracellular OPN played a role in lysosomal damage through the association of OPN with GAL-3 in microglia. It is noteworthy that OPN is not a sole cytokine, but rather a compound arrangement of multiple peptides that includes splice variants and several active proteolytic cleavage products. For example, OPN cleaved by MMP forms the OPN-N-terminal fragment and the OPN-C terminal fragment, the molecular weights of which are approximately ~40 kDa, ~32 kDa, and ~25 kDa, respectively [49]. Recent studies have found that the MMP-3 or 7 cleavage of OPN-C fragments results has aα9β1 binding site, which facilitates its role in the development of inflammatory arthritis [50]. Another study has shown that OPN-C promotes inflammation by the activation of the NF-κB pathway [51]. We found that OPN-WT and cleaved forms of OPN coexisted in the ipsilateral cortex following HI injury in neonatal mice, and the C-terminal domain of OPN was the molecular basis for the direct binding between OPN and GAL-3. WT-OPN and the cleavage of OPN increased pro-inflammatory cytokines and CTSB release. Taken together, our data indicate an amplifying loop of OPN and GAL-3 in HI-associated lysosomal damage that drives autophagosome accumulation and brain injury. OPN is known to be involved in various pathophysiological events and has been studied as a secreted protein (secreted OPN). Secreted OPN can bind to multiple integrins, such as αvβ3, αvβ5, αvβ1, and α5β1, and to certain variant forms of CD44 [52]. OPN has multiple effects in different cell types, with distinct outcomes for disease phenotypes because of these varied receptors. For instance, OPN regulates hepatitis C virus replication and assembly by binding to the receptors αVβ3 and CD44 [53]. OPN-upregulated cyclooxygenase-2 expression in tumor-associated macrophages leads to enhanced angiogenesis and tumor growth via α9β1 integrin [54]. Macrophage-secreted OPN binds to CD44 on the tumor cells and promotes tumor invasion and clonal growth [55]. In the current study, we observed that secreted OPN dramatically up-regulated NLRP3 IL-1β and CTSB expression. Small interfering-mediated knockdown and antibody neutralization experiments identified CD44 as the OPN receptor that stimulated CTSB expression and NLRP3 inflammasome activation in microglia. Collectively, our observations indicated a critical role for secreted OPN in microglia-mediated neuroinflammation through binding to CD44. ## 5. Conclusions There were also some limitations in our study. Firstly, lysosomal membrane destabilization may lead to the release of various lysosomal cysteine proteases into the cytosol. There might be other signaling factors that are affected by OPN. Second, several studies, including our previous study, showed that activated microglia/macrophages are the main cellular sources of OPN [7]; therefore, the effects of OPN deficiency on neuroinflammation in this study were examined by shRNA. In the future, the use of microglia-specific depletion of OPN mice might be considered to accurately measure the effects of OPN deficiency on neuroinflammation. Thirdly, in the current study, we were able to detect the band corresponding to the cleaved form without the need for any artificial digestion. The commercial murine antibodies directed against the OPN-N-terminal fragment or OPN-C-terminal fragment are not available, and we did not specifically detect OPN cleavage. Finally, the role of OPN cleavage in HI injury has not been investigated yet. Furthermore, the underlying mechanisms and modulating signals are poorly understood. A striking observation of our studies is the regulation of lysosomal function by endogenous OPN, as well as secreted forms of OPN through CD44-mediated signaling pathways. We demonstrated the potential involvement of intracellular OPN in lysosomal damage and autophagic flux through the interaction of OPN and/or OPN cleavage with GAL-3 in the microglia. Secreted OPN was able to bind to microglial CD44 in an autocrine/paracrine manner and induce lysosomal damage. The knockdown of OPN expression could restore lysosomal damage with significant improvement in the autophagic flux and protect the brain against HI insult. ## References 1. Bellahcène A., Castronovo V., Ogbureke K.U., Fisher L.W., Fedarko N.S.. Small integrin-binding ligand N-linked glycoproteins (SIBLINGs): Multifunctional proteins in cancer. Nat. Rev. Cancer (2008) 8 212-226. DOI: 10.1038/nrc2345 2. Gliem M., Krammes K., Liaw L., van Rooijen N., Hartung H.P., Jander S.. Macrophage-derived osteopontin induces reactive astrocyte polarization and promotes re-establishment of the blood brain barrier after ischemic stroke. Glia (2015) 63 2198-2207. DOI: 10.1002/glia.22885 3. Rentsendorj A., Sheyn J., Fuchs D.T., Daley D., Salumbides B.C., Schubloom H.E., Hart N.J., Li S., Hayden E.Y., Teplow D.B.. A novel role for osteopontin in macrophage-mediated amyloid-β clearance in Alzheimer’s models. Brain Behav. Immun. (2018) 67 163-180. DOI: 10.1016/j.bbi.2017.08.019 4. Chan J.L., Reeves T.M., Phillips L.L.. Phillips. Osteopontin expression in acute immune response mediates hippocampal synaptogenesis and adaptive outcome following cortical brain injury. Exp. Neurol. (2014) 261 757-771. DOI: 10.1016/j.expneurol.2014.08.015 5. Shen X., Qiu Y., Wight A.E., Kim H.J., Cantor H.. Definition of a mouse microglial subset that regulates neuronal development and proinflammatory responses in the brain. Proc. Natl. Acad. Sci. USA (2022) 119 e2116241119. DOI: 10.1073/pnas.2116241119 6. Yim A., Smith C., Brown A.M.. Osteopontin/secreted phosphoprotein-1 harnesses glial-, immune-, and neuronal cell ligand-receptor interactions to sense and regulate acute and chronic neuroinflammation. Immunol. Rev. (2022) 311 224-233. DOI: 10.1111/imr.13081 7. Xin D., Li T., Chu X., Ke H., Liu D., Wang Z.. MSCs-extracellular vesicles attenuated neuroinflammation, synapse damage and microglial phagocytosis after hypoxia-ischemia injury by preventing osteopontin expression. Pharmacol. Res. (2021) 164 105322. DOI: 10.1016/j.phrs.2020.105322 8. Li T.-T., Xin D.-Q., Ke H.-F., Chu X.-L., Zhao Y.-J., Yue S.-W., Liu D.-X., Wang Z.. L-Cysteine attenuates osteopontin-mediated neuroinflammation following hypoxia-ischemia insult in neonatal mice by inducing S-sulfhydration of Stat3. Acta Pharmacol. Sin. (2021) 43 1658-1669. DOI: 10.1038/s41401-021-00794-2 9. Luzio J.P., Pryor P.R., Bright N.A.. Lysosomes: Fusion and function. Nat. Rev. Mol. Cell Biol. (2007) 8 622-632. DOI: 10.1038/nrm2217 10. Boya P., Kroemer G.. Lysosomal membrane permeabilization in cell death. Oncogene (2008) 27 6434-6451. DOI: 10.1038/onc.2008.310 11. Liu C., Yao Q., Hu T., Cai Z., Xie Q., Zhao J., Yuan Y., Ni J., Wu Q.Q.. Cathepsin B deteriorates diabetic cardiomyopathy induced by streptozotocin via promoting NLRP3-mediated pyroptosis. Mol. Ther. Nucleic Acids (2022) 30 198-207. DOI: 10.1016/j.omtn.2022.09.019 12. Hu Y., Shi Y., Chen H., Tao M., Zhou X., Li J., Ma X., Wang Y., Liu N.. Blockade of autophagy prevents the progression of hyperuricemic nephropathy through inhibiting NLRP3 inflammasome-mediated pyroptosis. Front. Immunol. (2022) 13 858494. DOI: 10.3389/fimmu.2022.858494 13. Stone B.S., Zhang J., Mack D.W., Mori S., Martin L.J., Northington F.J.. Delayed neural network degeneration after neonatal hypoxia-ischemia. Ann. Neurol. (2008) 64 535-546. DOI: 10.1002/ana.21517 14. Li C., Sun G., Chen B., Xu L., Ye Y., He J., Bao Z., Zhao P., Miao Z., Zhao L.. Nuclear receptor coactivator 4-mediated ferritinophagy contributes to cerebral ischemia-induced ferroptosis in ischemic stroke. Pharmacol. Res. (2021) 174 105933. DOI: 10.1016/j.phrs.2021.105933 15. Li Y., Zhou D., Ren Y., Zhang Z., Guo X., Ma M., Xue Z., Lv J., Liu H., Xi Q.. Mir223 restrains autophagy and promotes CNS inflammation by targeting ATG16L1. Autophagy (2019) 15 478-492. DOI: 10.1080/15548627.2018.1522467 16. Maejima Y., Kyoi S., Zhai P., Liu T., Li H., Ivessa A., Sciarretta S., Del Re D.P., Zablocki D.K., Hsu C.-P.. Mst1 inhibits autophagy by promoting the interaction between Beclin1 and Bcl-2. Nat. Med. (2013) 19 1478-1488. DOI: 10.1038/nm.3322 17. Henn A., Lund S., Hedtjärn M., Schrattenholz A., Pörzgen P., Leist M.. The suitability of BV2 cells as alternative model system for primary microglia cultures or for animal experiments examining brain inflammation. Altern. Anim. Exp. (2009) 26 83-94. DOI: 10.14573/altex.2009.2.83 18. Zhang X., Ding K., Ji J., Parajuli H., Aasen S.N., Espedal H., Huang B., Chen A., Wang J., Li X.. Trifluoperazine prolongs the survival of experimental brain metastases by STAT3-dependent lysosomal membrane permeabilization. Am. J. Cancer Res. (2020) 10 545-563. PMID: 32195026 19. Bai X., Liu S., Yuan L., Xie Y., Li T., Wang L., Wang X., Zhang T., Qin S., Song G.. Hydrogen-rich saline mediates neuroprotection through the regulation of endoplasmic reticulum stress and autophagy under hypoxia-ischemia neonatal brain injury in mice. Brain Res. (2016) 1646 410-417. DOI: 10.1016/j.brainres.2016.06.020 20. Kim J.Y., Grunke S.D., Levites Y., Golde T.E., Jankowsky J.L.. Intracerebroventricular viral injection of the neonatal mouse brain for persistent and widespread neuronal transduction. J. Vis. Exp. JoVE (2014) 91 51863 21. Li T., Chu X., Xin D., Ke H., Wang S., Liu D., Chen W., Wang Z.. H. Biomed. Pharmacother. (2021) 135 111207. DOI: 10.1016/j.biopha.2020.111207 22. Rotshenker S.. Galectin-3 (MAC-2) controls phagocytosis and macropinocytosis through intracellular and extracellular mechanisms. Front. Cell. Neurosci. (2022) 16 949079. DOI: 10.3389/fncel.2022.949079 23. Cai B., Zhao J., Zhang Y., Liu Y., Ma C., Yi F., Zheng Y., Zhang L., Chen T., Liu H.. USP5 attenuates NLRP3 inflammasome activation by promoting autophagic degradation of NLRP3. Autophagy (2022) 18 990-1004. DOI: 10.1080/15548627.2021.1965426 24. Siew J.J., Chen H.M., Chen H.Y., Chen H.L., Chen C.M., Soong B.W., Wu Y.R., Chang C.P., Chan Y.C., Lin C.H.. Galectin-3 is required for the microglia-mediated brain inflammation in a model of Huntington’s disease. Nat. Commun. (2019) 10 3473. DOI: 10.1038/s41467-019-11441-0 25. Xin D., Chu X., Bai X., Ma W., Yuan H., Qiu J., Liu C., Li T., Zhou X., Chen W.. l-Cysteine suppresses hypoxia-ischemia injury in neonatal mice by reducing glial activation, promoting autophagic flux and mediating synaptic modification via H2S formation. Brain Behav. Immun. (2018) 73 222-234. DOI: 10.1016/j.bbi.2018.05.007 26. Ni J., Wu Z., Peterts C., Yamamoto K., Qing H., Nakanishi H.. The critical role of proteolytic relay through cathepsins B and E in the phenotypic change of microglia/macrophage. J. Neurosci. (2015) 35 12488-12501. DOI: 10.1523/JNEUROSCI.1599-15.2015 27. Jia J., Claude-Taupin A., Gu Y., Choi S.W., Peters R., Bissa B., Mudd M.H., Allers L., Pallikkuth S., Lidke K.A.. Galectin-3 Coordinates a Cellular System for Lysosomal Repair and Removal. Dev. Cell (2020) 52 69-87.e8. DOI: 10.1016/j.devcel.2019.10.025 28. Lindsey M.L., Zouein F.A., Tian Y., Padmanabhan Iyer R., de Castro Brás L.E.. Osteopontin is proteolytically processed by matrix metalloproteinase 9. Can. J. Physiol. Pharmacol. (2015) 93 879-886. DOI: 10.1139/cjpp-2015-0019 29. Peraramelli S., Zhou Q., Zhou Q., Wanko B., Zhao L., Nishimura T., Leung T.H., Mizuno S., Ito M., Myles T.. Thrombin cleavage of osteopontin initiates osteopontin’s tumor-promoting activity. J. Thromb. Haemost. (2022) 20 1256-1270. DOI: 10.1111/jth.15663 30. Yang Y., Rosenberg G.A.. Matrix metalloproteinases as therapeutic targets for stroke. Brain Res. (2015) 1623 30-38. DOI: 10.1016/j.brainres.2015.04.024 31. Wang H., Zhan Y., Xu L., Feuerstein G.Z., Wang W.. Use of suppression subtractive hybridization for differential gene expression in stroke: Discovery of CD44 gene expression and localization in permanent focal stroke in rats. Stroke (2001) 32 1020-1027. DOI: 10.1161/01.STR.32.4.1020 32. Kang W.S., Choi J.S., Shin Y.J., Kim H.Y., Cha J.H., Lee J.Y., Chun M.H., Lee M.Y.. Differential regulation of osteopontin receptors, CD44 and the alpha(v) and beta(3) integrin subunits, in the rat hippocampus following transient forebrain ischemia. Brain Res. (2008) 1228 208-216. DOI: 10.1016/j.brainres.2008.06.106 33. Troncoso M., Bannoud N., Carvelli L., Asensio J., Seltzer A., Sosa M.A.. Hypoxia-ischemia alters distribution of lysosomal proteins in rat cortex and hippocampus. Biol. Open (2018) 7 bio036723. DOI: 10.1242/bio.036723 34. Ruiz-Blázquez P., Pistorio V., Fernández-Fernández M., Moles A.. The multifaceted role of cathepsins in liver disease. J. Hepatol. (2021) 75 1192-1202. DOI: 10.1016/j.jhep.2021.06.031 35. Bai H., Yang B., Yu W., Xiao Y., Yu D., Zhang Q.. Cathepsin B links oxidative stress to the activation of NLRP3 inflammasome. Exp. Cell Res. (2018) 362 180-187. DOI: 10.1016/j.yexcr.2017.11.015 36. Bruchard M., Mignot G., Derangère V., Chalmin F., Chevriaux A., Végran F., Boireau W., Simon B., Ryffel B., Connat J.L.. Chemotherapy-triggered cathepsin B release in myeloid-derived suppressor cells activates the Nlrp3 inflammasome and promotes tumor growth. Nat. Med. (2013) 19 57-64. DOI: 10.1038/nm.2999 37. Tang T.T., Lv L.L., Pan M.M., Wen Y., Wang B., Li Z.L., Wu M., Wang F.M., Crowley S.D., Liu B.C.. Hydroxychloroquine attenuates renal ischemia/reperfusion injury by inhibiting cathepsin mediated NLRP3 inflammasome activation. Cell Death Dis. (2018) 9 351. DOI: 10.1038/s41419-018-0378-3 38. Zhang X., Wei M., Fan J., Yan W., Zha X., Song H., Wan R., Yin Y., Wang W.. Ischemia-induced upregulation of autophagy preludes dysfunctional lysosomal storage and associated synaptic impairments in neurons. Autophagy (2021) 17 1519-1542. DOI: 10.1080/15548627.2020.1840796 39. Fleming A., Bourdenx M., Fujimaki M., Karabiyik C., Krause G.J., Lopez A., Martín-Segura A., Puri C., Scrivo A., Skidmore J.. The different autophagy degradation pathways and neurodegeneration. Neuron (2022) 110 935-966. DOI: 10.1016/j.neuron.2022.01.017 40. Yoo H.-I., Kim E.-G., Lee E.-J., Hong S.-Y., Yoon C.-S., Hong M.-J., Park S.-J., Woo R.-S., Baik T.-K., Song D.Y.. Neuroanatomical distribution of galectin-3 in the adult rat brain. J. Mol. Histol. (2017) 48 133-146. DOI: 10.1007/s10735-017-9712-9 41. Puigdellívol M., Allendorf D.H., Brown G.C.. Sialylation and galectin-3 in microglia-mediated neuroinflammation and neurodegeneration. Front. Cell. Neurosci. (2020) 14 162. DOI: 10.3389/fncel.2020.00162 42. Tan Y., Zheng Y., Xu D., Sun Z., Yang H., Yin Q.. Galectin-3: A key player in microglia-mediated neuroinflammation and Alzheimer’s disease. Cell Biosci. (2021) 11 78. DOI: 10.1186/s13578-021-00592-7 43. Lalancette-Hébert M., Swarup V., Beaulieu J.M., Bohacek I., Abdelhamid E., Weng Y.C., Sato S., Kriz J.. Galectin-3 is required for resident microglia activation and proliferation in response to ischemic injury. J. Neurosci. (2012) 32 10383-10395. DOI: 10.1523/JNEUROSCI.1498-12.2012 44. Wang Q., Wang K., Ma Y., Li S., Xu Y.. Serum Galectin-3 as a potential predictive biomarker is associated with poststroke cognitive impairment. Oxidative Med. Cell. Longev. (2021) 2021 5827812. DOI: 10.1155/2021/5827812 45. Sävman K., Heyes M.P., Svedin P., Karlsson A.. Microglia/macrophage-derived inflammatory mediators galectin-3 and quinolinic acid are elevated in cerebrospinal fluid from newborn infants after birth asphyxia. Transl. Stroke Res. (2013) 4 228-235. DOI: 10.1007/s12975-012-0216-3 46. Wang Q., Diao S., Qiu H., Gao R., Wang M., Chen Q., Xiao M., Li Z., Chen C.. Galectin-3 administration drives remyelination after hypoxic-ischemic induced perinatal white matter injury. Front. Cell. Neurosci. (2022) 16 976002. DOI: 10.3389/fncel.2022.976002 47. Burguillos M.A., Svensson M., Schulte T., Boza-Serrano A., Garcia-Quintanilla A., Kavanagh E., Santiago M., Viceconte N., Oliva-Martin M.J., Osman A.M.. Microglia-secreted galectin-3 acts as a toll-like receptor 4 ligand and contributes to microglial activation. Cell Rep. (2015) 10 1626-1638. DOI: 10.1016/j.celrep.2015.02.012 48. Doverhag C., Hedtjärn M., Poirier F., Mallard C., Hagberg H., Karlsson A., Sävman K.. Galectin-3 contributes to neonatal hypoxic-ischemic brain injury. Neurobiol. Dis. (2010) 38 36-46. DOI: 10.1016/j.nbd.2009.12.024 49. Gao Y.A., Agnihotri R., Vary C.P.H., Liaw L.. Expression and characterization of recombinant osteopontin peptides representing matrix metalloproteinase proteolytic fragments. Matrix Biol. (2004) 23 457-466. DOI: 10.1016/j.matbio.2004.09.003 50. Kon S., Nakayama Y., Matsumoto N., Ito K., Kanayama M., Kimura C., Kouro H., Ashitomi D., Matsuda T., Uede T.. A novel cryptic binding motif, LRSKSRSFQVSDEQY, in the C-terminal fragment of MMP-3/7-cleaved osteopontin as a novel ligand for α9β1 integrin is involved in the anti-type II collagen antibody-induced arthritis. PLoS ONE (2014) 9. DOI: 10.1371/journal.pone.0116210 51. Hamias R., Rudich A., Greenberg G., Szendro G., Wolak T.. Angiotensin 1-7, but not the thrombin-cleaved osteopontin C-terminal fragment, attenuates osteopontin-mediated macrophage-induced endothelial-cell inflammation. Inflamm. Res. (2018) 67 265-275. DOI: 10.1007/s00011-017-1120-9 52. Rangaswami H., Bulbule A., Kundu G.C.. Osteopontin: Role in cell signaling and cancer progression. Trends Cell Biol. (2006) 16 79-87. DOI: 10.1016/j.tcb.2005.12.005 53. Iqbal J., Sarkar-Dutta M., McRae S., Ramachandran A., Kumar B., Waris G.. Osteopontin regulates Hepatitis C virus (HCV) replication and assembly by interacting with HCV proteins and lipid droplets and by binding to receptors αVβ3 and CD44. J. Virol. (2018) 92 e02116-e02117. DOI: 10.1128/JVI.02116-17 54. Kale S., Raja R., Thorat D., Soundararajan G., Patil T.V., Kundu G.C.. Osteopontin signaling upregulates cyclooxygenase-2 expression in tumor-associated macrophages leading to enhanced angiogenesis and melanoma growth via α9β1 integrin. Oncogene (2014) 33 2295-2306. DOI: 10.1038/onc.2013.184 55. Ahmed M., Sottnik J.L., Dancik G.M., Sahu D., Hansel D.E., Theodorescu D., Schwartz M.A.. An osteopontin/CD44 axis in RhoGDI2-mediated metastasis suppression. Cancer Cell (2016) 30 432-443. DOI: 10.1016/j.ccell.2016.08.002
--- title: 'Effects of a Diabetes Sports Summer Camp on the Levels of Physical Activity and Dimensions of Health-Related Quality of Life in Young Patients with Diabetes Mellitus Type 1: A Randomized Controlled Trial' authors: - Lida Skoufa - Eleni Makri - Vassilis Barkoukis - Maria Papagianni - Panagiota Triantafyllou - Evangelia Kouidi journal: Children year: 2023 pmcid: PMC10046943 doi: 10.3390/children10030456 license: CC BY 4.0 --- # Effects of a Diabetes Sports Summer Camp on the Levels of Physical Activity and Dimensions of Health-Related Quality of Life in Young Patients with Diabetes Mellitus Type 1: A Randomized Controlled Trial ## Abstract Physical activity (PA) is considered an important part of the treatment of children with diabetes mellitus type 1 (T1DM). Furthermore, health-related quality of life (HRQoL) affects both the physical and mental health of patients with T1DM. The purpose of the study was to evaluate through a randomized controlled trial the impact of participation in a summer diabetes sports camp on the PA and HRQoL of children and adolescents with T1DM. Eighty-four children and adolescents with T1DM were randomly assigned into an intervention ($M = 12.64$, SD = 1.82, 30 female) and a control group ($M = 12.67$, SD = 2.50, 30 female). Intervention group participants attended a ten-day summer diabetes sports camp which included an intensive program of PA (6 h of daily PA), educational and entertaining activities as well as education on the importance of PA in the management of the disease. At baseline and at the end of the study, participants completed measures of physical activity, self-esteem, depression, health status, intention to change behavior, and life satisfaction. Results of the two-way repeated measures analysis showed no statistically significant group differences in PA levels ($p \leq 0.05$) and HRQoL parameters ($p \leq 0.05$ for all parameters). In conclusion, the results did not support the effectiveness of a 10-day diabetes sports camp on PA levels and HRQoL for children with T1DM. Longer interventions may be more effective in exerting positive influence on trait parameters of children with T1DM’s quality of life. Participation in such programs on multiple occasions should be evaluated in the future. ## 1. Introduction T1DM is a chronic metabolic disease that may occur at any age but usually has a juvenile onset and accounts for the 5–$10\%$ of diabetic patients [1]. T1DM is characterized as a disease of the developed countries, but there has been an increasing trend in its occurrence over the last three decades worldwide. It has been estimated that worldwide, the prevalence of T1DM is $9.5\%$ [1]. In T1DM diabetes, high blood glucose levels pose serious health risks, and its proper regulation delays or even prevents the appearance of its complications [2]. Research evidence has shown high rates of mental disorders in children and adolescents with diabetes, including anxiety disorders, eating disorders and particularly major depressive disorder, that range from 33–$47\%$ [3,4]. The prevalence of depression in children with diabetes is two-fold greater than their peers in the general population, while the prevalence in adolescents is three-fold greater [5]. Evidence has shown that $32\%$ of youth with diabetes experience symptoms of anxiety [6], which has been linked to worse treatment adherence, symptom control and management of the disease [7]. HRQoL is considered to significantly affect physical and mental health as well as the mentality and behavior of patients with T1DM in their everyday life [8]. Almost a decade ago, studies showed that HRQoL of young patients with T1DM was poor compared to their healthy peers, while good metabolic control has been associated with a better QoL [9,10]. According to research evidence, engaging in frequent PA has several health benefits for both healthy and diseased people. Specifically, the evidence that is currently available suggests that PA has a beneficial impact on body composition, functional capacity, insulin sensitivity, glycemic management, glycosylated hemoglobin levels, and stress levels in individuals with T1DM [11,12]. It has been found that only a small percentage of children and adolescents with T1DM take part in exercise programs as part of their leisure and school activities [13]. In addition, previous research has demonstrated that they are unable to reach recommended levels of PA and have lower PA levels when compared to their healthy peers [14]. In order to design exercise programs for children and adolescents with T1DM, diabetes camps have been organized worldwide [15]. These camps aim to promote healthy habits and encourage children with T1DM to adopt an active and healthy lifestyle, and improve their overall HRQoL [15,16]. Researchers have only lately begun to demonstrate the benefits of camps for children and adolescents who suffer from chronic diseases [17], despite the fact that these camps have been around for a very long time. There are contradictory findings in the research about the impact of camps on the HRQoL of children and adolescents with T1DM. Participation in diabetes camps has been shown to positively contribute to the education of young T1DM patients about the disease and its management [17,18]. This, in turn, leads to significant improvements in the glycemic control, self-esteem, and mobility of patients who attend such camps. A recent study [19] showed that although there was a decrease in PA levels one month after participation in an adult diabetes camp, there was a significant increase in a three and six month follow up. This was the case despite the fact that there was a decrease in PA levels one month after participation. A comparative study in children with T1DM who either participated in diabetes camps or attended regular schools showed that the first group met the recommended number of steps per day. In addition, it was discovered that attending diabetes camps offered considerable benefits for health-related issues for children diagnosed with T1DM [20]. On the other hand, participation in diabetes camps did not seem to improve quality of life, anxiety, or psychological status in any of the previous investigations [21,22]. Nevertheless, despite the ambiguous results, there is a general consensus that camp programs for children and adolescents with T1DM offer them the opportunity to interact with their peers in a safe environment while also being supported and supervised by a medical team that specializes in their condition [23]. Furthermore, studies have shown that participation in camps contributes positively to the education of T1DM children and adolescents about the disease and its management, leading to significant improvements in their glycaemic control, their self-esteem and their mobility [18,22]. In general, physical activity has been suggested as an important component of the treatment of children and adolescents with T1DM. Previous research on diabetes camps for young patients has demonstrated that participation is useful in boosting PA, providing benefits for certain quality of life and psychological measures, and increasing the participants’ illness awareness. To the best of our knowledge, no previous research has investigated whether diabetes camps have any positive benefits for young people living with T1DM. In order to fill this gap, the purpose of the present study was to evaluate whether participation of children with T1DM in a short-term summer diabetes sports camp, alongside their healthy peers and under the supervision of a specialized medical team, would assist these children in increasing their levels of PA, thereby improving their HRQoL and psychological status. ## 2.1. Participants Children attending the Endocrine Unit of the 3rd Pediatric Department of the Aristotle University of Thessaloniki in the Hippokration Hospital of Thessaloniki, outpatient clinics or private physicians’ practices were approached to participate in the study ($$n = 134$$). In order to determine the adequate sample size that was required to detect our effects, we calculated an a priori power analysis using GPower [24,25]. The F test was selected as the test family, and the repeated measures within subjects ANOVA as the statistical test. Alpha was set to 0.05, and the effect size was set to $f = 0.25.$ The analysis showed that a sample size of 36 participants was required. Participants in the study were required to have been diagnosed with T1DM for at least a year prior to the beginning of the research. Eligible participants ranged in age from 7 to 18 years old. In addition, they should have been maintaining a stable insulin regimen. Exclusion criteria included suffering from another kind of diabetes or another condition that affects ability to exercise (e.g., cardiorespiratory and musculoskeletal diseases). The sample of the study consisted of eighty-four children and adolescents with T1DM who met the eligibility criteria for the study and provided their consent to take part in it. A medical history was obtained from all volunteers, and they also underwent a clinical evaluation and HbA1c measurement in the Diabetes Outpatient Clinic of the 3rd Pediatric Department of the Aristotle University of Thessaloniki. They were all on intensive insulin treatment (a multiple daily insulin injection regimen or continuous subcutaneous insulin infusion via pump). A total of 63 participants were using either continuous or flash glucose monitoring, while the remaining 21 were using a glucose meter. Some 43 participants were on therapy with pens, and 41 were on therapy with pumps. None of the participants had any diabetes-related complications. ## 2.2. Design and Procedure The study was a randomized controlled trial (Figure 1). After the baseline evaluation, the online statistical computing web program “www.randomizer.org (accessed on 30 June 2021)” was used for the randomization process. Children and adolescents with T1DM were randomly assigned into two groups: the intervention group ($$n = 42$$), who participated in a 10-day summer diabetes sports camp, and the control group ($$n = 42$$), who were asked to continue their daily routine without participating in any physical activity intervention. The study protocol was approved by the Ethics Committee of the School of Physical Education and Sport Science of the Aristotle University of Thessaloniki. After being thoroughly informed about the procedures, all parents gave their written informed consent for their children’s participation in the study. ## 2.3. Measurements At baseline and at the end of the 10-day study, all participants were asked to complete a battery of questionnaires measuring PA and four parameters of traits of HRQoL: life satisfaction, self-esteem, depression, health status, coping strategies and intention to change health-related behavior. The questionnaires had been used in Greece previously [26]. ## 2.3.1. Physical Activity Physical activity was measured with the Godin–Shephard Leisure-Time Physical Activity Questionnaire [27], consisting of two questions. The first question assessed the frequency and intensity of the PA in which the respondents participated during the period of a week. More specifically, the frequency of PA was estimated in three types of exercise, intense, moderate and mild, for more than 15 min during leisure time. The second question assessed the frequency of participation (often, sometimes, rarely or never) in any PA in leisure time, during which the participant sweated or his heart “beat” quickly, for a 7-day period. ## 2.3.2. Life Satisfaction Life Satisfaction was assessed with the Life Satisfaction Assessment Scale [28] which is designed to measure global cognitive crises of life satisfaction and was developed as a measure of the critical component of subjective well-being [29]. Participants indicated how much they agree or disagree with each of five items (e.g., “I am satisfied with my life”) using a 7-point scale ranging from 7 (I totally agree) to 1 (I totally disagree). In the present study, Cronbach’s alpha was calculated at 0.30. ## 2.3.3. Self-Esteem Self-esteem was measured with the Rosenberg Self-Esteem Scale [30]. The scale consisted of ten items evaluating both positive (e.g., “I feel that I have a number of good qualities”) and negative (e.g., “At times I think I am no good at all”) feelings towards the self. Questions were answered using a 4-point Likert scale from 1 (I strongly disagree) to 4 (I totally agree). Cronbach’s alpha was calculated at 0.76. ## 2.3.4. Depression Depressive symptoms were assessed via the Centre for Epidemiological Studies Depression scale (CES-D) [31]. The scale consists of 20 items estimating the frequency of depressive symptoms during the past week. An example of such one such item is ‘I felt tearful’. This instrument has received support with regard to its validity and reliability in older populations [32]. In the present study, Cronbach’s alpha was calculated at 0.64. ## 2.3.5. Health Status and Its Treatment and Intent to Change Health-Related Behaviour (HAPA) The health status assessment, its treatment and the intention to change health-related behavior questionnaire was used [33]. The questionnaire consists of a combination of items classified into three categories. The first category includes seven items measuring health status (e.g., “In general, how would you say your health is?”), changes that have occurred in the health status (e.g., “Compared to my best health status ever, my health in general now is…”) and perceptions about health status in everyday life (e.g., “How much is your everyday life affected by your health?”). Answers were given on 5-point scale, from 1 (bad) to 5 (excellent) for health status and changes in the health status, and on a 5-point scale from 1 (not at all) to 5 (too much) for health status in everyday life. Cronbach’s alpha was calculated at 0.64. The next category concerned action plans for the next period and the adoption of health-promoting behaviors that interpret the participants’ intention to change their eating habits, PA, smoking cessation and participation in medical tests (e.g., “I intend to live a healthier life”). The ten questions were answered on a 7-point scale from 1 (not at all) to 7 (very much). Cronbach’s alpha was calculated at 0.75. The third category evaluated the planning of physical activities and action plans to address any difficulties or obstacles in their implementation. It consisted of nine items (e.g., “I already have concrete plans to exercise”) assessed on a 4-point scale from 1 (not at all) to 4 (very much.). Cronbach’s alpha was calculated at 0.85. The overall high score from the three categories showed a strong intention to change health-related behaviors. ## 2.4. Intervention Design Children and adolescents diagnosed with T1DM who made up the intervention group attended a 10 day diabetes summer sports camp with their healthy peers. All children took part in the exact same pursuits and shared the exact same environment and sleeping quarters. An intensive program of daily physical activity that included three hours in the morning and three hours in the afternoon was part of the intervention (Table 1). Activities such as swimming, sailing, diving, climbing, sports games, football, tennis volleyball, basketball, handball, ping pong, canoeing, archery, team games, and athletics were among those included in the program. During this period, students also took part in a variety of other events, which included both informative and enjoyable activities (e.g., singing, dancing, daily trips). During the study period, children and adolescents with T1DM had medical supervision. Every day, sessions were held to educate them on the importance of PA for the achievement of good glycaemic control and a better general health status, and the role of a healthy lifestyle in the disease management. Participants in the control group were not subjected to any physical activity intervention. Instead, they were instructed to carry on with their typical day-to-day routines, while also taking a vacation with their parents over the same time period. ## 2.5. Statistical Analysis The analyses were conducted with SPSS 25.0. A Shapiro–Wilk test was used to verify the normality of the distributions of the parameters under study. The correlations between variables were examined with the Pearson correlation coefficient test. A two-way repeated measures analysis of variance was conducted to test for differences in the dependent variables between the intervention and control groups. A two-tailed p value < 0.05 was considered statistically significant. ## 3.1. Preliminary Analyses The characteristics of the participants are presented in Table 2. The mean body mass index (BMI) was 20.15 (SD = 3.36) for the intervention group and 20.15 (SD = 4.39) for the control group, indicating a normal weight status. HbA1c levels were above the recommended target for both groups. There were no statistically significant differences ($p \leq 0.050$) between the two groups in patients’ characteristics, PA and HRQoL indices at baseline. Correlations between the studied variables for the first and second measurement in the total sample are presented in Table 3. Descriptive statistics of the studied variables in the total sample and the two groups for the two measurements are presented in Table 4. There were no statistically significant differences within groups (first vs. second measurements) and between the two groups in any parameter studied (Table 4). ## 3.2. Effectiveness of the Intervention The results of the repeated measures analysis for PA did not show a significant interaction (F1,82 = 0.94, $p \leq 0.05$, n2 = 0.01) or main effect for time and for group. With respect to the indicators of quality of life, namely satisfaction, depression and self-esteem, no significant interaction (F1,82 = 0.05, $p \leq 0.05$, n2 = 0.001 for satisfaction, F1,82 = 2.78, $p \leq 0.05$, n2 = 0.03 for self-esteem and F1,82 = 0.06, $p \leq 0.05$, n2 = 0.001 for depression) or main effects for time or group were revealed in the analysis. Regarding the Health Action Process Approach variables, namely health status, changes in health, relationship of health with everyday life, intention, activities planning and planning to address difficulties, the results demonstrated no significant interaction for all variables (F1,82 = 0.95, $p \leq 0.05$, n2 = 0.01 for health status, F1,82 = 0.22, $p \leq 0.05$, n2 = 0.03 for changes in health, F1,82 = 0.87, $p \leq 0.05$, n2 = 0.00 for relation of health with everyday life, F1,82 = 0.40, $p \leq 0.05$, n2 = 0.01 for intention, F1,82 = 0.06, $p \leq 0.05$, n2 = 0.001 for activities planning and F1,82 = 0.01, $p \leq 0.05$, n2 = 0.00 for planning to address difficulties). A statistically significant main effect for the groups emerged only for changes in health (F1,82 = 4.67, $p \leq 0.05$, n2 = 0.05) but not for any of the other variables. The analyses also revealed a statistically significant main effect for time only for activity planning (F1,82 = 6.01, $p \leq 0.05$, n2 = 0.07), but not for any of the other variables. ## 4. Discussion This study aimed to examine the effects of a 10-day summer diabetes sports camp on the level of PA and dimensions of HRQoL, in children and adolescents with T1DM. The results of the study indicated that participation in the summer diabetes camp had no impact on the outcome variables. With respect to the effect of the diabetes sport camp on the PA levels of children and adolescents with T1DM, the findings demonstrated that even though the children in the intervention group received training and information on the role of PA in the regulation of the glycemic index and the improvement of their health status, their actual PA levels did not increase significantly. This was the case despite the fact that the children in the intervention group had attended the diabetes sports camp. This finding is consistent with findings from an earlier study [20] suggesting that participation in camps does not promote immediate increases in levels of PA. On the other hand, in the research conducted by Sikora et al., the positive effects on PA were observed three to six months after the end of the camp. This evidence may suggest a delay in the effects of the camp on the behavior of young people. Another plausible explanation may lie in the fact that the post-intervention survey was administered on the last day of the camp, so it could not capture behavior outside the camp. In the present study, no follow-up was conducted, but it will be important for future studies to investigate whether PA levels may increase after a greater amount of time has passed. In addition, there was no discernible change in levels of life satisfaction. Similarly to PA, it is possible that the length of the intervention and the interval between the first and second measurements were responsible for these insignificant results. Furthermore, no effect was observed on young people’s dimensions of HRQoL. This finding contradicts previous findings which showed that participation in such programs led to a significant improvement in children and adolescents’ self-esteem [34,35]. These findings may be attributed to the QoL dimensions measured. To be more specific, low self-esteem and depression are rather stable traits that take a longer period of time to change. This is especially true in situations in which no explicit actions or techniques targeting low self-esteem and/or depression have been encountered, such as in our research, wherein neither of these issues were investigated. We expected that the increase in PA and the socialization among the participants would influence self-esteem and depression. On the other hand, no specific strategies were utilized in order to change either the participants’ levels of depression or self-esteem. Thus, it would appear that tailored techniques may be more effective in trying to generate changes in such stable traits. The incorporation of techniques such as behavioral activation and cognitive re-structuring into the curriculum of the camp might be beneficial to the conduct of subsequent research. With respect to intention to change health-related behavior, the null findings of the present study suggest that it is challenging to achieve improvements in HRQoL by participating in a short-period camp. Again, for intentions, specific actions targeting attitudes, norms and self-efficacy are needed [36]. In this case, persuasion, information, modeling, goal setting, social support, and planning are among the strategies that have been found effective in producing more favorable intentions toward healthy behavior [36]. This study is not free of limitations. To begin with, there were no subsequent follow-ups conducted to attest to the long-term effects of the intervention. Past evidence revealed that people’s attitudes and behaviors would be affected in a delayed manner; hence, future research should include such measurements in order to investigate the long-term effects of diabetes camps. Secondly, the quality of life assessment was carried out based on relatively stable characteristics that take a longer amount of time to change. Future studies examining the effect of short interventions would benefit from measuring more situational dimensions of QoL, such as vitality. Nevertheless, the study provides valuable information that can be used in the development of practices aiming to improve T1DM patients’ HRQoL. To be more specific, in order for diabetes camps to be effective, participants need to either attend the camps for longer periods of time or attend multiple camps throughout the year (for example, camps in the fall, winter, and spring) so that improvements in health-related behaviors may occur and eventually become habits [18]. Barone et al. [ 15] argued that the changes that occur as a result of participation in a camp are not often documented. Nonetheless, life in such camps presents an opportunity to promote PA as a means of treating T1DM [15], develop skills to cope with their disease more effectively, and develop socially and psychologically [16]. In addition, the integration of specific strategies targeting these goals would further increase the effectiveness of diabetes camps in improving T1DM patients’ HRQoL. ## References 1. Mobasseri M., Shirmohammadi M., Amiri T., Vahed N., Hosseini Fard H., Ghojazadeh M.. Prevalence and incidence of type 1 diabetes in the world: A systematic review and meta-analysis. Health Promot. Perspect. (2020) 10 98-115. DOI: 10.34172/hpp.2020.18 2. Melendez-Ramirez L.Y., Richards R.J., Cefalu W.T.. Complications of type 1 diabetes. Endocrinol. Metab. Clin. N. Am. (2010) 39 625-640. DOI: 10.1016/j.ecl.2010.05.009 3. Blanz B.J., Rensch-Riemann B.S., Fritz-Sigmund D.I., Schmidt M.H.. IDDM is a risk factor for adolescent psychiatric disorders. Diabetes Care (1993) 16 1579-1587. DOI: 10.2337/diacare.16.12.1579 4. Kovacs M., Goldston D., Obrosky D.S., Bonar L.K.. Psychiatric disorders in youths with IDDM: Rates and risk factors. Diabetes Care (1997) 20 36-44. DOI: 10.2337/diacare.20.1.36 5. Grey M., Whittemore R., Tamborlane W.. Depression in type 1 diabetes in children. J. Psychosom. Res. (2002) 53 907-911. DOI: 10.1016/S0022-3999(02)00312-4 6. Buchberger B., Huppertz H., Krabbe L., Lux B., Mattivi J.T., Siafarikas A.. Symptoms of depression and anxiety in youth with type 1 diabetes: A systematic review and meta-analysis. Psychoneuroendocrinology (2016) 70 70-84. DOI: 10.1016/j.psyneuen.2016.04.019 7. Samuels S., Menand E.V., Mauer E.A., Hernandez S., Terekhova D., Mathews T.L., Albright M.S., Antal M.D., Kanellopoulos D.. Anxiety and type 1 diabetes management: Guardian and child report in a pediatric endocrinology clinic. Psychosomatics (2020) 61 231-237. DOI: 10.1016/j.psym.2019.11.006 8. Pita R., Grigoriadou E., Marina Ε., Kouvatsou Z., Didaggelos T., Karamitsos D.. Quality of life and type 1 diabetes mellitus (Greek ed.). Greek Diabetol. Ann. (2006) 19 282-294 9. Hoey H., Aanstoot H.-J., Chiarelli F., Daneman D., Danne T., Dorchy H., Fitzgerald M., Garandeau P., Greene S., Holl R.. Good metabolic control is associated with better quality of life in 2,101 adolescents with type 1 diabetes. Diabetes Care (2001) 24 1923-1928. DOI: 10.2337/diacare.24.11.1923 10. Kalyva E., Malakonaki E., Eiser C., Mamoulakis D.. Health-related quality of life (hrqol) of children with type 1 diabetes mellitus (T1DM): Self and parental perceptions. Pediatr. Diabetes (2011) 12 34-40. DOI: 10.1111/j.1399-5448.2010.00653.x 11. Leclair E., de Kerdanet M., Riddell M., Heymen E.. Type 1 diabetes and physical activity in children and adolescents. J. Diabetes Metab. (2013) 1 1163-1173. DOI: 10.4172/2155-6156.s10-004 12. Quirk H., Blake H., Tennyson R., Randell T.L.. Physical activity interventions in children and young people with Type 1 diabetes mellitus: A systematic review with meta-analysis. Diabet. Med. (2014) 31 1163-1173. DOI: 10.1111/dme.12531 13. Robertson K., Adolfsson P., Scheiner G., Hanas R., Riddell M.C.. Exercise in children and adolescents with diabetes. Pediatr. Diabetes (2009) 10 154-168. DOI: 10.1111/j.1399-5448.2009.00567.x 14. Valerio G., Spagnuolo M.I., Lombardi F., Spadaro R., Siano M., Franzese A.. Physical activity and sports participation in children and adolescents with type 1 diabetes mellitus. Nutr. Metab. Cardiovasc. Dis. (2007) 17 376-382. DOI: 10.1016/j.numecd.2005.10.012 15. Barone M.T., Vivolo M.A., Madden P.B.. Are diabetes camps effective?. Diabetes Res. Clin. Pract. (2016) 114 15-22. DOI: 10.1016/j.diabres.2016.01.013 16. Tumini S., Anzellotti M.T., Chiarelli F.. Camps for children with T1DM. Acta Biomed. (2003) 74 32-34. PMID: 12817800 17. Hasan I., Chowdhury A.B., Haque M.I., Patterson C.C.. Changes in glycated hemoglobin, diabetes knowledge, quality of life, and anxiety in children and adolescents with type 1 diabetes attending summer camps: A systematic review and meta-analysis. Pediatr. Diabetes (2020) 22 124-131. DOI: 10.1111/pedi.13147 18. Mancuso M., Caruso-Nicoletti M.. Summer camps and quality of life in children and adolescents with type 1 diabetes. Acta Biomed. (2003) 74 35-37. DOI: 10.4172/2155-6156.S10-004 19. Correale L., Ferraro O.E., Pellino V.C., Ricagno E., Liberali G., Puci M.V., Vandoni M.. Educational physical activity camp in type 1 diabetes: Can be useful to promote an active lifestyle?. Afr. J. Diabetes Med. (2020) 28 1-3. DOI: 10.13140/RG.2.2.10302.54083 20. Sikora M., Zwierzchowska A., Jaworska M., Solich-Talanda M., Mikolajczyk R., Zebrowska A.. The effects of physical activity on glycaemic control in children and adolescents with type 1 diabetes mellitus participating in diabetes camp. Balt. J. Health Phys. Act. (2018) 10 151-161. DOI: 10.29359/BJHPA.10.4.14 21. García-Pérez L., Perestelo-Pérez L., Serrano-Aguilar P., Trujillo-Martín M.. del Effectiveness of a psychoeducative intervention in a summer camp for children with type 1 diabetes mellitus. Diabetes Educ. (2010) 36 310-317. DOI: 10.1177/0145721710361784 22. Santiprabhob J., Kiattisakthavee P., Likitmaskul S., Chaichanwattanakul K., Wekawanich J., Dumrongphol H., Sriwijitkamol A., Peerapatdit T., Nitiyanant W.. Glycemic control, quality of life and self-care behavior among adolescents with type 1 diabetes who attended a diabetes camp. Southeast Asian J. Trop. Med. Public Health (2012) 43 172-184. PMID: 23082568 23. Anarte M.T., Carreira M., Gea A.L., Varela E., Mateo-Anarte M., López Siguero J.P., Gea I.L.. Diabetes summer camp in children and adolescents with type 1 diabetes: Effects on well-being and quality of life. Endocrinol. Diabetes Nutr. (Engl. ed.) (2020) 67 326-332. DOI: 10.1016/j.endien.2019.08.008 24. Faul F., Erdfelder E., Lang A.G., Buchner A.. *GPower 3: A flexible statistical power analysis program for the social, Behavioral, and Biomedical Sciences*. Beh. Res. Methods* (2007) 39 175-191. DOI: 10.3758/BF03193146 25. Faul F., Erdfelder E., Buchner A., Lang A.G.. *Statistical Power analyses using GPower 3.1: Tests for correlation and regression analyses*. Beh. Res. Methods* (2009) 41 1149-1160. DOI: 10.3758/BRM.41.4.1149 26. Kechagia C.. Association of Physical Activity and Psychological Parameters with Quality of life in Patients with Diabetes Mellitus. Master’s Thesis (2019) 27. Godin G., Shephard R.J.. A simple method to assess exercise behavior in the community. Can. J. Appl. Sport Sci. (1985) 10 141-146. PMID: 4053261 28. Diener E., Emmons R.A., Larsen R.J., Griffin S.. The satisfaction with life scale. J. Pers. Assess. (1985) 49 71-75. DOI: 10.1207/s15327752jpa4901_13 29. Pavot W., Diener E., Colvin C.R., Sandvik E.. Further validation of the satisfaction with life scale: Evidence for the cross-method convergence of well-being measures. J. Pers. Assess. (1991) 57 149-161. DOI: 10.1207/s15327752jpa5701_17 30. Rosenberg M.. Society and the Adolescent Self-Image (2016) 31. Radloff L.S.. The CES-D scale. Appl. Psychol. Meas. (1977) 1 385-401. DOI: 10.1177/014662167700100306 32. Beekman A.T., Deeg D.J., Van Limbeek J., Braam A.W., De Vries M.Z., Van Tilburg W.. Criterion validity of the Center for Epidemiologic Studies Depression scale (CES-D): Results from a community-based sample of older subjects in The Netherlands. Psychol. Med. (1997) 27 231-235. DOI: 10.1017/S0033291796003510 33. Schwarzer R.. Health Action Process Approach (HAPA) as a theoretical framework to understand behavior change. Actual. Psicol. (2016) 30 119. DOI: 10.15517/ap.v30i121.23458 34. Bolu S., Danis A., Arslanoglu I., Bolu F., Akcali F., Aytar G.. Assesment of the effect of summer camp on the life quality of diabetic children. J. Turgut. Ozal. Med. Cent. (2017) 24 305-309. DOI: 10.5455/jtomc.2017.04.061 35. Török S., Kökönyei G., Károlyi L., Ittzés A., Tomcsányi T.. Outcome effectiveness of therapeutic recreation camping program for adolescents living with cancer and diabetes. J. Adolesc. Health (2006) 39 445-447. DOI: 10.1016/j.jadohealth.2005.12.018 36. Steinmetz H., Knappstein M., Ajzen I., Schmidt P., Kabst R.. How effective are behavior change interventions based on the theory of planned behavior?. Z. Psychol. (2016) 224 216-233. DOI: 10.1027/2151-2604/a000255
--- title: 'The Mother-Baby Bond: Role of Past and Current Relationships' authors: - Emanuela Bianciardi - Francesca Ongaretto - Alberto De Stefano - Alberto Siracusano - Cinzia Niolu journal: Children year: 2023 pmcid: PMC10046950 doi: 10.3390/children10030421 license: CC BY 4.0 --- # The Mother-Baby Bond: Role of Past and Current Relationships ## Abstract During the perinatal period, up to $25\%$ of women experience difficulties in relating to their child. The mother-child bond promotes the transition to motherhood, protects the woman from depression, and protects the child from the intergenerational transmission of the disease. This study prospectively investigated if the relationship with the co-parent, the attachment style, and the bond that women had with their parents influenced the mother-fetus and then mother-child bond. We also explored the role of depression and anxiety. One hundred nineteen pregnant women were enrolled. We administered clinical interviews and psychometric tools. A telephone interview was conducted at 1, 3, and 6 months of follow-up. Maternal insecure attachment style (r = −0.253, $$p \leq 0.006$$) and women’s dyadic adjustment in the couple’s relationships ($r = 0.182$, $$p \leq 0.049$$) were correlated with lower maternal–fetal attachment. Insecure attachment styles and depression correlate with bottle-feeding rather than breastfeeding. The bond women had with their mothers, not their fathers, was associated with breastfeeding. Depression (OR = 0.243, $$p \leq 0.008$$) and anxiety (OR = 0.185, $$p \leq 0.004$$; OR = 0.304, $p \leq 0.0001$) were related to mother-infant bonding. Close relationships, past and present, affect the bond with the fetus and the child differently. Psychotherapy can provide reassuring and restorative intersubjective experiences. ## 1. Introduction It has been reported that in the perinatal period from 10 to $25\%$ of women have relational problems with the unborn child characterized by feelings of anxiety and hostility towards the newborn up to the neglect and rejection of the child [1]. The mother-child relationship begins before birth and can facilitate the transition to motherhood, which is a critical physical, psychological and social transitional stage in women’s lives [2]. A strong mother-child bond is thought to protect women from depression and children from intergenerational transmission of the disease [3]. In line with the theory of prenatal attachment, “the fetus becomes more human to the woman as the pregnancy progresses, becoming loved both as an extension of itself and as an independent object” with which to form a relationship [4]. The mother-baby bond can be viewed as a bodily, immunological, perceptive, and affective relationship. The maternal physical change of pregnancy initiates the bodily relationship, which continues after delivery through mother’s care and breastfeeding. The immunological relationship consisted of bi-directional communication determined on the one hand by fetal antigen presentation and on the other hand by recognition of and reaction to these antigens by the maternal immune system [5]. After delivery, it is required for the transmission of health and disease [6]. It is perceptible through the mother’s touch, fetal movements, ultrasound examinations, and eye-to-eye contact between mother and infant [7,8]. Finally, it has an impact on maternal emotional, brain, and behavioral changes that begin with conception [9,10]. A good mother-foetus relationship helps women take care of their pregnancy status by adopting healthy behaviors [11]. Moreover, the bond of the pregnant mother toward her unborn infant may be a predictor of an early mother-infant relationship postnatally [12]. According to the theory of the fetal origins of mental health [13], an individual’s mental and physical health status begin during intrauterine life [14]. Accordingly, maternal emotional well-being and the mother-foetus bond may influence neurodevelopmental outcomes in the offspring [15]. The maternal bond with the child laid the groundwork for secure attachment in offspring throughout the lifespan, promoted infant social-emotional development, and facilitated later parenting [16]. It is worth noting that the negative effects of prenatal adversity can be sensitive to the quality of the postnatal environment; in fact, the mother-infant relationship can be protective and reverse the course of brain dysfunction [17], as was clearly demonstrated in both animal and human studies [18]. Moreover, fathers can also protect the mother-child bond both by preventing maternal perinatal depression (PND) and by strengthening the couple relationship [19]. Furthermore, the risk of mental disorders in women’s lives increased during the difficult perinatal period, potentially having a negative impact on the mother-infant bond [20]. The aim of this prospective study was to deepen the understanding of the ongoing mother-baby relationship from pregnancy to the postpartum period. Prospectively, we explored how affective dimensions, such as anxiety and depression, and interpersonal functioning dimensions, such as attachment style (AS), couple relationships, and the bond that expectant mothers had with their parents influenced the relationship first with the fetus and then with the child. Furthermore, we investigated whether these personality and affective factors could account for the type and duration of breastfeeding. ## 2. Materials and Methods The data for this study was collected as part of a larger longitudinal study of perinatal depression and infant development, which was advanced by the University of Rome “Tor Vergata” in Italy and promoted by the non-profit Volunteers Association of Tor Vergata Hospital organization. This data comes from an arm of the study that was conducted with the cooperation of the DSMDP ASL Roma 5 in Rome, Italy. From June 2018 to January 2019, 119 women were enrolled at the Mothers Clinics of the DSMDP ASL Roma 5 (Italy) Department of Gynecology and Obstetrics, which were affiliated with the University of Rome “Tor Vergata”. The inclusion criterion was being over 18 years old. The exclusion criteria were the diagnosis of psychotic disorders and insufficient knowledge of the Italian language. The study was conducted according to the standards of the Declaration of Helsinki and was approved by the Institutional Ethics Review Board of the University of Rome, Tor Vergata. All women signed informed consent. The study included four phases. In the first phase (T0), women were enrolled in the childbirth preparation course. A detailed, structured clinical interview and self-report questionnaires were administered. A telephone interview was conducted one month (T1), three months (T2), and six months (T3) after delivery to collect information about the delivery and the type of breast- or bottle-feeding (BF1st, BF3rd, BF6th). At the T1 session, participants completed the Mother-Infant Bonding Scale (MIBS) questionnaire via telephonic interview. ## 2.1. Structured Clinical Interview The interview was administered by an obstetric nursing student (F.O.) with training relevant to perinatal depression. Questions were chosen based on the perinatal depression literature and included, but were not limited to, sociodemographic data and personal and family history of psychiatric disorders. ## 2.2. Edinburgh Postnatal Depression Scale The Edinburgh Postnatal Depression Scale (EPDS) is a 10-item test that was originally developed to screen for postpartum depression but has since been adopted to screen for depression in pregnancy. Each question is scored from 0 to 3, and the total score ranges from 0 to 30, with higher values indicative of a more severe risk of depression. Scores of 14 or higher during pregnancy and 12 or higher after delivery have been shown to have the highest sensitivity and specificity for detecting depression. We used the validated Italian version of the EPDS [21]. ## 2.3. Parental Bonding Instrument The Parental Bonding Instrument (PBI) was used to measure parental behavior as perceived by the offspring. The instrument consists of 25 items: 12 “care” items and 13 “protection” items. Women are asked to rate their own parental behavior as they recall it from the first 16 years of their lives. There are 25 items each for the father figure and mother figure separately, with recalled child-rearing attitudes evaluated on a four-point (0–3) scale for 12 care items, 7 overprotection items, and 6 control items [22]. To date, there has been no consensus about the factor structure of the PBI. We extracted raw scores for each subscale with respect to the mother and father and analyzed them separately as follows: maternal care, maternal control, maternal overprotection, paternal care, paternal control, and paternal overprotection. Higher scores indicate preferred parenting attitudes in all dimensions. ## 2.4. Relationship Questionnaire Adult attachment style was assessed using the relationship questionnaire (RQ). Women were instructed to answer the questionnaire with reference to all their close relationships with peers (whether romantic or not). The RQ is a single-item measure comprising four short paragraphs, each describing a prototypical attachment pattern concerning close adult relationships. For each of the four descriptions, the respondents indicate how well it describes or relates to themselves on a seven-point rating scale. RQ provides a four-category model of AS based on the four combinations obtained by dichotomizing the subject’s mental representations of the self (self “internal working model” on one axis) and the subject’s image of the other (other “internal working model” on the orthogonal axis) into “positive” and “negative” based on their interpersonal relationships. This yields four attachment patterns: RQ1- secure (positive self, positive other), RQ2- preoccupied (negative self, positive other), RQ3- fearful (negative self, negative other), and RQ4- dismissing avoidant (positive self, negative other) [23,24]. ## 2.5. State-Trait Anxiety Inventory (STAI) The state-trait anxiety inventory (STAI) investigates anxiety state and anxiety trait (forms Y1 and Y2, respectively) by means of 20 questions, scored on a scale from 1 to 4. A score of 0 to 29 indicates no anxiety, a score of 30 to 37 indicates mild anxiety, a score of 38 to 44 indicates moderate anxiety, and a score of > 44 indicates severe anxiety [25]. ## 2.6. Dyadic Adjustment Scale (DAS) The Dyadic Adjustment Scale (DAS) is a 32-item, self-reported dyadic adjustment scale. This scale was designed to detect changes in the partner relationship and includes four scales: consensus (thirteen items), satisfaction (ten items), cohesion (five items), and affective expression (four items). The instrument also gives a total score of dyadic adjustment that ranges from 0 to 151. High total and subscale scores indicate a positive appraisal of the couple’s relationship. The internal consistency of the DAS total score is 0.96 [26]. ## 2.7. Maternal Foetal Attachment Scale (MFAS) The MFAS is a 24-item questionnaire organized into five subscales corresponding to aspects of the relationship between mother and fetus: differentiation of self from the fetus; interaction with the fetus; attributing characteristics and intentions to the fetus; giving of self; and role-taking. Women are rated on a five-point scale (from 1 absolutely no to 5 absolutely yes), and higher scores are associated with higher levels of maternal–fetal attachment [27]. ## 2.8. Mother-to-Infant Bonding Scale (MIBS) The Mother-to-Infant Bonding Scale (MIBS) is an 8-item self-report measure designed to assess a mother’s feelings towards her baby during the early postpartum period. Each item consists of an adjective (loving, resentful, protective, neutral or felt nothing, joyful, dislike, disappointed, aggressive) and is rated on a four-point scale [28]. ## 2.9. Statistical Analyses We used descriptive analysis to study the frequency of dichotomous and continuous variables. Variables were treated either as continuous (MFAS total score, MIBS total score, RQ subscales, PBI subscales, age, STAI Y1-2 total score, EPDS total score, DAS total scores) or binary (couple relationship, employment, EPDS cut-off, RQ secure subscale versus RQ insecure subscales, breastfeeding at T1–BF1st, T2–BF3rd, and at T3–BF6th). We used the student’s t-test to compare the MFAS mean score in EPDS ≥ 12 versus EPDS < 12, and RQ = 1 versus RQ ≥ 1. The ANOVA test was used to compare the MFAS score among the four RQ1/RQ2/RQ3/RQ4 subscales of the RQ. A post hoc Games-Howell comparison and Kruskal-Wallis test were performed to compare the MFAS score to RQ$\frac{1}{2}$/$\frac{3}{4}$ subscales. We used the student’s t-test to analyze the differences between the groups of breastfeeding “no vs yes” at the T0, T1 and T3 times of the study (BF1st, BF3rd, BF6th) and the continuous variables (MFAS total score, Age, DAS total score, EPDS total score, STAI Y-1 total score, STAI Y-2 total score, RQ-$\frac{1}{2}$/$\frac{3}{4}$ subscales total score, PBI MOTHER CARE subscales total score, PBI MOTHER OVER PROTECTION subscales total score, PBI FATHER CARE subscales total score, PBI FATHER OVERPROTECTION subscales total score). Bivariate correlations (Pearson r, two-tailed) were used to explore the association between the MFAS total score and the continuous variables that were listed above. We used an alpha level of 0.05 for significance (two-tailed). Correlation coefficients are considered to represent a small effect from 0.1 to 0.3, a medium effect from 0.3 to 0.5, and a large effect if greater than 0.5 [29]. Multiple linear regression, using the forward stepwise method, was used to test whether MFAS and MIBS could be predicted by the different variables. For dependent variables, standardized coefficients and regression coefficients beta were calculated. The statistical significance level was set a priori at $p \leq 0.05$ and calculations were done with the software IBM SPSS Statistics version 26 for Mac. ## 3. Results The descriptive statistic is shown in Table 1 and Table 2. According to the EPDS score, $13.3\%$ of women ($\frac{16}{119}$) suffered from depression during pregnancy. The prevalence of exclusive breastfeeding at the first month postpartum (BF1st), after three months (BF3th), and at the sixth month (BF6th) after delivery were 41.5, 40.1, and $28.4\%$, respectively. The student’s t-test showed that the MFAS score was not different in women with and without depression according to the EPDS score. The ANOVA test revealed that the MFAS score was lower in the RQ3 group of women with a preoccupied attachment style (Table 3, Figure 1). As reported in Table 4, the correlation analyses (Pearson r two-tailed) demonstrated that the MFAS score was significantly and negatively related to the RQ- RQ3 fearful subscale (r = −0.253, $$p \leq 0.006$$). In addition, the DAS total score ($r = 0.182$, $$p \leq 0.049$$) was significantly and positively related to the MFAS score. Women who were not breastfeeding three months after partum had higher EPDS, lower PBI-mother care, and higher PBI-mother overprotection scores compared to those with exclusive breastfeeding (Table 5 and Table 6). Women who were not breastfeeding six months after partum had higher EPDS scores, lower PBI-mother care scores, and were higher in the RQ3 subscale levels compared to those with exclusive breastfeeding (Table 7). The first linear regression model (Table 8) demonstrated that RQ3 ($$p \leq 0.024$$, adjR2 = 0.037) was related to the MFAS score. The second multiple linear regression model (Table 9) demonstrated that the EPDS (OR = 0.243, $$p \leq 0.008$$) and STAI Y1-2 (OR = 0.185, $$p \leq 0.004$$; OR = 0.304, $p \leq 0.0001$) total scores were related to the MIBS score. ## 4. Discussion We found a high prevalence of depression in pregnant women, as has been extensively documented in the literature [30]. It should be noted that the women who participated in the study were not under psychotherapeutic or pharmacological treatment. Thus, we confirm that a large proportion of women with perinatal depression go undetected and untreated [31]. We discuss our findings in two blocks: those on the mother-fetus bond and those on the mother-child relationship. The maternally insecure attachment style and women’s dyadic adjustment in the couple’s relationships were correlated with lower maternal–fetal attachment. Pregnancy represents a life event in which the couple dynamically adapts to change to achieve a new shared and lasting harmony. Failure to achieve this dyad adjustment is a risk factor for maternal-fetal bonding because the mother may perceive that the forthcoming baby is an obstacle to the couple’s relationship [32,33]. Moreover, difficulties in the couple’s relationship may constitute an additional risk factor, as it has been shown that the partner supports the mental health of the mother and the development of the child [19]. A person’s attachment style is determined by their own and others’ models of expectations, needs, feelings, and behaviors in close relationships. During pregnancy, the woman can face various stressors, including the pregnancy itself, which activates the attachment system [34]. While the adult attachment style is activated in close and bidirectional relationships, prenatal attachment is based on the unidirectional and abstract bond between mother and unborn child that is established during pregnancy [35]. For women with an insecure attachment style, mentally representing the unborn child, whose feelings and actions are less obvious than those of the flesh-and-blood child, can be especially difficult [36]. Depression and anxiety during pregnancy had no effect on the mother-fetus bond, contrary to what we expected [37]. However, the insecure attachment style has emerged as the best explanation for the weaker mother-fetus attachment, as well as the couple relationship [38]. In this study, we explored the mother-infant bond using two measures: a psychometric instrument that was completed by the women one month postpartum, and, in addition, we assessed the proportion of breastfed infants at one, three, and six months of age [39]. Our results demonstrated that depression in pregnancy was associated with bottle-feeding compared with breast-feeding at 1, 3, and 6 months postpartum, which can be explained either as a direct effect of depressive symptoms, such as anhedonia and fatigue, or an indirect effect of depression mediated by the poor mother-infant relationship [40]. In fact, we found that women with depression and anxiety during pregnancy rated their relationship with the baby worse. Our findings confirm that perinatal depression has a negative impact on offspring, both through the depression itself and especially through the mother-infant relationship. This last aspect is fundamental because the newborn’s brain is not mature at the time of birth but develops rapidly through experiences [41] that the child has with the external environment, which is the mother, on whom it totally depends [42]. Interactions with the mother are therefore essential to modulate brain development and influence the child’s cognitive and emotional development. Furthermore, depression and anxiety in pregnancy affect fetal development through biological pathways, causing what has been defined as a “meta-plastic” brain state [13] in which sensitivity to external stimuli is heightened. The potential spectrum of consequences is dangerous when we consider that most mothers with depression go untreated. In our study, however, poor mother-fetal bonding was not associated with decreased mother-infant bonding or breastfeeding, as we would have expected [12]. We got some interesting results. First, we found that mothers with insecure attachment styles preferred bottle-feeding over breastfeeding at 1 month and 6 months postpartum (Table 5 and Table 7). The secure attachment style supports women’s interpersonal functioning, protects them from depression, and is associated with more effective emotional regulation strategies that improve the relationship with the child. Conversely, the insecure attachment style increased the risk of mood disorders, as bonding with the child can be perceived as a source of stress [43]. Furthermore, for the first time, we demonstrated that the relationship that the women in this study had with their parents in the first 15 years of their lives influenced their relationship with the newborn and breastfeeding. In particular, we found that the risk factor was poor maternal care and an overcontrolling attitude, characterized by intrusiveness and a limitation of autonomy. So, only the bond the women in the study had with their mothers, not their fathers, was significant. This figure highlights the primacy of the maternal line in women’s mental health. Moreover, it is important to consider that the literature studies have mainly been based on the attachment theory of John Bowlby [44], which refers to the emotional response of parents towards their children, paying less attention to the effects of overprotection and control by parents. Furthermore, a continuity has been described between having had overcontrolling mothers and an insecure attachment style characterized by low self-esteem and a negative self-model [45]. From this perspective, the experience of a poor relationship with the mother may have determined negative internal models of self and others typical of the fearful attachment style, with the risk of depression and a negative impact on the bond with the child. Our results confirm that the perinatal phase can be conceptualized as a psychopathological continuum across generations. In fact, the intergenerational transmission—from the grandmother to the grandson who has not been directly exposed—and the transgenerational transmission—from the mother to the child—of insecure attachment could affect the children of the women in this study [16]. Finally, future research should look into the predictive value of an insecure attachment style not only in the mother-infant dyad but also in other attachment relationships such as the doctor-patient relationship and, as a result, treatment adherence [46,47]. Although the implications of our study are compelling, we recognize some limitations. To begin with, we could have explored attachment style using clinical interviews compared to self-report questionnaires. However, self-report tests are a reliable and widely accepted tool for research purposes [24]. Furthermore, we could have explored the couple’s dyadic adjustment from the partner’s point of view, considering that fathers protect the woman from depression and mitigate the negative effect of maternal depression on the offspring [48]. Besides these limitations, we highlighted the relevance of our study, which investigated the role of current and past relationships, attachment style, and maternal mental health on the antenatal and postnatal mother-baby bond. Maternal depression is amenable to treatment and thus is a modifiable risk factor to prevent poor child outcomes, although depression is not the only treatment target [49]. The International Guidelines on Women’s Mental Health recommend considering and promoting the mother-child bond [50,51]. As we have emphasized, dimensions of interpersonal functioning, such as the woman’s bonding with the mother, adjustment of the dyadic relationship with the partner, and attachment style, were essential for an adequate mother-infant relationship beyond perinatal depression. The psychotherapeutic path of women in the perinatal period should provide the conditions for accessing a reassuring and reparative intersubjective experience that responds to their needs. Furthermore, it would be appropriate to include the co-parents in the therapeutic project. ## References 1. Fernandez Arias P., Yoshida K., Brockington I.F., Kernreiter J., Klier C.M.. Foetal abuse. Arch. Womens Ment. Health (2019) 22 569-573. DOI: 10.1007/s00737-018-0922-2 2. Maguire J., McCormack C., Mitchell A., Monk C.. Neurobiology of maternal mental illness. Handbook of Clinical Neurology (2020) 3. Kohlhoff J., Barnett B.. Parenting self-efficacy: Links with maternal depression, infant behaviour and adult attachment. Early Hum. Dev. (2013) 89 249-256. DOI: 10.1016/j.earlhumdev.2013.01.008 4. Brandon A.R., Pitts S., Denton W.H., Stringer C.A., Evans H.M.. History of the Theory of Prenatal Attachment. J. Prenat. Perinat. Psychol. Health (2009) 23 201-222. PMID: 21533008 5. Bianciardi E., Barone Y., Lo Serro V., De Stefano A., Giacchetti N., Aceti F., Niolu C.. Inflammatory markers of perinatal depression in women with and without history of trauma. Riv. Psichiatr. (2021) 56 237-245. PMID: 34663990 6. Niarchou M., Zammit S., Lewis G.. The Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort as a resource for studying psychopathology in childhood and adolescence: A summary of findings for depression and psychosis. Soc. Psychiatry Psychiatr. Epidemiol. (2015) 50 1017-1027. DOI: 10.1007/s00127-015-1072-8 7. Kuboshita R., Fujisawa T.X., Makita K., Kasaba R., Okazawa H., Tomoda A.. Intrinsic brain activity associated with eye gaze during mother–child interaction. Sci. Rep. (2020) 10 18903. DOI: 10.1038/s41598-020-76044-y 8. Crowell J.. Editorial: Maternal Bonding: How and Why Does It Matter?. J. Am. Acad. Child Adolesc. Psychiatry (2022) 61 751-753. DOI: 10.1016/j.jaac.2022.03.007 9. Hillerer K.M., Jacobs V.R., Fischer T., Aigner L.. The maternal brain: An organ with peripartal plasticity. Neural Plast. (2014) 2014 574159. DOI: 10.1155/2014/574159 10. Bartels A., Zeki S.. The neural correlates of maternal and romantic love. Neuroimage (2004) 21 1155-1166. DOI: 10.1016/j.neuroimage.2003.11.003 11. Zhang L., Wang L., Yuan Q., Huang C., Cui S., Zhang K., Zhou X.. The mediating role of prenatal depression in adult attachment and maternal-fetal attachment in primigravida in the third trimester. BMC Pregnancy Childbirth (2021) 21. DOI: 10.1186/s12884-021-03779-5 12. Petri E., Palagini L., Bacci O., Borri C., Teristi V., Corezzi C., Faraoni S., Antonelli P., Cargioli C., Banti S.. Maternal–foetal attachment independently predicts the quality of maternal–infant bonding and post-partum psychopathology. J. Matern. Neonatal Med. (2018) 31 3153-3159. DOI: 10.1080/14767058.2017.1365130 13. O’Donnell K.J., Meaney M.J.. Fetal origins of mental health: The developmental origins of health and disease hypothesis. Am. J. Psychiatry (2017) 174 319-328. DOI: 10.1176/appi.ajp.2016.16020138 14. Briana D.D., Malamitsi-Puchner A.. Developmental origins of adult health and disease: The metabolic role of BDNF from early life to adulthood. Metabolism (2018) 81 45-51. DOI: 10.1016/j.metabol.2017.11.019 15. Brummelte S.. Introduction: Early adversity and brain development. Neuroscience (2017) 342 1-3. DOI: 10.1016/j.neuroscience.2016.09.041 16. McMahon C., Trapolini T., Barnett B.. Maternal state of mind regarding attachment predicts persistence of postnatal depression in the preschool years. J. Affect. Disord. (2008) 107 199-203. DOI: 10.1016/j.jad.2007.07.017 17. Branchi I., Cirulli F.. Early experiences: Building up the tools to face the challenges of adult life. Dev. Psychobiol. (2014) 56 1661-1674. DOI: 10.1002/dev.21235 18. Bergman K., Sarkar P., Glover V., O’Connor T.G.. Maternal Prenatal Cortisol and Infant Cognitive Development: Moderation by Infant-Mother Attachment. Biol. Psychiatry (2010) 67 1026-1032. DOI: 10.1016/j.biopsych.2010.01.002 19. Seedat S.. Paternal perinatal mental disorders are inextricably linked to maternal and child morbidity. World Psychiatry (2020) 19 337-338. DOI: 10.1002/wps.20782 20. Alderdice F.. Supporting psychological well-being around the time of birth: What can we learn from maternity care?. World Psychiatry (2020) 19 332-333. DOI: 10.1002/wps.20778 21. Benvenuti P., Ferrara M., Niccolai C., Valoriani V., Cox J.L.. The Edinburgh Postnatal Depression Scale: Validation for an Italian sample. J. Affect. Disord. (1999) 53 137-141. DOI: 10.1016/S0165-0327(98)00102-5 22. Parker G., Tupling H., Brown L.. A parental bonding instrument. Br. J. Med. Psychol. (1979) 52 1-10. DOI: 10.1111/j.2044-8341.1979.tb02487.x 23. Niolu C., Bianciardi E., Di Lorenzo G., Nicolai S., Celi M., Ribolsi M., Pietropolli A., Ticconi C., Tarantino U., Siracusano A.. Insecure attachment style predicts low bone mineral density in postmenopausal women. A pilot study. Riv. Psichiatr (2016) 51 143-148. PMID: 27727264 24. D’Argenio A., Mazzi C., Pecchioli L., Di Lorenzo G., Siracusano A., Troisi A.. Early trauma and adult obesity: Is psychological dysfunction the mediating mechanism?. Physiol. Behav. (2009) 98 543-546. DOI: 10.1016/j.physbeh.2009.08.010 25. Ramanaiah N.V., Franzen M., Schill T.. A Psychometric Study of the State-Trait Anxiety inventory. J. Personal. Assess. (1983) 47 531-535. DOI: 10.1207/s15327752jpa4705_14 26. Spanier G.B.. Measuring Dyadic Adjustment: New Scales for Assessing the Quality of Marriage and Similar Dyads. J. Marriage Fam. (1976) 38 15. DOI: 10.2307/350547 27. Busonera A., Cataudella S., Lampis J., Tommasi M., Zavattini G.C.. Psychometric properties of a 20-item version of the Maternal-Fetal Attachment Scale in a sample of Italian expectant women. Midwifery (2016) 34 79-87. DOI: 10.1016/j.midw.2015.12.012 28. Taylor A., Atkins R., Kumar R., Adams D., Glover V.. A new mother-to-infant bonding scale: Links with early maternal mood. Arch. Womens Ment. Health (2005) 8 45-51. DOI: 10.1007/s00737-005-0074-z 29. Cohen J.. Statistical Power Analysis for the Behavioural Science (1988) 30. Craig M., Howard L.. Postnatal depression. BMJ Clin. Evid. (2009) 2009 1407 31. Khanlari S., Barnett Am B., Ogbo F.A., Eastwood J.. Re-examination of perinatal mental health policy frameworks for women signalling distress on the Edinburgh Postnatal Depression Scale (EPDS) completed during their antenatal booking-in consultation: A call for population health intervention. BMC Pregnancy Childbirth (2019) 19. DOI: 10.1186/s12884-019-2378-4 32. Smythe K.L., Petersen I., Schartau P.. Prevalence of Perinatal Depression and Anxiety in Both Parents. JAMA Netw. Open (2002) 5 e2218969. DOI: 10.1001/jamanetworkopen.2022.18969 33. Fredriksen E., von Soest T., Smith L., Moe V.. Depressive symptom contagion in the transition to parenthood: Interparental processes and the role of partner-related attachment. J. Abnorm. Psychol. (2019) 128 397-403. DOI: 10.1037/abn0000429 34. Robakis T.K., Zhang S., Rasgon N.L., Li T., Wang T., Roth M.C., Humphreys K.L., Gotlib I.H., Ho M., Khechaduri A.. Epigenetic signatures of attachment insecurity and childhood adversity provide evidence for role transition in the pathogenesis of perinatal depression. Transl. Psychiatry (2020) 10 48. DOI: 10.1038/s41398-020-0703-3 35. Zaki N., Ruiz-Ruano García A.M., López Puga J.. Attachment style and prenatal expectations from a Bayesian perspective. Psicothema (2019) 32 138-144 36. Røhder K., Væver M.S., Aarestrup A.K., Jacobsen R.K., Smith-Nielsen J., Schiøtz M.L.. Maternal-fetal bonding among pregnant women at psychosocial risk: The roles of adult attachment style, prenatal parental reflective functioning, and depressive symptoms. PLoS ONE (2020) 15. DOI: 10.1371/journal.pone.0239208 37. Rollè L., Giordano M., Santoniccolo F., Trombetta T.. Prenatal attachment and perinatal depression: A systematic review. Int. J. Environ. Res. Public Health (2020) 17. DOI: 10.3390/ijerph17082644 38. Papapetrou C., Panoulis K., Mourouzis I., Kouzoupis. A.. Pregnancy and the perinatal period: The impact of attachment theory. Psychiatriki. (2020) 31 257-270. DOI: 10.22365/jpsych.2020.313.257 39. Kossakowska K., Bielawska-Batorowicz E.. Postpartum Depressive Symptoms and Their Selected Psychological Predictors in Breast-, Mixed and Formula-Feeding Mothers. Front. Psychiatry (2022) 13 20. DOI: 10.3389/fpsyt.2022.813469 40. Dubber S., Reck C., Müller M., Gawlik S.. Postpartum bonding: The role of perinatal depression, anxiety and maternal–fetal bonding during pregnancy. Arch. Womens Ment. Health (2015) 18 187-195. DOI: 10.1007/s00737-014-0445-4 41. Reissland N., Froggatt S., Reames E., Girkin J.. Effects of maternal anxiety and depression on fetal neuro-development. J. Affect. Disord. (2018) 241 469-474. DOI: 10.1016/j.jad.2018.08.047 42. Webb A.R., Heller H.T., Benson C.B., Lahav A.. Mother’s voice and heartbeat sounds elicit auditory plasticity in the human brain before full gestation. Proc. Natl. Acad. Sci. USA (2015) 112 3152-3157. DOI: 10.1073/pnas.1414924112 43. Bianciardi E., Vito C., Betrò S., De Stefano A., Siracusano A., Niolu C.. The anxious aspects of insecure attachment styles are associated with depression either in pregnancy or in the postpartum period. Ann. Gen. Psychiatry (2020) 19 51. DOI: 10.1186/s12991-020-00301-7 44. Karbowa-Płowens M.. John Bowlby and Attachment Theory. Encyclopedia of Evolutionary Psychological Science (2019) 45. Wilhelm K.. Parental Bonding and Adult Attachment Style: The Relationship between Four Category Models. Int. J. Women’s Heal. Wellness (2016) 2. DOI: 10.23937/2474-1353/1510016 46. Niolu C., Bianciardi E., Di Lorenzo G., Marchetta C., Barone Y., Sterbini N., Ribolsi M., Reggiardo G., Siracusano A.. Enhancing adherence, subjective well-being and quality of life in patients with schizophrenia: Which role for long-acting risperidone?. Ther. Adv. Psychopharmacol. (2015) 5 278-288. DOI: 10.1177/2045125315596897 47. Niolu C., Barone Y., Bianciardi E., Ribolsi M., Marchetta C., Robone C., Ambrosio A., Sarchiola L., Reggiardo G., Di Lorenzo G.. Predictors of poor adherence to treatment in inpatients with bipolar and psychotic spectrum disorders. Riv. Psichiatr (2015) 50 285-294. PMID: 26780202 48. Gentile S., Fusco M.L.. Untreated perinatal paternal depression: Effects on offspring. Psychiatry Res. (2017) 252 325-332. DOI: 10.1016/j.psychres.2017.02.064 49. Jannini T.B., Lorenzo G.D., Bianciardi E., Niolu C., Toscano M., Ciocca G., Jannini E.A., Siracusano A.. Off-label Uses of Selective Serotonin Reuptake Inhibitors (SSRIs). Curr. Neuropharmacol. (2021) 20 693-712. DOI: 10.2174/1570159X19666210517150418 50. Stephenson L.A., Macdonald A.J.D., Seneviratne G., Waites F., Pawlby S.. Mother and Baby Units matter: Improved outcomes for both. BJPsych Open (2018) 4 119-125. DOI: 10.1192/bjo.2018.7 51. Antenatal and postnatal mental health: Clinical management and service guidance. Essent. MIDIRS (2015) 6 14
--- title: Effect of 11 Weeks of Physical Exercise on Physical Fitness and Executive Functions in Children authors: - Mingyang Zhang - Hanna Garnier - Guoping Qian - Shunchang Li journal: Children year: 2023 pmcid: PMC10046957 doi: 10.3390/children10030485 license: CC BY 4.0 --- # Effect of 11 Weeks of Physical Exercise on Physical Fitness and Executive Functions in Children ## Abstract Object: The aim of our study was to evaluate and compare the effects of physical exercise interventions on physical fitness and executive functions in children. Methods: Six-year-old children participated in the study and were randomly divided into physical exercise group (PE group, $$n = 43$$) and control group (C group, $$n = 46$$). The children in the PE group participated in a physical exercise program for 45 min daily, four days a week for 11 weeks. The children in the C group continued with their usual routines. Then, all the children were tested before and after the experiment for body composition (height, weight, BMI), physical fitness (20-m shuttle run test, standing long jump test, grip strength test, 4 × 10 m shuttle run test and sit and reach tests), and executive functions test (animal go/no-go task, working memory span task, simple reaction test and flexible item selection task) before and after the 11-week period. Results: The 11 weeks of physical exercise did not significantly affect the body composition of the children ($p \leq 0.05$). The physical fitness and executive functions test results showed that 11 weeks of physical exercise interventions improves physical fitness (cardiopulmonary fitness, muscle strength, speed sensitivity and flexibility quality) and executive functions parameters (inhibitory control, working memory, the reaction time, and cognitive flexibility) in children ($p \leq 0.05$, $p \leq 0.01$). Conclusion: 11 weeks of physical exercise can improve the physical fitness and executive functions of six-year-old children. ## 1. Introduction Physical fitness (PF) is defined as the body’s condition related to lifestyle that includes strength, speed sensitivity, and cardiopulmonary function [1]. According to a PF survey in China, the trend of continuous decline in children’s PF has not changed in recent years [2]. Moreover, evidence indicates that PF in childhood moderately affects PF in adulthood [3]. A growing body of evidence focuses not only on the PF of healthy children and those with disease, but also on the executive functions (EFs) of these children [4,5]. EFs are represented by a set of higher-order cognition of the prefrontal cortex brain region for goal-directed thought and action [6]. Evidence suggests that EFs have significant effects, including attention, reaction, and spatial memory, on children [7,8]. The development of EFs is associated with the maturation of the prefrontal cortex during childhood [7], and prefrontal cortical thickness mediates the association between cortisol reactivity and EFs in children [9]. Inversely, children exhibit partially compromised EFs, which may be partly explained by the reduced cortical thickness in the prefrontal cortex [10]. Overall, EFs play a pivotal role in children’s brain health and school performance, and the importance of EFs in children begins in early childhood and continues throughout development. Based on the above, there has been a strong call to improve PF and EFs among children and adolescents for global health and social development. Therefore, it is of great importance to develop effective measures to improve children’s PF and EFs. Physical exercise has many physical and mental health benefits, such as helping an individual to maintain a healthy body weight by maintaining energy balance, to enhance the endurance performance of the muscles, and to relieve anxious or depressive states. Neurobiology research has shown that physical exercise, with almost no side effects, can be beneficial to children. Several studies have emphasized the significant contribution of physical exercise to PF. For example, the study performed by Neil et al. [ 11] showed that school-based physical exercise interventions may improve PF in children and adolescents. In addition, physical exercise can directly improve children’s EFs and indirectly enhance EFs mediated by physical fitness, such as limb strength, speed sensitivity, agility, balance, and flexibility [12]. Moreover, the frequency and duration of physical activity, as well as physical exercise intensity are consistently and favorably associated with multiple PF indicators [13]. Accumulating evidence has shown that physical exercise interventions in children not only improves their PF, but also improves brain health, academic performance, and EFs. A meta-analysis reported the positive effect of physical exercise interventions on EFs in children, such as inhibition/interference control, working memory cognitive flexibility, and planning [14,15]. Mehren’s [16] study was the first to show that different exercise intensities (moderate intensity vs. high intensity) have different effects on EFs as measured by functional magnetic resonance imaging (fMRI). Furthermore, children are in a critical period of basic motor development, and there are closed interrelation between motor ability, cognitive function development and prefrontal cortex [17]. It is important to note that EFs in children are malleable and are influenced by the plasticity of the cerebral cortex structure [18]. Such evidence directly highlights the positive effect of physical exercise on EFs. The literature on the intervention of physical exercise on EFs in children is abundant, but current research on the comparative study of pre-test and post-test results as well as intragroup difference is lacking. Therefore, it is necessary to conduct comparative research to explore the effects of physical exercise on EFs in children. In order to explore the effect of 11-week physical exercise interventions on PF and EFs in children, we intervened in the form of 11 weeks of physical exercise on children to observe the effect of improvement on the PF and EFs in six-year-old children. Therefore, for the test of this study, we adopted repeated measurements to observe the improvement of the PF and EFs in children, which were measured once before the start of the training and once at the end (at 11 weeks). Our study hypothesized that physical exercise interventions would improve PF and EFs in children. Based on this, we designed an experiment on the intervention effect of physical exercise on children. ## 2.1. Participants The participants in our study were first grade elementary school students in Beijing, China. We excluded children with medical conditions that affect body composition, PF, and EFs. This study adopted a randomized controlled design. A total of 100 healthy right-handed children (Boys = 50; Girls = 50; six-year-olds) were enrolled in our study before the experiment and randomly divided into a control group (C group, $$n = 49$$) and physical exercise group (PE group, $$n = 50$$). Finally, a total of 89 students completed the whole experiment, and their data were considered effective samples (C group, $$n = 46$$; PE group, $$n = 43$$; Figure 1). We ensured that the sex ratios of the two groups were as equal as possible during participant grouping. All the children were tested before and after the experiment based on four PF indicators and four EFs task parameters. The four PF indicators in this study included cardiopulmonary fitness (20-m shuttle run test), muscle strength (standing long jump test and grip strength test), speed sensitivity (4 × 10 m shuttle run test), and flexibility quality (sit and reach). The four EFs task parameters assessed through EF touch testing and the Psykey psychological system included the animal go/no-go task, working memory span task, simple reaction test, and flexible item selection task. In addition, we obtained informed consent from the children and their parents/guardians before their participation in the study. All experimental procedures and materials were reviewed and approved by the institutional review board of Capital University of Physical Education and Sports. ## 2.2. Physical Exercise Interventions The physical exercise protocol was developed based on previous literatures [19,20]. In this experiment, the protocol was implemented in the school setting, and aimed to increase physical activity in children. All the children in the PE group participated in the physical exercise program four days a week for 11 weeks. Two teachers (one man and one woman) guided the children through the exercises during the experiment. The exercise program consisted of three stages including the warm-up, aerobic exercise, and relaxing activity stages. The warm-up stage lasted for 5 min and included activities, such as running and stretching. Subsequently, the children participated in a 30-min aerobic exercise training, including jumping rope and sport games ($60\%$–$70\%$ of maximum heart rate). Finally, the children participated in a relaxation activity, such as stretching exercises of the upper and lower limb muscles for 5 min (Figure 2). To increase motivation and adherence to the protocol, all exercise sessions were selected based on ease of comprehension, enjoyment, and safety. In addition, all the three stages of the physical exercise intervention program were included in a school-based exercise program with guidance from a physical therapist who works in a hospital to optimize physical health. ## 2.3. Body Composition and Sedentary Behavior In this study, body composition and sedentary behavior were assessed for each child. Height (cm) and weight (kg) were measured using a Holtain stadiometer and digitized weighing scales, respectively, with the participants lightly dressed and barefooted. Body mass index (BMI) was calculated by dividing body weight in kilograms by height squared in meters (kg/m2). ## 2.4. PF Test The detailed procedure for the PF test has been previously described in detail [21,22]. The PF test in this study included cardiopulmonary fitness (20-m shuttle run test), muscle strength (standing long jump and grip strength tests), speed sensitivity (4 × 10 m shuttle run test), and flexibility quality (sit and reach). All the PF tests were conducted by qualified personnel during school hours (9 am–4 pm), and all participants performed routine warm-up exercises before the tests. The 4 × 10 m shuttle run test was always the last test performed. The 20-m shuttle run test was performed to assess cardiopulmonary fitness. The participants stood 0.3 m in front of the starting line. This test required participants to run back and forth between two lines set 20 m apart. The initial running speed was 6.5 km/h, which was increased by 0.5 km/h. Children were encouraged to exhibit their best performance during tests. This test ended when the child demonstrated fatigue. The final test score was recorded as the number of laps completed. Muscle strength was assessed using two different tests: [1] The standing long jump test was used to assess lower limb muscle strength. The jump distance for this test was measured in cm. Children jumped a distance with both feet simultaneously off the ground on an international standard playground. They performed three jumps with 30 s rest intervals between attempts. The best distance among the three jumps was used for the analysis. [ 2] Grip strength test was used to evaluate upper limb muscular strength, which was measured using a handgrip dynamometer (TKK 5001, gripA, Takei, Tokyo). The children stood upright with their shoulders in a neutral position, and arms at their side. Then, the children were instructed to use each hand to squeeze the handle of the dynamometer with maximum effort for 3 s. Each hand was tested three times, alternating the hands between trials, with 30 s rests intervals between measurements on the same hand. Absolute grip strength was calculated as the highest registered value for each hand and is expressed in kilograms (to the nearest 0.1 kg). The 4 × 10 m shuttle run test was used to assess speed sensitivity. The children run back and forth four times along a 10-m track at maximum speed. The children performed this test twice, and the best result (minimum time in seconds) was recorded for the analysis. Flexibility quality was assessed using the sit and reach test. The children sat on the floor with their head, back, and hips in contact with the wall and both legs fully extended. They stretched out their upper limbs as far as possible to touch their toes. The test was performed twice, and the better result was recorded. ## 2.5. EFs Test All the EFs tests were conducted in the classrooms, and the children performed these tests before and after the 11-weeks physical exercise interventions. All the test tasks were performed in sequence according to the procedure. Four main EFs parameters, including animal go/no-go task, working memory span task, simple reaction test, and flexible item selection task, were measured using EF touch testing and Psykey psychological system. The data obtained from the children for the EF tests were considered not valid if the procedure was terminated due to child fatigue/stress or due to experimental error. It is worth noting that all the participants were eligible for the hand grip strength component of this test. Individuals were excluded if they had undergone hand surgery within the previous 3 months prior to the study and if they had any pain, or stiffness in their right hand (e.g., arthritis or tendinitis) in the previous 7 days prior to the study. Details of the EF tests are as follows: The animal go/no-go task is a commonly used test to evaluate inhibitory control [23]. The task involves the random presentation of seven animals above a green button. If the animal is not a pig, the children are required to quickly press the green button. When a pig is presented, no button is pressed. Only one animal was presented at a time for a maximum of 3 s, and the process was 40 times. The correct rate and reaction time were recorded. The working memory span task was employed to analyze working memory as previously described [24]. First, a house was appeared on the screen, and the children were required to note and remember the information in the house, such as the name of the animal in the house and the color of the windows. Subsequently, the experimenter asked the children some information about the house. Next, an empty house appeared on the screen, and the children had to answer some questions about the house. The number of houses increased from one to three during the test. The simple reaction test was used to evaluate the reaction time in our experiment [25]. When the stimulus of a green circle presents on the screen of a computer, the children were required to quickly press a green button as soon as possible. A total of 30 trials were performed with a 2 s interval. The flexible item selection task was performed to assess cognitive flexibility [26]. The children were required to sort the cards by color, size, or category. Two pictures of the same category appeared on the screen, and the children had to indicate the category of the pictures. Next, when a new picture was shown on the screen, and the children had to indicate its category as with the first two images. Finally, three pictures were shown together, and the children chose the two pictures that were similar in a certain dimension. ## 2.6. Statistical Analysis All data were presented as mean ± SD. Statistical analysis was performed by SPSS 20.0 and GraphPad Prism software. In body composition, PF test, the comparison among different groups at the same time was tested by independent-samples T test. The paired sample T test was used to compare before and after the experiment in the same group. In order to examine the effect of the physical exercise intervention, a 2 × 2 repeated measures ANOVA was used with the within factor (before/after intervention) and the between factor EXERCISE (exercise/no exercise). A statistically significant level was defined as $p \leq 0.05.$ ## 3.1. Basic Information of Children The children’s body composition and sedentary behavior details for each group are presented in Table 1. There was no statistically significant difference between the C and PE groups before the physical exercise interventions in body composition indices, such as height, weight, BMI ($p \leq 0.05$). A similar phenomenon was observed after the 11-week period as no significant changes in body composition indices were observed over time in both groups. In addition, no remarkable difference of body composition index in each group with the paired sample t-tests ($p \leq 0.05$). Thus, there was no significant effect of 11 weeks physical exercise on body composition in children. ## 3.2. Effect of 11 Weeks of Physical Exercise on PF in Children Figure 3 shows the different PF parameters for each group. After 11 weeks of physical exercise interventions, the PF parameters including the 20-m shuttle run test, standing long jump test, grip strength test, 4 × 10 m shuttle run test, sit and reach tests were significantly improved in the PE group than that observed before the 11 weeks ($p \leq 0.05$, $p \leq 0.01$). However, there were no significant differences in the C group measured before and after the 11 weeks ($p \leq 0.05$). In addition, pre-test results revealed no significant difference of PF parameters between the control and PE groups before physical exercise interventions, including the 20-m shuttle run test, standing long jump test, grip strength test, 4 × 10 m shuttle run test, sit and reach tests ($p \leq 0.05$). When differences in physical fitness were compared in children using cardiopulmonary fitness, muscle strength, speed sensitivity, flexibility quality, all variables showed a significant intergroup difference ($p \leq 0.05$). These data supported that 11 weeks of physical exercise interventions improve PF parameters in children. ## 3.3. Effect of 11 Weeks of Physical Exercise on EFs on Children Before the physical exercise interventions, variance analysis was performed on each function included in the experiment, and we found that the pre-test accuracy of the inhibitory control, the working memory, the reaction time, and the cognitive flexibility were not significantly different between the two groups, indicating no difference in the EFs level of the children in both groups before the 11-week period ($p \leq 0.05$). To explore the influence of 11 weeks of physical exercise interventions on the EFs of the children, a mixed variance design of 2 × 2 repeated measures ANOVA was used with the within factor (before/after intervention) and the between factor EXERCISE (exercise/no exercise). Physical exercise was used as the inter-subject variable (Accuracy Working memory span: F[1, 87] = 22.09; Accuracy The simple reaction test: F[1, 87] = 49.53; Accuracy Go/No go: F[1, 87] = 15.69; Accuracy Flexible item selection task: F[1, 87] = 103.22. Time was used as the internal variable of the subject (Accuracy Working memory span: F[1, 87] = 22.09; Accuracy The simple reaction test: F[1, 87] = 88.05; Accuracy Go/No go: F[1, 87] = 26.27; Accuracy Flexible item selection task: F[1, 87] = 62.83). Physical exercise x time had an interactive effect (Accuracy Working memory span: F[1, 87] = 25.21; Accuracy The simple reaction test: F[1, 87] = 44.88; Accuracy Go/No go: F[1, 87] = 12.76; Accuracy Flexible item selection task: F[1, 87] = 61.27). We observed the changes in the EFs parameters, which are presented in Figure 4. After 11 weeks of physical exercise interventions, the EFs parameters in the PE group were significantly improved than those before the 11 weeks ($p \leq 0.05$, $p \leq 0.01$), but there were no significant differences in the C group ($p \leq 0.05$). In addition, significant differences were observed in the post-test EFs parameter results between the C and PE groups after physical exercise interventions ($p \leq 0.05$, $p \leq 0.01$). These results indicate that 11 weeks of physical exercise interventions improved EFs parameters in children. ## 4. Discussion The aim of our study was to reveal the effects of physical exercise interventions on the PF and EFs of six-year-old Chinese elementary school children for the duration of 11 weeks. The body composition, PF and EFs parameters were measured before and after the 11-weeks physical exercise interventions or control condition. The present findings support the effectiveness of physical exercise and expanded previous research in several important ways. Moreover, the design of this study allowed us to obtain a more complete picture of the effect of physical exercise on PF and EFs. Physical fitness and cognitive development in children are influenced by various sport and play activities. The three measurement results showed physical exercise improved the PF and EFs of the children. The following results will be discussed. Body composition parameters in childhood can have lifelong consequences on terms overall health, academic performance, and work ability [27]. In this study, we first investigated the children’s body composition. To this end, six-year-old children were classified (PE = 43, $C = 46$) to identify the differences in body composition, and to explore the effect of physical exercise on body composition in children. The results of our study demonstrated no statistically significant difference of body composition parameters between two groups before physical exercise interventions, including height, weight, BMI. Those phenomenon can also be demonstrated after 11 weeks physical exercise interventions. It is widely known that children are at a critical stage of growth and development; their height and weight changes accordingly with the developing of skeletal and muscular systems [28]. However, the body composition parameters in the C and PE groups also showed no statistically significant difference after the 11-week physical exercise interventions. Furthermore, the paired sample t-tests in our experiment also revealed that there was no statistically significant differences in the body composition parameters before and after the 11-week physical exercise interventions in each group. Contrastingly, Alberty et al. demonstrated that 24 months of physical exercise interventions (twice a week, 60 min per session) was associated with a significantly small increase in body composition parameters [29]. We presumed that 11 weeks of physical exercise in our study is not sufficient to observe changes in body composition. Thus, we concluded that there was no significant effect of physical exercise interventions on body composition in children. Next, our study was conducted with the hypothesis that the PE group, compared with the C group, would have more positive effect in PF. Therefore, we examined the potential mechanisms of physical exercise-induced on PF in the children. Childhood is a critical stage for PF development, which is necessary for being physically active and achieving health-related benefits both in the short and long term [30]. Thus, long term exercise interventions inevitably have different degrees of positive influence on PF, such as muscle strength, flexibility, and VO2max [31]. Consistent with this study, we found that the PF parameters, including cardiopulmonary fitness, muscle strength, speed sensitivity, and flexibility quality, of the PE group significantly improved after the physical exercise training. Meanwhile, pre-test results in our study revealed that the baseline PF parameters were not statistically significantly different. This is consistent with our hypothesis, but the PF parameters were remarkably different between the two groups post-test. Moreover, our conclusion was later substantiated by Lee’s study. Consistent with a previous study, Lee’s article pointed out that 16 weeks of physical exercise interventions caused the greatest change in PF, which was demonstrated by the significant increase in PF variables, such as muscular strength, flexibility, muscular endurance, and balance showed a significant increase [32]. The latest research by Ortega et al. further investigated the effects of physical exercise on cardiorespiratory fitness among children who are overweight or obese. Their findings showed that cardiorespiratory fitness is improved among children aged 8 to 11 years who are overweight or obese after 20 weeks of exercise of relatively high intensity exercise for more than 1 h, 3 times per week [33]. These data directly supported our conclusion that 11 weeks of physical exercise interventions in our study improves PF parameters in children. Last, EFs test was performed to examine the dynamic change in EFs parameters. Our study revealed that 11 weeks of physical exercise had significant effects on EFs parameters in children, including their inhibitory control, working memory, the reaction time, and cognitive flexibility. Several previous studies have tried to prove the positive impact of physical exercise interventions on children’s EFs. A meta-analysis demonstrated that a suitable exercise time and intensity can effectively improve the EFs among children and adolescents [34,35]. In addition, physical exercise has been reported to significantly change brain structure, which is critical for cognitive development including EFs [36,37]. Our study further showed that EFs including the inhibitory control, the working memory, the reaction time, and the cognitive flexibility of the PE group were significantly improved compared with that observed in the C group after 11 weeks physical exercise intervention. Similarly, after 11 weeks of physical exercise, the EF in the PE group had also improved. Many studies have shown that physical exercise can effectively improve people’s cognitive function [38,39], which is consistent with the results of this study that physical exercise can improve the EFs of children. Li L et al. provided neurological evidence for the moderating role of the left globus pallidus, which is a crucial structure for complex cognitive processing, in the positive effect of physical exercise on EFs, such as spatial learning and working memory. Increased left globus pallidus activity may be associated with the improvement of EFs due to its relation with the cerebral cortex and the thalamic neuron network [40]. Such evidence collectively indicates that 11 weeks of physical exercise interventions improve EFs parameters in children. The advantage of our study is the comparison the EFs of children through physical exercise training, and the comparison of the changes of EFs after the intervention. However, our study also has some limitations, which need further discussion. First, we only tested four main EFs parameters (inhibitory control, working memory, the reaction time, and cognitive flexibility) in this study, but these parameters cannot completely evaluate the whole content of EFs. Second, we did not discuss whether the influence of the repeated measurements of the EFs test can be eliminated. Last, but most important, is that the effect of physical exercise on the EFs of children with obesity and other diseases needs to be studied further. ## 5. Conclusions The results of our study demonstrated that 11 weeks of physical exercise can improve the PF and EFs of six-year-old children. Thus, physical exercise intervention can be considered as a safe and an economic choice for improving the PF and EFs in children. ## References 1. Hohmann A., Yuan X., Schmitt M., Zhang H., Pietzonka M., Siener M.. Physical Fitness and Motor Competence in Chinese and German Elementary School Children in Relation to Different Physical Activity Settings. Children (2021.0) 8. DOI: 10.3390/children8050391 2. Huang W., Luo J., Chen Y.. Effects of Kindergarten, Family Environment, and Physical Activity on Children’s Physical Fitness. Front. Public Health (2022.0) 10 904903. DOI: 10.3389/fpubh.2022.904903 3. Yip K.M., Wong S.W.S., Chua G.T., So H.-K., Ho F.K., Wong R.S., Tung K.T.S., Chan E.Y.N., Tso W.W.Y., Chow B.-C.. Age-and Sex-Specific Physical Fitness Reference and Association with Body Mass Index in Hong Kong Chinese Schoolchildren. Int. J. Environ. Res. Public Health (2022.0) 19. DOI: 10.3390/ijerph192215346 4. Li Y., Xia X., Meng F., Zhang C.. Association between physical fitness and anxiety in children: A moderated mediation model of agility and resilience. Front. Public Health (2020.0) 8 468. DOI: 10.3389/fpubh.2020.00468 5. Kuiper K.C., Swaab H., Tartaglia N., van Buggenhout G., Wouters C., van Rijn S.. The developmental impact of sex chromosome trisomies on emerging executive functions in young children: Evidence from neurocognitive tests and daily life skills. Genes Brain Behav. (2022.0) 21 e12811. DOI: 10.1111/gbb.12811 6. Kim S.Y., Song J., Wen W., Chen S., Zhang M., Yan J., Lopez B.G., Arredondo M.M., Ip K.I.. Culturally relevant stressors as moderators of intergenerational transmission of mother-adolescent executive function in Mexican immigrant families. Cogn. Res. Princ. Implic. (2021.0) 6 70. DOI: 10.1186/s41235-021-00333-x 7. Ronan L., Alexander-Bloch A., Fletcher P.C.. Childhood Obesity, Cortical Structure, and Executive Function in Healthy Children. Cereb. Cortex. (2020.0) 30 2519-2528. DOI: 10.1093/cercor/bhz257 8. Papastergiou A., Pappas V., Sanoudaki E.. The executive function of bilingual and monolingual children: A technical efficiency approach. Behav. Res. Methods (2022.0) 54 1319-1345. DOI: 10.3758/s13428-021-01658-7 9. Feola B., Dougherty L.R., Riggins T., Bolger D.J.. Prefrontal cortical thickness mediates the association between cortisol reactivity and executive function in childhood. Neuropsychologia (2020.0) 148 107636. DOI: 10.1016/j.neuropsychologia.2020.107636 10. Pop T.L., Maniu D., Rajka D., Lazea C., Cismaru G., Ştef A., Căinap S.S.. Prevalence of underweight, overweight and obesity in school-aged children in the urban area of the northwestern part of Romania. Int. J. Environ. Res. Public Health (2021.0) 18. DOI: 10.3390/ijerph18105176 11. Neil-Sztramko S.E., Caldwell H., Dobbins M.. School-based physical activity programs for promoting physical activity and fitness in children and adolescents aged 6 to 18. Cochrane Database Syst. Rev. (2021.0) 2021 CD007651 12. Wang H., Ge W., Zhu C., Sun Y., Wei S.. How pom cheerleading improves the executive function of preschool children: The mediating role of speed and agility. BMC Psychol. (2022.0) 10. DOI: 10.1186/s40359-022-00944-z 13. Carson V., Lee E.Y., Hewitt L., Jennings C., Hunter S., Kuzik N., Stearns J.A., Unrau S.P., Poitras V.J., Gray C.. Systematic review of the relationships between physical activity and health indicators in the early years (0–4 years). BMC Public Health (2017.0) 17 33-63. PMID: 28056890 14. De Greeff J.W., Bosker R.J., Oosterlaan J., Visscher C., Hartman E.. Effects of physical activity on executive functions, attention and academic performance in preadolescent children: A meta-analysis. J. Sci. Med. Sport (2018.0) 21 501-507. DOI: 10.1016/j.jsams.2017.09.595 15. Verburgh L., Königs M., Scherder E.J.A., Oosterlaan J.. Physical exercise and executive functions in preadolescent children, adolescents and young adults: A meta-analysis. Br. J. Sport Med. (2014.0) 48 973-979. DOI: 10.1136/bjsports-2012-091441 16. Mehren A., Diaz Luque C., Brandes M., Lam A.P., Thiel C.M., Philipsen A., Özyurt J.. Intensity-dependent effects of acute exercise on executive function. Neural Plast. (2019.0) 2019 8608317. DOI: 10.1155/2019/8608317 17. Diamond A.. Close interrelation of motor development and cognitive development and of the cerebellum and prefrontal cortex. Child Dev. (2000.0) 71 44-56. DOI: 10.1111/1467-8624.00117 18. Ji C., Yang J., Lin L., Chen S.. Executive Function Improvement for Children with Autism Spectrum Disorder: A Comparative Study between Virtual Training and Physical Exercise Methods. Children (2022.0) 9. DOI: 10.3390/children9040507 19. Xiong X., Zhu L.N., Dong X.X., Wang W., Yan J., Chen A.G.. Aerobic Exercise Intervention Alters Executive Function and White Matter Integrity in Deaf Children: A Randomized Controlled Study. Neural Plast. (2018.0) 2018 3735208. DOI: 10.1155/2018/3735208 20. Chen A., Zhu L., Jin L., Xiong X., Wang W., Yan J.. Effects of Aerobic Exercise on Executive Control and Its Brain Network in Deaf Children. Sport. Sci. (2016.0) 37 94-101 21. Martinez-Tellez B., Sanchez-Delgado G., Cadenas-Sanchez C., Mora-Gonzalez J., Martín-Matillas M., Löf M., Ortega F.B., Ruiz J.R.. Health-related physical fitness is associated with total and central body fat in preschool children aged 3 to 5 years. Pediatr. Obes. (2016.0) 11 468-474. DOI: 10.1111/ijpo.12088 22. Houwen S., Visscher C., Hartman E., Lemmink K.A.. Test-Retest Reliability of Eurofit Physical Fitness Items for Children with Visual Impairments. Pediatr. Exerc. Sci. (2006.0) 18 300-313. DOI: 10.1123/pes.18.3.300 23. Durston S., Thomas K.M., Yang Y., Uluğ A.M., Zimmerman R.D., Casey B.J.. A neural basis for the development of inhibitory control. Dev. Sci. (2002.0) 5 F9-F16. DOI: 10.1111/1467-7687.00235 24. Kane M.J., Engle R.W.. Working-memory capacity and the control of attention: The contributions of goal neglect, response competition, and task set to Stroop interference. J. Exp. Psychol. Gen. (2003.0) 132 47-70. DOI: 10.1037/0096-3445.132.1.47 25. Wang Y., Zhu H., Chen C., Meng W., Rong X., Liu K.. The Relationship between Body Balance and Cognition of College Students in Different Majors. Proceedings of the 2014 Asia-Pacific Congress on Sports Technology and Engineering (STE 2014) 131-135 26. Jacques S., Zelazo P.D.. The Flexible Item Selection Task (FIST): A measure of executive function in preschoolers. Dev. Neuropsychol. (2001.0) 20 573-591. DOI: 10.1207/S15326942DN2003_2 27. Badiane A., Diouf A., Sylla P.M.D.D., Cissé N.S., Idohou-Dossou N., Dramaix M., Wade S., Donnen P.. Body composition and determinant factors among mother-child pairs (6–8 months) in rural areas of Senegal. Matern. Child Nutr. (2021.0) 17 e13174. DOI: 10.1111/mcn.13174 28. Wang L., Shi C., Zhao Z.. The research and development of growth curve for children’s height and weight on Android platform. Proceedings of the 2017 10th International Congress on Image and Signal Processing 29. Alberty R., Čillík I.. Effect of after-school physical activity on body composition in primary school children: The Slovak “PAD” project. Physiol. Rep. (2023.0) 11 e15540. DOI: 10.14814/phy2.15540 30. Wang G., Zi Y., Li B., Su S., Sun L., Wang F., Ren C., Liu Y.. The Effect of Physical Exercise on Fundamental Movement Skills and Physical Fitness among Preschool Children: Study Protocol for a Cluster-Randomized Controlled Trial. Int. J. Environ. Res. Public Health (2022.0) 19. DOI: 10.3390/ijerph19106331 31. Park S.K., Lee K.S., Heo S.J., Jee Y.S.. Effects of High Intensity Plank Exercise on Physical Fitness and Immunocyte Function in a Middle-Aged Man: A Case Report. Medicina (2021.0) 57. DOI: 10.3390/medicina57080845 32. Lee C.K., Sim Y.K., Lee J.-H., Yook J.S., Ha S.-M., Seo E.C., So W.-Y., Kim H.-R., Jeong W.-M., Goo B.O.. The Relationship between Body Composition and Physical Fitness and the Effect of Exercise According to the Level of Childhood Obesity Using the MGPA Model. Int. J. Environ. Res. Public Health (2022.0) 19. DOI: 10.3390/ijerph19010487 33. Ortega F.B., Mora-Gonzalez J., Cadenas-Sanchez C., Esteban-Cornejo I., Migueles J.H., Solis-Urra P., Verdejo-Román J., Rodriguez-Ayllon M., Molina-Garcia P., Ruiz J.R.. Effects of an Exercise Program on Brain Health Outcomes for Children With Overweight or Obesity: The ActiveBrains Randomized Clinical Trial. JAMA Netw. Open (2022.0) 5 e2227893. DOI: 10.1001/jamanetworkopen.2022.27893 34. Xue Y., Yang Y., Huang T.. Effects of chronic exercise interventions on executive function among children and adolescents: A systematic review with meta-analysis. Brit. J. Sport Med. (2019.0) 53 1397-1404. DOI: 10.1136/bjsports-2018-099825 35. Liu S., Yu Q., Li Z., Cunha P.M., Zhang Y., Kong Z., Lin W., Chen S., Cai Y.. Effects of acute and chronic exercises on executive function in children and adolescents: A systemic review and meta-analysis. Front. Psychol. (2020.0) 11 3482. DOI: 10.3389/fpsyg.2020.554915 36. Bliss E.S., Wong R.H., Howe P.R., Mills D.E.. Benefits of exercise training on cerebrovascular and cognitive function in ageing. J. Cereb. Blood Flow Metab. (2021.0) 41 447-470. DOI: 10.1177/0271678X20957807 37. Stillman C.M., Esteban-Cornejo I., Brown B., Bender C.M., Erickson K.I.. Effects of Exercise on Brain and Cognition Across Age Groups and Health States. Trends Neurosci. (2020.0) 43 533-543. DOI: 10.1016/j.tins.2020.04.010 38. Sanders L.M.J., Hortobágyi T., Karssemeijer E.G.A., Van der Zee E.A., Scherder E.J.A., van Heuvelen M.J.G.. Effects of low- and high-intensity physical exercise on physical and cognitive function in older persons with dementia: A randomized controlled trial. Alzheimers Res. Ther. (2020.0) 12 28. DOI: 10.1186/s13195-020-00597-3 39. Srinivas N.S., Vimalan V., Padmanabhan P., Gulyas B.. An overview on cognitive function enhancement through physical exercises. Brain Sci. (2021.0) 11. DOI: 10.3390/brainsci11101289 40. Li L., Zhang S., Cui J., Chen L.Z., Wang X., Fan M., Wei G.X.. Fitness-Dependent Effect of Acute Aerobic Exercise on Executive Function. Front. Physiol. (2019.0) 10 902. DOI: 10.3389/fphys.2019.00902
--- title: Integrated Analysis of N1-Methyladenosine Methylation Regulators-Related lncRNAs in Hepatocellular Carcinoma authors: - Danjun Song - Xi Wang - Yining Wang - Weiren Liang - Jun Luo - Jiaping Zheng - Kai Zhu journal: Cancers year: 2023 pmcid: PMC10046959 doi: 10.3390/cancers15061800 license: CC BY 4.0 --- # Integrated Analysis of N1-Methyladenosine Methylation Regulators-Related lncRNAs in Hepatocellular Carcinoma ## Abstract ### Simple Summary The relationship between m1A-related lncRNAs and HCC is still unclear. In this study, five m1A-related lncRNAs (AL031985.3, NRAV, WAC-AS1, AC026412.3, and AC099850.4) were identified and used to develop a prognostic signature. The prognostic signature was an independent risk factor related to OS in HCC patients. Synergistic effects on patient survival were observed after combining with TP53 or TMB. In addition, we also screened small molecules which could be potential drugs for HCC patients. Our results suggested that five m1A-related lncRNAs generated a prognostic signature that could be a promising prognostic prediction approach and therapeutic response assessment tool for HCC patients. To the best of our knowledge, this is the first study of m1A-related lncRNAs in HCC. ### Abstract N1-methyladenosine (m1A) and long non-coding RNAs (lncRNAs) play significant roles in tumor progression in hepatocellular carcinoma (HCC). However, their association with HCC is still unclear. In this study, lncRNAs related to m1A were extracted from the mRNA expression matrix in The Cancer Genome Atlas (TCGA) database. Five m1A-related lncRNAs (AL031985.3, NRAV, WAC-AS1, AC026412.3, and AC099850.4) were identified based on lasso Cox regression and they generated a prognostic signature of HCC. The prognostic signature was identified as an independent prognosis factor in HCC patients. Moreover, the prognostic signature achieved better performance than TP53 mutation status or tumor mutational burden (TMB) scores in the stratification of patient survival. The immune landscape indicated that most immune checkpoint genes and immune cells were distributed differently between both risk groups. A higher IC50 of chemotherapeutics (sorafenib, nilotinib, sunitinib, and gefitinib) was observed in the high-risk group, and a lower IC50 of gemcitabine in the low-risk group, suggesting the potential of the prognostic signature in chemosensitivity. In addition, fifty-five potential small molecular drugs were found based on drug sensitivity and NRAV expression. Together, five m1A-related lncRNAs generated a prognostic signature that could be a promising prognostic prediction approach and therapeutic response assessment tool for HCC patients. ## 1. Introduction Hepatocellular carcinoma (HCC) is the most common hepatic cancer, accounting for about $90\%$ of primary liver cancers. It ranks sixth in incidence and fourth in cancer-related deaths worldwide [1]. However, HCC ranks fifth in incidence and second in cancer-related deaths in China [2]. Currently, the mechanism of HCC pathogenesis is unclear. Chronic hepatic B/C virus infection, alcoholic consumption, and non-alcoholic steatohepatitis are common risk factors for HCC development [3,4,5,6]. Although there have been significant advances in surgical and comprehensive treatments, patient prognosis remains unsatisfactory [7,8]. Therefore, it is crucial to explore the detailed carcinogenesis mechanism and identify novel biomarkers for prognosis assessment and individualized treatment. RNA methylation modifications play a significant role in regulating post-transcriptional gene expression [9]. N1-methyladenosine (m1A) methylation in non-coding RNA (tRNA and rRNA) and mRNA has been found to be one of the essential dynamic reversible modification processes [10,11]. Similar to m6A modifications, m1A methylation is regulated by methyltransferases (writers), demethylases (erasers), and binding proteins (readers). m1A methylation is formed by adding a methyl group to the adenosine N1 position by the methyltransferases TRMT6, TRMT61A, TRMT61B, and TRMT10C [12,13], and the removal process is catalyzed by the demethylases ALKBH1 and ALKBH3 [13,14]. In addition, four specific RNA-binding proteins, YTHDF1, YTHDF2, YTHDF3, and YTHDC1, are required to complete the process [13,15]. Moreover, dysregulation of m1A methylation regulators can impact biological processes, leading to abnormal pathological development, such as cell proliferation, impaired self-renewal ability, cell apoptosis and death, and carcinogenesis [11,16,17]. Long non-coding RNAs (lncRNAs) are transcripts of more than 200 nucleotides generally not capable of encoding proteins [18], and their aberrant expression is related to tumorigenesis and progression [18]. Recently, m1A methylation regulators-related lncRNAs (m1A-related lncRNAs) were significantly correlated with cancer progression and patient survival [19,20]. However, the relationship between m1A-related lncRNAs and HCC is still unclear. Our study used The Cancer Genome Atlas (TCGA) database to perform an in-depth analysis of m1A-related lncRNAs in HCC. Additionally, characteristic clinicopathological correlation, survival analysis, predictive model construction, correlation of somatic mutations, and potential drug screening were also conducted. ## 2.1. Data Collection and Processing The flowchart of our study is shown in Figure 1. TCGA database (https://gdc.cancer.gov, accessed on 1 March 2021) was used to download the RNA-seq expression profile (FPKM values), somatic mutation data, and clinicopathological and survival information. In addition, ten m1A methylation regulators were obtained from the published literature [13,21,22], including four methyltransferases (TRMT6, TRMT61A, TRMT61B, and TRMT10C), two demethylases (ALKBH1 and ALKBH3), and four RNA-binding proteins (YTHDF1, YTHDF2, YTHDF3, and YTHDC1). The lncRNA expression matrixes and these 10 m1A methylation regulators were extracted from the mRNA expression profile using the “limma” R package, and further, 126 m1A-related lncRNAs were screened with a criterion of the absolute value of the correlation coefficient > 0.50 and $p \leq 0.001.$ Univariate Cox regression analysis was performed with $p \leq 0.05$ as a cut-off value to investigate the prognosis values of 126 m1A methylation regulators-related lncRNAs. In addition, the Pan-Cancer Atlas Hub (UCSC Xena, http://xena.ucsc.edu, accessed on 1 March 2021) was used to obtain DNA and RNA stemness scores for subsequent analysis. ## 2.2. Unsupervised Clustering Analysis of m1A-Related lncRNAs Based on the extracted m1A-related lncRNAs, unsupervised clustering analysis was performed to classify patients through the “ConsensusClusterPlus” R package. Gene set variation analysis (GSVA) was performed using the “c2.cp.kegg.v7.4.symbols” gene set obtained from the MSigDB database to compare enriched functional differences between the two clusters. In addition, the distribution of 23 immune cells between both clusters was also identified using the R packages “GSVA” and “GSEABase”. The optimal number of clusters at which the magnitude of the cophenetic correlation coefficient starts to decrease is the k value. ## 2.3. m1AScore Construction All patients ($$n = 370$$) were randomly divided into two cohorts, including the training ($$n = 186$$) and testing ($$n = 184$$) cohorts, to develop and validate prognostic risk models. The lasso Cox regression analysis is a method to improve prediction accuracy and interpretability of statistical models and realize variable selection and regularization. Our study used lasso regression to screen the most valuable prognostic predictors in the m1A-related lncRNAs and used it to develop an m1AScore. m1AScore = (βlncRNA1 × expression level of lncRNA1) + (βlncRNA2 × expression level of lncRNA2) + ⋯ + (βlncRNAn × expression level of lncRNAn). The prognostic signature was developed based on the median cut-off of m1AScore, and patients were allocated into high- or low-risk groups. ## 2.4. Predictive Performance of the Prognostic Signature Clinicopathological risk factors related to overall survival (OS) were investigated among the training, testing, and entire cohorts using univariate and multivariate analyses. In addition, subgroup analysis was used to detect the predictive ability of the prognostic signature in patients with different characteristics. A time-dependent receiver operating characteristic (ROC) curve was applied to investigate the predictive ability on 2-year survival using the “timeROC” R package. Gene Set Enrichment Analysis (GSEA) was conducted using the gene set “c2.cp.kegg.v7.4.symbols” to determine significantly enriched functional pathways between the high- and low-risk groups, using the R packages “org.Hs.eg.db” and “clusterProfiler”. The top five enriched pathways in both groups were displayed. ## 2.5. Generating a Nomogram After integrating with clinicopathological parameters and m1AScore, a predictive nomogram was built to assess 1-year, 3-year, and 5-year OS using the “rms” R package. Calibration curves were used to identify the predictive accuracy of the established nomogram. In addition, ROC curve and decision curve analyses were used to compare predictive performance between the different clinical parameters and the prognostic signature. ## 2.6. Correlation between Single-Nucleotide Variants and the Prognostic Signature The distribution of somatic variation between the high- and low-risk groups was investigated, and the top 20 driver genes with the highest mutational frequencies were displayed using the R package “maftool”. Subsequently, tumor mutational burden (TMB) was calculated by counting the total non-synonymous mutations, and the median TMB score was used as a cut-off value to differentiate high- and low-TMB patients. Prognostic differences among the high-TMB + high-risk (H-TMB + H-Risk) group, the high-TMB + low-risk (H-TMB + L-Risk) group, the low-TMB + high-risk (L-TMB + H-Risk) group, and the low-TMB + low-risk (L-TMB + L-Risk) group were analyzed. Furthermore, after combining TP53 mutation status with risk levels, all patients were divided into four classes. Similarly, survival analysis among the TP53-mutation + high-risk (TP53-M + H-Risk), TP53-mutation + low-risk (TP53-M + L-Risk), TP53-wild type + high-risk (TP53-W + H-Risk), and TP53-wild type + low-risk (TP53-W + L-Risk) groups was performed. ## 2.7. Assessment of the Prognostic Signature in Immune Landscapes Correlation analysis was used to investigate the interrelationship between 23 immune cells and m1AScore using Spearman’s method. In addition, the expression patterns of 29 immune checkpoint genes between both risk groups were also compared, according to the previous study [23]. ## 2.8. Therapy Response Assessment of the Prognostic Signature The “pRRophetic” R package was applied to investigate drug sensitivity between the various risk groups based on the half-maximal inhibitory concentration (IC50) [24]. In addition, HCC patients’ immunophenoscores (IPS) were obtained from The Cancer Immunome Database (TCIA, https://tcia.at/home, accessed on 2 March 2021) and compared between the two risk groups. IPS is a representative gene score related to immunogenicity, comprised of four determinants (MHC molecules, immunomodulators, effector cells, and suppressor cells) [25]; higher IPS indicates improved immunogenicity. The correlation between drug sensitivity and identified m1A-related lncRNAs was analyzed using the CellMiner database (version 2021.2, database 2.7) (https://discover.nci.nih.gov/cellminer/, accessed on 2 March 2021) [26] and Pearson’s correlation coefficient analysis. All drugs used for correlation analysis were approved by the Food and Drug Administration or identified by clinical trials. ## 2.9. Specimen Collection Ten pairs of fresh tumor tissues and their adjacent paratumor tissues from ten surgically resected HCC patients were obtained from Zhongshan Hospital, Fudan University, and stored at –80 °C until RNA extraction. Informed consent forms were collected from all patients. The Ethics Committee of the Zhongshan Hospital, Fudan University, approved the protocol of this study. ## 2.10. Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) Total RNA was extracted using Trizol reagent (Invitrogen) and reverse transcribed to cDNA using PrimeScript RT Reagent Kit (Takara). SYBR Premix Ex Taq (Takara) was used to perform qRT-PCR according to the manufacturer’s instructions. Our primers were designed with reference to the previous study [27]. We import the sequence into Primer3 for primer design. The primer sequences used in this study are listed in Supplementary Table S1. GAPDH was the endogenous control. ## 2.11. Statistical Analysis The t-test was used to compare differences among continuous variables, and the Chi-square test or Fisher’s exact test was applied to compare differences in categorical variables. The log-rank test and Cox regression analysis were used to detect the prognostic value of these variables, while PCA was used to identify the outstanding performance of clustering information and prognostic signature. The correlation between the m1AScore and the stemness index of the tissue samples containing DNA methylation-based stemness scores (DNAss) and mRNA expression-based stemness scores (RNAss) was analyzed by the Spearman’s correlation test. All data were analyzed using R (version 4.1.1), and a two-tailed p-value < 0.05 indicated a significant difference. ## 3.1. Identification of Two Clusters Based on the Co-Expressed lncRNAs The expression matrixes of ten m1A methylation regulators and all lncRNAs were obtained from the mRNA profile. On correlation analysis, 126 co-expressed lncRNAs were identified as m1A-related lncRNAs. Forty-three co-expressed lncRNAs were classified as prognosis-related genes using univariate Cox regression analysis (Supplementary Table S2). The network-linked lncRNAs and m1A methylation regulators are displayed in Supplementary Figure S1A. Most co-expressed lncRNAs were related to YTHDC1 expression. The expression patterns of these prognosis-related lncRNAs were significantly different between the normal and tumor tissues (Supplementary Figure S1B). In all tumor samples, two clusters were identified through unsupervised clustering analysis based on the expression matrix of screened lncRNAs (Supplementary Figure S1C). Patients were distinguished visually through PCA analysis (Supplementary Figure S1D). Survival analysis demonstrated that patients in cluster 2 suffered a worse OS (Figure 2A). A heatmap containing clinical parameters, cluster types, and lncRNA expression data was displayed to compare patient characteristics (Figure 2B). There were significant differences in T stage and pathological grade between cluster 1 and cluster 2. GSVA analysis indicated that pathways such as RNA_DEGRADATION, SPLICEOSOME, NUCLEOTIDE_EXCISION_REPAIR, and CELL_CYCLE, were enriched in cluster 2; while OLFACTORY_TRANSDUCTION, FOLATE_BIOSYNTHESIS, ARGININE_AND_PROLINE_METABOLISM, and ARACHIDONIC_ACID_METABOLISM were enriched in cluster 1 (Figure 2C). The immune cell infiltrations were compared between both clusters, and the infiltration levels of activated CD4 T cells and type2 T helper cells were higher in cluster 2 than cluster 1. In contrast, activated CD8 T cells, CD56+ natural killer cells, eosinophils, MDSCs, mast cells, natural killer cells, neutrophils, plasmacytoid dendritic cells, and type1 T helper cells were higher in cluster 1 than cluster 2 (Figure 2D). ## 3.2. Generating an m1AScore for Prognostic Prediction A total of 370 HCC patients were allocated randomly to the training ($$n = 186$$) and testing ($$n = 184$$) cohorts, and no statistical differences were found between them (Supplementary Table S3). Furthermore, based on the forty-three co-expressed m1A-related lncRNAs, five m1A-related lncRNAs were screened as the most valuable predictors for prognostic assessment to generate an m1AScore using lasso Cox regression (Supplementary Figure S1E,F); the correlation coefficients are shown in Table 1. Figure 3A–C shows the distribution of the relative m1A-related lncRNAs expression and m1AScore among the training, testing, and entire cohorts, respectively. Patients were distinguished visually through PCA analysis based on the prognostic signature (Supplementary Figure S1G). The Kaplan–Meier curves showed that patients with low risk displayed superior OS compared with high risk in the training, testing, and entire cohorts (log-rank test: $p \leq 0.001$, $$p \leq 0.003$$, and $p \leq 0.001$, respectively) (Figure 3D–F). The univariate and multivariate regression analysis also indicated that m1AScore was an independent prognostic indicator in both the training and testing cohorts ($p \leq 0.001$) (Supplementary Table S4). In addition, the time-dependent ROC curve analysis showed that the area under curves (AUCs) for 2-year OS were 0.752, 0.716, and 0.718 in the training, testing, and entire cohorts, respectively (Figure 3G–I); thus, suggesting the prediction accuracy of prognostic signature. Furthermore, subgroup analysis demonstrated that the established prognostic signature was a significant risk parameter, and that it can stratify patient prognosis by different clinicopathological characteristics (Supplementary Figure S2A–H). ## 3.3. Function Analysis and Nomogram Construction A Sankey diagram was performed to analyze the interrelationship between the cluster types and the prognostic signature. It showed that most patients in cluster 1 belonged to the low-risk group with favorable outcomes (Supplementary Figure S3A), and the m1AScore in cluster 2 was higher than in cluster 1 (Supplementary Figure S3B). GSEA function analysis revealed that patients in the high-risk group were enriched in pathways such as CELL_CYCLE, while metabolism-related pathways were involved in the low-risk group patients. Thus, these findings supported the GSVA results in the above clustering analysis (Figure 4A,B). A nomogram was established to assess the 1-, 3-, and 5-year OS (Figure 4C), and the calibration curve revealed the nomogram’s ideal consistency for predicting 1-, 3-, and 5-year survival probability (Figure 4D). The ROC and decision curves validated the nomogram’s predictive performance (Figure 4E,F). Overall, these results suggested the predictive accuracy of the generated nomogram. ## 3.4. Survival Stratification Based on the Prognostic Signature and Single-Nucleotide Variant Somatic alterations were found to affect the expression of oncogenes and tumor suppressor genes associated with tumor progression [28]. In this study, mutational frequencies of the top 20 driver genes were compared between the groups, and the results showed that the driver genes’ alteration frequencies in the high-risk group were significantly higher than the low-risk group, especially in TP53 ($43\%$ vs. $14\%$) (Figure 5A,B). Because TP53 mutation is associated with advanced tumor biological features and poor prognosis, high mutation frequencies of TP53 and others could be the potential reasons of dismal prognosis in the high-risk group. More and more studies have suggested that TMB could be an independent indicator for prognosis and immunotherapy [29,30]. Therefore, the combination of TMB and m1AScore could be useful for prognostic stratification and immunotherapeutic response assessment. The correlation between m1AScore and TMB was insignificant (Figure 5C). However, their combination could further stratify patient survival (Figure 5D). The OS of patients with H-TMB + H-Risk or L-TMB + H-Risk group was worse than H-TMB + L-Risk or L-TMB + L-Risk group, suggesting that prognostic signature could differentiate the prognosis in the H-TMB or L-TMB patients. These findings demonstrated synergistic effects of TMB and m1AScore in prognosis prediction. Similarly, we further detect the synergistic effects of m1AScore and TP53 in prognostic assessment. The prognostic signature could stratify outcomes in patients with TP53-M or TP53-W, while TMB scores failed to differentiate patient survival in high- or low-risk patients (Figure 5E), suggesting promising prognostic values of the prognostic signature. ## 3.5. The Immune Landscape of m1AScore To explore the differences in immune landscape between the two groups, we also performed correlation analysis to investigate the interrelationship between the m1AScore and twenty-three immune cells. Activated CD4 T cells, activated dendritic cells, immature dendritic cells, plasmacytoid dendritic cells, regulatory T cells, type17 T helper cells, and type2 T helper cells were found to be positively associated with m1AScore. In contrast, activated CD8 T cells and eosinophils were negatively correlated to m1AScore (Supplementary Figure S3E). Immune checkpoint genes are closely related to the immunotherapy of malignant tumors. In this study, the expression levels of 20 immune checkpoint molecules were significantly higher in the high-risk group than the low-risk group, while FGL1 expression was lower in the high-risk group (Figure 5F). The correlation between m1AScore and DNAss or RNAss was also investigated, and it was found that there were no strong correlations between them (Supplementary Figure S3C,D). ## 3.6. Therapeutic Response Assessment and Drug Sensitivity The IC50 is a marker of response to chemotherapeutic drugs of tumor cells. The sensitivities of sorafenib, lapatinib, nilotinib, sunitinib, and gefitinib were enhanced in the high-risk group compared to the low-risk group (Figure 6A–E). In contrast, the sensitivity of gemcitabine was improved in the low-risk group, suggesting the potential therapeutic value of these drugs in various groups. Moreover, IPS-CTLA(−)-PD1(−), IPS-CTLA(−)-PD1(+), and IPS-CTLA(+)-PD1(−) showed significant differences between both groups (Figure 6F, $p \leq 0.05$), suggesting that the prognostic signature could be a promising predictor for assessing therapeutic drug responses. The RNA expression data and activity of one m1A-related lncRNA, NRAV, were extracted from the CellMiner database. Detailed information of significantly related drugs is listed in Supplementary Table S5. The top eight drugs with the highest correlation between NRAV expression and drug sensitivity are displayed in Figure 6G. ## 3.7. Validation of the Expression Patterns of Five Screened lncRNAs The expression patterns showed that AL031985.3, NRAV, WAC-AS1, AC026412.3, and AC099850.4 were expressed more in the tumor tissues than normal tissues (Figure 7A). PCR also validated similar expression patterns using six tumor samples and paired normal samples (Figure 7B). The Kaplan–Meier curves demonstrated that low expression of AL031985.3, NRAV, WAC-AS1, AC026412.3, and AC099850.4 is associated with better survival in the TCGA database (Figure 7C). These findings indicated the potential of these genes for diagnosis and prognostic assessment. ## 4. Discussion Dynamic RNA methylation modifications are associated with tumor progression and regulators of m1A methylation are involved in cell apoptosis, death, and carcinogenesis [11,16,17]. A recent study demonstrated that m1A regulatory genes may play a critical role in HCC progression and could be used as biomarkers for diagnosis and prognostic assessment [13]. Previous studies indicated that RNA methylation modifications could mediate lncRNA expression [31,32], and lncRNAs affect RNA methylation modification regulators [33,34]. Wang et al. reported that the prognostic signature based on six m6A/m5C/m1A-related lncRNAs were associated with survival, immune microenvironment, TMB, and immunotherapy in head and neck squamous cell carcinoma patients [19]. However, the role of m1A methylation regulators-related lncRNAs in HCC remains unclear. To the best of our knowledge, this is the first study of m1A-related lncRNAs in HCC. In this study, two clusters were identified based on m1A lncRNA expression, and patients in cluster 2 showed worse survival than cluster 1. An m1A-related lncRNA risk model was generated through lasso Cox regression to improve the performance of prognosis prediction. Survival analysis identified that the prognostic signature could discriminate patient outcomes among the training, testing, and entire cohorts. Univariate and multivariate Cox regression analysis demonstrated that m1AScore might be a valuable predictor for HCC patients independent of age, gender, grade, and tumor stage. A nomogram was conducted by integrating clinical parameters and the prognostic signature, and the calibration curves revealed good consistency for 1-, 3-, and 5-year survival probability. These findings thus suggested the reliable performance of established prognostic signatures in HCC patients. The prognostic signature contained five m1A-lncRNAs, including AL031985.3, NRAV, WAC-AS1, AC026412.3, and AC099850.4. However, previous studies reported that the lncRNAs AL031985.3, NRAV, and WAC-AS1 were independent prognostic risk factors in HCC and involved in the functions of pyroptosis, glycolysis, immune function, and ferroptosis [35,36,37,38]. WAC-AS1 regulates ARPP19 to promote glycolysis and tumor proliferation by sponging miR-320d in HCC [36]. These results verified the prognostic values of three lncRNAs, and there are no reports on the roles of lncRNA AC026412.3 and AC099850.4. However, our study could help understand the potential function of these two lncRNAs. HCC development and progression are associated with gene mutations [39]. Survival analysis demonstrated that the H-TMB + H-Risk group showed the worst survival and the L-TMB + L-Risk group the best prognosis, suggesting a synergistic effect after combining TMB and the prognostic signature. Moreover, the prognostic signature could discriminate OS in the H-TMB or L-TMB patients, while TMB status failed to stratify survival in the patients with low-risk; similar results were observed in the combination of TP53 status and the prognostic signature. These findings indicated that the prognostic signature is better than TMB or TP53 status in predicting patient OS. Moreover, combining the prognostic signature and TMB or TP53 status can improve the prognostic values. Exploration of response rates of chemotherapeutic drugs to tumor cells is valuable for drug screening. Drug sensitivity analysis demonstrated that sorafenib, nilotinib, sunitinib, and gefitinib had higher sensitivity in the high-risk group, while gemcitabine displayed enhanced sensitivity in the low-risk group. Sorafenib is one of the first-line therapeutic strategies for advanced HCC, and gemcitabine is an effective chemotherapeutic drug for advanced HCC in clinical practice [40,41]. Nilotinib, an orally available receptor tyrosine kinase inhibitor, can induce autophagy in HCC through AMPK activation [42]. Sunitinib is a multi-targeted receptor tyrosine kinase inhibitor, similar to sorafenib. The PRODIGE 16 study showed that TACE plus sunitinib as first-line therapy was feasible for HCC patients when surgical resection was not suitable [43]. Gefitinib, a selective EGFR tyrosine kinase inhibitor, blocks EGFR activity and has an antitumor effect on HCC development in DEN-exposed rats [44]. The sensitivity of screened lncRNAs to different small molecule drugs was also investigated. Sapitinib was highly correlated with NRAV expression and could induce apoptosis and suppress phospho-EGFR and its downstream pathways [45]. These findings suggested the prospect of targeting NRAV for HCC treatment. Nevertheless, there are several limitations to this study. First, the m1AScore was generated based on the TCGA database, and although the testing cohort was used to verify its values, it is necessary to examine it in an independent database. Second, the expression patterns of five m1A-lncRNAs were identified by qRT-PCR; however, their function should be studied in vivo and in vitro. ## 5. Conclusions In this study, five m1A-related lncRNAs were identified that generated prognostic signatures and can be used as an independent prognostic risk factor for HCC patients. The prognostic signature revealed its potential performance for diagnosis, prognostic prediction, and assessment of therapeutic responses. ## References 1. Villanueva A.. Hepatocellular Carcinoma. N. Engl. J. Med. (2019) 380 1450-1462. DOI: 10.1056/NEJMra1713263 2. He Y., Liang D., Li D., Shan B., Zheng R., Zhang S., Wei W., He J.. Incidence and mortality of laryngeal cancer in China, 2015. Chin. J. Cancer Res. (2020) 32 10-17. DOI: 10.21147/j.issn.1000-9604.2020.01.02 3. Levrero M., Zucman-Rossi J.. Mechanisms of HBV-induced hepatocellular carcinoma. J. Hepatol. (2016) 64 S84-S101. DOI: 10.1016/j.jhep.2016.02.021 4. Marengo A., Rosso C., Bugianesi E.. Liver Cancer: Connections with Obesity, Fatty Liver, and Cirrhosis. Annu. Rev. Med. (2016) 67 103-117. DOI: 10.1146/annurev-med-090514-013832 5. Ganne-Carrié N., Nahon P.. Hepatocellular carcinoma in the setting of alcohol-related liver disease. J. Hepatol. (2019) 70 284-293. DOI: 10.1016/j.jhep.2018.10.008 6. Zoller H., Tilg H.. Nonalcoholic fatty liver disease and hepatocellular carcinoma. Metabolism (2016) 65 1151-1160. DOI: 10.1016/j.metabol.2016.01.010 7. Flattau A., Cristallo J., Duggan M., Gbur M., Fabienne Daguilh M.L., Selwyn P.. Clinical Redeployment of an Academic Family Medicine Department in an Early, Severe COVID-19 Pandemic in the Bronx, NY. J. Am. Board Fam. Med. (2021) 34 466-473. DOI: 10.3122/jabfm.2021.03.200562 8. Pinato D.J., Fessas P., Sapisochin G., Marron T.U.. Perspectives on the Neoadjuvant Use of Immunotherapy in Hepatocellular Carcinoma. Hepatology (2021) 74 483-490. DOI: 10.1002/hep.31697 9. Anwanwan D., Singh S.K., Singh S., Saikam V., Singh R.. Challenges in liver cancer and possible treatment approaches. Biochim. Biophys. Acta Rev. Cancer (2020) 1873 188314. DOI: 10.1016/j.bbcan.2019.188314 10. Dominissini D., Nachtergaele S., Moshitch-Moshkovitz S., Peer E., Kol N., Ben-Haim M.S., Dai Q., Di Segni A., Salmon-Divon M., Clark W.C.. The dynamic N(1)-methyladenosine methylome in eukaryotic messenger RNA. Nature (2016) 530 441-446. DOI: 10.1038/nature16998 11. Zhang C., Jia G.. Reversible RNA Modification N(1)-methyladenosine (m(1)A) in mRNA and tRNA. Genom. Proteom. Bioinform. (2018) 16 155-161. DOI: 10.1016/j.gpb.2018.03.003 12. Safra M., Sas-Chen A., Nir R., Winkler R., Nachshon A., Bar-Yaacov D., Erlacher M., Rossmanith W., Stern-Ginossar N., Schwartz S.. The m1A landscape on cytosolic and mitochondrial mRNA at single-base resolution. Nature (2017) 551 251-255. DOI: 10.1038/nature24456 13. Shi Q., Xue C., Yuan X., He Y., Yu Z.. Gene signatures and prognostic values of m1A-related regulatory genes in hepatocellular carcinoma. Sci. Rep. (2020) 10 15083. DOI: 10.1038/s41598-020-72178-1 14. Liu F., Clark W., Luo G., Wang X., Fu Y., Wei J., Wang X., Hao Z., Dai Q., Zheng G.. ALKBH1-Mediated tRNA Demethylation Regulates Translation. Cell (2016) 167 816-828.e816. DOI: 10.1016/j.cell.2016.09.038 15. Dai X., Wang T., Gonzalez G., Wang Y.. Identification of YTH Domain-Containing Proteins as the Readers for N1-Methyladenosine in RNA. Anal. Chem. (2018) 90 6380-6384. DOI: 10.1021/acs.analchem.8b01703 16. Engel M., Chen A.. The emerging role of mRNA methylation in normal and pathological behavior. Genes Brain Behav (2018) 17 e12428. DOI: 10.1111/gbb.12428 17. Wang Y., Wang J., Li X., Xiong X., Wang J., Zhou Z., Zhu X., Gu Y., Dominissini D., He L.. N(1)-methyladenosine methylation in tRNA drives liver tumourigenesis by regulating cholesterol metabolism. Nat. Commun. (2021) 12 6314. DOI: 10.1038/s41467-021-26718-6 18. Sanchez Calle A., Kawamura Y., Yamamoto Y., Takeshita F., Ochiya T.. Emerging roles of long non-coding RNA in cancer. Cancer Sci. (2018) 109 2093-2100. DOI: 10.1111/cas.13642 19. Wang E., Li Y., Ming R., Wei J., Du P., Zhou P., Zong S., Xiao H.. The Prognostic Value and Immune Landscapes of a m(6)A/m(5)C/m(1)A-Related LncRNAs Signature in Head and Neck Squamous Cell Carcinoma. Front. Cell Dev. Biol. (2021) 9 718974. DOI: 10.3389/fcell.2021.718974 20. Liu J., Chen C., Wang Y., Qian C., Wei J., Xing Y., Bai J.. Comprehensive of N1-Methyladenosine Modifications Patterns and Immunological Characteristics in Ovarian Cancer. Front. Immunol. (2021) 12 746647. DOI: 10.3389/fimmu.2021.746647 21. Zheng Q., Yu X., Zhang Q., He Y., Guo W.. Genetic characteristics and prognostic implications of m1A regulators in pancreatic cancer. Biosci. Rep. (2021) 41 BSR20210337. DOI: 10.1042/BSR20210337 22. Zhao M., Shen S., Xue C.. A Novel m1A-Score Model Correlated With the Immune Microenvironment Predicts Prognosis in Hepatocellular Carcinoma. Front. Immunol. (2022) 13 805967. DOI: 10.3389/fimmu.2022.805967 23. Zhang C., Zhang Z., Zhang G., Zhang Z., Luo Y., Wang F., Wang S., Che Y., Zeng Q., Sun N.. Clinical significance and inflammatory landscapes of a novel recurrence-associated immune signature in early-stage lung adenocarcinoma. Cancer Lett. (2020) 479 31-41. DOI: 10.1016/j.canlet.2020.03.016 24. Geeleher P., Cox N., Huang R.S.. pRRophetic: An R package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PLoS ONE (2014) 9. DOI: 10.1371/journal.pone.0107468 25. Charoentong P., Finotello F., Angelova M., Mayer C., Efremova M., Rieder D., Hackl H., Trajanoski Z.. Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade. Cell Rep. (2017) 18 248-262. DOI: 10.1016/j.celrep.2016.12.019 26. Shankavaram U.T., Varma S., Kane D., Sunshine M., Chary K.K., Reinhold W.C., Pommier Y., Weinstein J.N.. CellMiner: A relational database and query tool for the NCI-60 cancer cell lines. BMC Genom. (2009) 10. DOI: 10.1186/1471-2164-10-277 27. Lee T.L., Xiao A., Rennert O.M.. Identification of novel long noncoding RNA transcripts in male germ cells. Methods Mol. Biol. (2012) 825 105-114. DOI: 10.1007/978-1-61779-436-0_9 28. Huang W., Skanderup A.J., Lee C.G.. Advances in genomic hepatocellular carcinoma research. Gigascience (2018) 7 giy135. DOI: 10.1093/gigascience/giy135 29. Rizzo A., Ricci A.D., Brandi G.. PD-L1, TMB, MSI, and Other Predictors of Response to Immune Checkpoint Inhibitors in Biliary Tract Cancer. Cancers (2021) 13. DOI: 10.3390/cancers13030558 30. Condelli V., Calice G., Cassano A., Basso M., Rodriquenz M.G., Zupa A., Maddalena F., Crispo F., Pietrafesa M., Aieta M.. Novel Epigenetic Eight-Gene Signature Predictive of Poor Prognosis and MSI-Like Phenotype in Human Metastatic Colorectal Carcinomas. Cancers (2021) 13. DOI: 10.3390/cancers13010158 31. Shen J., Feng X.P., Hu R.B., Wang H., Wang Y.L., Qian J.H., Zhou Y.X.. N-methyladenosine reader YTHDF2-mediated long noncoding RNA FENDRR degradation promotes cell proliferation in endometrioid endometrial carcinoma. Lab. Investig. (2021) 101 775-784. DOI: 10.1038/s41374-021-00543-3 32. Liu H., Xu Y., Yao B., Sui T., Lai L., Li Z.. A novel N6-methyladenosine (m6A)-dependent fate decision for the lncRNA THOR. Cell Death Dis. (2020) 11 613. DOI: 10.1038/s41419-020-02833-y 33. Wang Y., Lu J.H., Wu Q.N., Jin Y., Wang D.S., Chen Y.X., Liu J., Luo X.J., Meng Q., Pu H.Y.. LncRNA LINRIS stabilizes IGF2BP2 and promotes the aerobic glycolysis in colorectal cancer. Mol. Cancer (2019) 18 174. DOI: 10.1186/s12943-019-1105-0 34. Wen S., Wei Y., Zen C., Xiong W., Niu Y., Zhao Y.. Long non-coding RNA NEAT1 promotes bone metastasis of prostate cancer through N6-methyladenosine. Mol. Cancer (2020) 19 171. DOI: 10.1186/s12943-020-01293-4 35. Wu Z.H., Li Z.W., Yang D.L., Liu J.. Development and Validation of a Pyroptosis-Related Long Non-coding RNA Signature for Hepatocellular Carcinoma. Front. Cell Dev. Biol. (2021) 9 713925. DOI: 10.3389/fcell.2021.713925 36. Xia X., Zhang H., Xia P., Zhu Y., Liu J., Xu K., Yuan Y.. Identification of Glycolysis-Related lncRNAs and the Novel lncRNA WAC-AS1 Promotes Glycolysis and Tumor Progression in Hepatocellular Carcinoma. Front. Oncol. (2021) 11 733595. DOI: 10.3389/fonc.2021.733595 37. Zhou P., Lu Y., Zhang Y., Wang L.. Construction of an Immune-Related Six-lncRNA Signature to Predict the Outcomes, Immune Cell Infiltration, and Immunotherapy Response in Patients With Hepatocellular Carcinoma. Front. Oncol. (2021) 11 661758. DOI: 10.3389/fonc.2021.661758 38. Chen Z.A., Tian H., Yao D.M., Zhang Y., Feng Z.J., Yang C.J.. Identification of a Ferroptosis-Related Signature Model Including mRNAs and lncRNAs for Predicting Prognosis and Immune Activity in Hepatocellular Carcinoma. Front. Oncol. (2021) 11 738477. DOI: 10.3389/fonc.2021.738477 39. Müller M., Bird T.G., Nault J.C.. The landscape of gene mutations in cirrhosis and hepatocellular carcinoma. J. Hepatol. (2020) 72 990-1002. DOI: 10.1016/j.jhep.2020.01.019 40. Llovet J.M., Kelley R.K., Villanueva A., Singal A.G., Pikarsky E., Roayaie S., Lencioni R., Koike K., Zucman-Rossi J., Finn R.S.. Hepatocellular carcinoma. Nat. Rev. Dis. Prim. (2021) 7 6. DOI: 10.1038/s41572-020-00240-3 41. Zaanan A., Williet N., Hebbar M., Dabakuyo T.S., Fartoux L., Mansourbakht T., Dubreuil O., Rosmorduc O., Cattan S., Bonnetain F.. Gemcitabine plus oxaliplatin in advanced hepatocellular carcinoma: A large multicenter AGEO study. J. Hepatol. (2013) 58 81-88. DOI: 10.1016/j.jhep.2012.09.006 42. Yu H.C., Lin C.S., Tai W.T., Liu C.Y., Shiau C.W., Chen K.F.. Nilotinib induces autophagy in hepatocellular carcinoma through AMPK activation. J. Biol. Chem. (2013) 288 18249-18259. DOI: 10.1074/jbc.M112.446385 43. Turpin A., de Baere T., Heurgué A., Le Malicot K., Ollivier-Hourmand I., Lecomte T., Perrier H., Vergniol J., Sefrioui D., Rinaldi Y.. Liver transarterial chemoembolization and sunitinib for unresectable hepatocellular carcinoma: Results of the PRODIGE 16 study. Clin. Res. Hepatol. Gastroenterol. (2021) 45 101464. DOI: 10.1016/j.clinre.2020.05.012 44. Schiffer E., Housset C., Cacheux W., Wendum D., Desbois-Mouthon C., Rey C., Clergue F., Poupon R., Barbu V., Rosmorduc O.. Gefitinib, an EGFR inhibitor, prevents hepatocellular carcinoma development in the rat liver with cirrhosis. Hepatology (2005) 41 307-314. DOI: 10.1002/hep.20538 45. Scheipl S., Barnard M., Cottone L., Jorgensen M., Drewry D.H., Zuercher W.J., Turlais F., Ye H., Leite A.P., Smith J.A.. EGFR inhibitors identified as a potential treatment for chordoma in a focused compound screen. J. Pathol. (2016) 239 320-334. DOI: 10.1002/path.4729
--- title: The Potential Inhibitory Role of Acetyl-L-Carnitine on Proliferation, Migration, and Gene Expression in HepG2 and HT29 Human Adenocarcinoma Cell Lines authors: - Sarah Albogami journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10046977 doi: 10.3390/cimb45030155 license: CC BY 4.0 --- # The Potential Inhibitory Role of Acetyl-L-Carnitine on Proliferation, Migration, and Gene Expression in HepG2 and HT29 Human Adenocarcinoma Cell Lines ## Abstract Malignancies of the liver and colon are the most prevalent forms of digestive system cancer globally. Chemotherapy, one of the most significant treatments, has severe side effects. Chemoprevention using natural or synthetic medications can potentially reduce cancer severity. Acetyl-L-carnitine (ALC) is an acetylated derivative of carnitine essential for intermediate metabolism in most tissues. This study aimed to investigate the effects of ALC on the proliferation, migration, and gene expression of human liver (HepG2) and colorectal (HT29) adenocarcinoma cell lines. The cell viability and half maximal inhibitory concentration of both cancer cell lines were determined using the 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay. Wound healing after treatment was assessed using a migration assay. Morphological changes were imaged using brightfield and fluorescence microscopy. Post treatment, apoptotic DNA was detected using a DNA fragmentation assay. The relative mRNA expressions of matrix metallopeptidase 9 (MMP9) and vascular endothelial growth factor (VEGF) were evaluated using RT-PCR. The results showed that ALC treatment affects the wound-healing ability of HepG2 and HT29 cell lines. Changes in nuclear morphology were detected under fluorescent microscopy. ALC also downregulates the expression levels of MMP9 and VEGF in HepG2 and HT29 cell lines. Our results indicate that the anticancer action of ALC is likely mediated by a decrease in adhesion, migration, and invasion. ## 1. Introduction Malignancies of the liver and colon are the most prevalent forms of digestive system cancer globally [1]. Liver cancer is the third most common type of cancer in males and eighth most common in females globally. It is the largest cause of mortality worldwide, and more than 800,000 new cases and 700,000 fatalities are reported annually [2,3,4,5]. Risk factors for liver cancer include chronic hepatitis B virus infection, high alcohol use, obesity, diabetes, and smoking [6,7,8]. Globally, colorectal cancer ranks fourth among males and third among females [9]; obesity, a diet lacking fruits and vegetables, a sedentary lifestyle, and smoking are risk factors [10,11,12]. As an imbalanced diet and altered cellular metabolism are important risk factors for the advancement of both liver and colorectal cancers, dietary adjustments are the first-line treatment [13]. A significant foundation for cancer chemoprevention has been the astonishing number of animal studies demonstrating that a range of chemical substances can prevent cancer [14]. In the pursuit of more effective inhibitors, both synthetic and natural substances are being studied [15]. Carnitine is a hydrophilic substance that plays a crucial function in the transport of long-chain fatty acids for beta-oxidation within mitochondria [16,17,18]. Carnitine is a powerful antioxidant and, given that it absorbs active oxygen species in tissues, carnitine has potential anticancer characteristics [19,20]. Carnitine—is hypothesized to cause a boost in cell respiration and apoptosis, along with a decrease in cell proliferation and inflammation in cancer cells through a variety of pathways [21,22,23]. The influence of carnitine on colon tumor progression has been studied in vivo, utilizing two experimental mouse models of colon cancer. One was an azoxymethane treatment model of carcinogen-induced colorectal cancer, whereas the other was a genetically generated model. These in vivo studies revealed that treatment with carnitine substantially elevated levels of carnitine and acylcarnitine in tissues. In azoxymethane-treated animals carnitine reduced the formation of premalignant lesions, while in a genetically induced model carnitine did not demonstrate tumor-protective properties [24]. Using mice, carnitine palmitoyltransferase I and II activity and cachectic cancer expression in the liver were evaluated in relation to L-carnitine supplementation [24]. The mRNA expression level and activity of liver carnitine palmito-yltransferase I and II, as well as serum levels of carnitine and acetyl-carnitine, were significantly lowered, which is associated with substantial elevations in serum concentrations of interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF) [24]. However, several forms of carnitine exist, including L-carnitine (LC), acetyl-L-carnitine (ALC), and propionyl-L-carnitine (PLC) [25]. Among these, ALC has received remarkable attention recently due to its various therapeutic properties. ALC is an acetylated derivative of L-carnitine produced by carnitine acetyltransferase with a high degree of bioavailability [26]. Numerous biological functions of ALC are induced by the metabolic effects of its acetyl and carnitine components, which are essential for a variety of intracellular and metabolic processes, including fatty acid transport into mitochondria, the stability of cell membranes, and a decrease in serum lipid concentrations, and its acetyl group can sustain cetyl-CoA levels [27,28]. ALC participates in the translocation of acetyl units across the mitochondrial membranes in both anabolic and catabolic pathways [29], and furthermore it is a common free radical scavenger and a regulator of energy metabolism and metabolic processes [30]. ALC also exhibits anti-apoptotic and anti-inflammatory properties [31,32,33,34,35] in addition to its stabilizing action on the mitochondrial membrane [36]. The clinical application of ALC has been demonstrated to have excellent outcomes in a range of diseases [22]. It has been proposed as a powerful, inexpensive, and safe alternative treatment for patients with cirrhosis [37] It functions at several levels to cure diabetic polyneuropathy type 1 [38], and may also restore equilibrium in diseases causing neuronal ceroid lipofuscinoses [39]. ALC and alpha-lipoic acid have been shown to enhance mitochondrial energy metabolism and reduce oxidative stress, resulting in enhanced memory in old rats [40,41]. Elmirini et al. [ 2015] demonstrated that ALC might have anticancer effects against colon cancer in vitro [42]; ALC was investigated at a molecular level to see if it functions as a “angiopreventive” substance. Previous in vitro and in vivo studies revealed that ALC inhibits inflammatory angiogenesis by decreasing triggered endothelial cell and macrophage infiltration; on a molecular level, Elmirini et al. demonstrated that ALC inhibits the vascular endothelial growth factor (VEGF), vascular endothelial growth factor receptor 2 (VEGFR2), C-X-C Motif Chemokine Ligand 12 (CXCL12), C-X-C chemokine receptor type 4 (CXCR-4), and focal adhesion kinase pathways. In addition, ALC inhibited the activation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) and intercellular adhesion molecule 1 (ICAM-1) and reduced the adherence of a monocyte cell line to endothelial cells [43]. Functional experiments simulating the pro-tumor development and behavior have shown that ALC inhibits the migration and invasion of four prostate cancer cell lines by reducing cell proliferation. In addition, these experiments revealed that ALC is capable of influencing the crucial functional phases of prostate carcinogenesis, and a number of the implicated molecular mediators were determined [29]. The impact of ALC on two ovarian cancer cell lines (OVCAR-3 and SKOV-3) was evaluated and ALC was found to have no effect on OVCAR-3 cell viability or proliferation and a minor reduction in SKOV-3 cell proliferation [44]. The purpose of this study was to evaluate the possible anticancer outcomes of ALC on human liver cancer cell line HepG2 and colorectal adenocarcinoma cell line HT29, through examining the potential role that ALC could play in preventing cell growth, migration, and gene expression. ## 2.1. Cells Culture and Reagents The human liver cancer cell line HepG2 and colorectal adenocarcinoma cell line HT-29 were obtained from the American Type Culture Collection (ATCC). Cells were cultured in Dulbecco’s modified Eagle medium (DMEM; Gibco-Invitrogen, Carlsbad, CA, USA) accompanied with $10\%$ fetal bovine serum (FBS), $1\%$ L-glutamine, and $1\%$ penicillin-streptomycin. Cells were maintained at 37 °C in a humidified atmosphere containing $5\%$ CO2. ALC was purchased from American International Lab, Inc. (Granada Hills, CA, USA). ## 2.2. Cell Viability Assay All cell lines were individually seeded in a 96-well plate at 1 × 105 cells/well and incubated overnight at 37 °C in a humidified $5\%$ CO2 atmosphere. After 24 h, the medium was replaced with a serum-free medium and incubated for 24 h. Cells were then treated with different dilutions of ALC (0, 0.5, 1,5, 10, 15, 30, and 60 μM) and each incubated for 24, 48, and 72 h. The media were removed after incubation, and the cells were treated with MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Sigma, St. Louis, MO, USA) dissolved at a concentration of 5 mg/mL in PBS. Further, 10 μL of MTT was added to each well, and the cells were incubated at 37 °C for 4 h. The medium was carefully removed, and 100 μL of dimethyl sulfoxide (DMSO; Sigma, St. Louis, MO, USA) was added to each well and mixed. The absorbance of the wells was obtained at 570 nm using a multimode microplate reader (BioTek, Winooski, VT, USA), and the half maximal inhibitory concentration (IC50) was calculated for each cell line. Each test was performed six times in triplicate. ## 2.3. Migration Assay Cells from different cell lines were seeded in a 6-well plate at a density of 5 × 105 cells/well and incubated for 24 h at 37 °C in a humidified $5\%$ CO2 atmosphere. Scratch wounds were created in the center of the monolayers of confluent cells using a sterile 200 μL pipette tip [45]. The culture media were removed, the cells were washed with PBS, and each cell line was treated with ALC at their IC50. Using an inverted microscope (Leica, Wetzlar, Germany), brightfield images were captured with a 10× objective lens and migration into the wound space was documented at 0, 24, 48, and 72 h. ## 2.4. Detection of Apoptotic DNA An Apoptosis DNA Ladder Assay Kit (abcam, Cambridge, UK) was used to detect internucleosomal DNA fragmentation in apoptotic cells according to manufacturing protocol. Cells from different cell lines were seeded in 6-well plates at a density of 10 × 105 cells/well for 24 h at 37 °C in a humidified $5\%$ CO2 environment. After 24 h, the existing media were replaced with serum-free media and incubated for another 24 h. Each cell line was treated with ALC at their IC50 and incubated for 48 h. Untreated cells were used as controls. Cells were trypsinized then pelleted in a 1.5 mL tube, washed with PBS, and pelleted by centrifugation for 5 min at 500× g. The supernatant was carefully removed, and cells were lysed with 35 μL of a lysis buffer with gentle pipetting. Then, 5 μL of Enzyme A solution was added to each sample and mixed by gentle vortexing, and cells were incubated at 37 °C for 10 min. Thereafter, 5 μL of Enzyme B solution was added to each sample and incubated at 50 °C for 30 min, followed by the addition of 5 μL of ammonium acetate solution. Then, 50 μL of isopropanol was added, and the solution was mixed well and placed at −20 °C for 10 min. Samples were centrifuged for 10 min, the supernatant was removed, and the DNA pellet was washed with 0.5 mL $70\%$ ethanol and air dried for 10 min. The DNA pellets were dissolved in 30 μL of suspension buffer, and 20 μL from each sample was electrophoresed on $1.5\%$ agarose gel stained with 0.5 μg/mL ethidium bromide. ## 2.5. Fluorescent Microscopy for Evaluating Apoptosis Cells from different cell lines were seeded in 96-well plates at a density of 0.5 × 105 cells/well for 24 h at 37 °C in a humidified $5\%$ CO2 atmosphere. After 24 h, the media were removed and serum-free media were added to each well and incubated for 24 h. Cells were then treated with ALC at their IC50 and, after 48 h, the media were removed and the cells were washed twice with PBS, fixed with methanol, and left to air dry at 25 °C for 30 min. The cells were then stained with a 500 nM solution of propidium iodide (PI; Invitrogen) in PBS for 5 min and then rinsed several times in PBS. DAPI (4′,6′-diamidino-2-phenylindole; Life Technologies, Carlsbad, CA, USA) was used to stain cells at a concentration of 1 µg/mL in PBS to detect nuclear morphological changes. The cells were incubated for 5 min in the dark and then rinsed several times with PBS. Cell images were captured using an inverted fluorescent microscope (Leica, Wetzlar, Germany) with appropriate filters for PI and DAPI. ImageJ software was used to overlay the images. ## 2.6. RNA Extraction and Reverse Transcription-Real-Time Polymerase Chain Reaction Cells were seeded in 6-well plates at a density of 1 × 106 cells/well and incubated for 24 h at 37 °C in a humidified $5\%$ CO2 atmosphere. After 24 h, the medium was replaced with a serum-free medium and incubated for another 24 h. Each cell line was treated with ALC at the IC50 and incubated for 48 h. TRIzol reagent (Invitrogen) was used to extract total RNA according to the manufacturer’s instructions. The QuantiNova reverse transcription kit (Qiagen, Hilden, Germany) was used to obtain cDNA from extracted RNA according to the manufacturer’s instructions. A 2x PCR master solution (i-Taq, iNtRON Biotechnology, Seoul, Republic of Korea) was used according to the manufacturer’s instructions with the following primers: MMP9 sense: 5′-TTGACAGCGACAAGAAGTGG-3′, antisense: 5′-GCCATTCACGTCGTCCTTAT-3′; VEGF: sense 5′-CCCACTGAGGAGTCCAACAT-3′, antisense: 5′-TTTCTTGCGCTTTCGTTTTT-3′; β-actin: sense 5′-GCTCTTTTCCAGCCTTCCTT-3′, antisense: 5′-GAGCCAGAGCAGTGATCTC-3′. The mRNA expression levels were adjusted to the β-actin expression level and compared with the mRNA expression in the control cells. ## 2.7. Statistical Analysis GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA) was utilized for data analysis. The absolute IC50 was calculated for each cell line at different time points by converting the concentration to a log concentration value, normalizing the results, and presenting them as percentages followed by nonlinear regression (curve fitting). The two-way ANOVA analysis was used to compare the wound closure % obtained from ImageJ for control vs. treated cells. Unpaired t-test analysis was used to compare the percentage of nucleus morphological changes between treated and control cells in each cell line and to compare the expression of each gene in comparison to the control after normalization. Wound closure % was measured using ImageJ version 1.36 (Image J-Fiji, Bethesda, MA, USA) at each time point using Equation [1]:[1]woundclosure%=AreaT0−AreaT(24, or 48 or 72)AreaT0×100 ## 3.1. Efficacy of ALC on the Viability of HepG2 and HT29 Cell Lines HepG2 and HT29 cells were treated with ALC at different concentrations (0, 0.5, 1,5, 10, 15, 30, and 60 μM/mL) to find the optimum IC50. As the concentration of ALC increased, the viability of the cells decreased at the tested time point (Figure 1). The IC50 of ALC was determined for each cell line to select the optimal treatment dose and duration for future investigations. The HepG2 cell line showed IC50 values of 43.12, 40.61, and 45.70 μM/mL after 24, 48, and 72 h, respectively (Figure 1A–C). The HT29 cell line showed IC50 values of 56.42, 54.71, and 56.28 μM/mL after 24, 48, and 72 h, respectively (Figure 1D–F). Based on the results obtained, an IC50 of 40.61 μM/mL after 48 h of treatment for the HepG2 cell line and an IC50 of 54.71 μM/mL after 48 h of treatment for the HT29 cell line were the optimum conditions for the treatment of both cell lines. ## 3.2. ALC Reduced Migration in HepG2 and HT29 Cell Lines The percentage of wound closure was determined after exposing each cell line to ALC at their IC50 at 0, 24, 48, and 72 h. The results in Figure 2A,B showed that treatment with ALC affected the ability of the HepG2 cell line to close the wound after all three tested time points (24, 48, and 72 h) when compared with the control (untreated cells). The HepG2 cell line (Figure 2C) showed a significant reduction when treated with ALC when compared to control cells ($p \leq 0.0001$). The results in Figure 3A,B clearly show the effect of ALC on the capability of the HT29 cells to migrate and close the wound. Similarly, HT29 cells showed a significant reduction in wound healing compared to control cells (Figure 3C) with $p \leq 0.0001.$ ## 3.3. ALC Influences Nucleus Morphology in HepG2 and HT29 Cell Lines and Induced DNA Fragmentation To investigate the effect of ALC on HepG2 and HT29 cell lines, changes in nuclear morphology were observed under both brightfield and fluorescence microscope. Both cell lines showed normal morphology with no treatment (control), while apoptotic cells were detected in both cells treated with ALC (Figure 4). Further changes to the cells include decrease in size, increase in cell density, and the chromatin condenses and migrates to the edges of the nucleus. Both cell lines showed DNA fragmentation after treatment with ALC; based on gel electrophoresis results, it is most likely that HT29 treated cells showed more DNA fragmentation than did HepG2 cells. The fluorescence images obtained after treating HepG2 cells with ALC support these findings, showing that treated cells were more likely to undergo apoptosis than untreated control cells (Figure 5). The findings for the HT29 cell line when exposed to ALC were similar (Figure 6), which may indicate that ALC has a nuclear morphological effect in both cell lines. In fact, the nucleus morphological change significantly increased ($p \leq 0.05$) in both cell lines after treatment with ALC, as shown in Figure 7A,B. ## 3.4. ALC Downregulates the Expression Level of Matrix Metallopeptidase 9 (MMP9) and Vascular Endothelial Growth Factor (VEGF) in HepG2 and HT29 Cell Lines To determine the effect of ALC on HepG2 and HT29 cells, the mRNA expression levels of MMP9 and VEGF were determined. The results obtained in Figure 8A,C show a significant decrease in mRNA expression levels of MMPs and VEGF in both cell lines treated with ALC ($p \leq 0.05$) compared with the control. ## 4. Discussion Although chemotherapy is currently one of the most important techniques used to treat cancer [46], the side effects are a serious disadvantage [47]. Chemoprevention of cancer with either natural or synthetic drugs is a potential method for reducing disease prevalence [48], and consequently, in the past few years, a significant amount of research has been conducted on the preventive and therapeutic abilities of several natural compounds and nutritional supplements against certain types of cancer [49]. ALC is an acetylated derivative of carnitine that plays a crucial function in intermediate metabolism in most tissues [50,51]. The anti-inflammatory and antioxidant effects of ALC have been proven in several studies [35,52,53], as well as its potent anticancer properties [29]. Some studies have demonstrated that ALC may play a significant role in some DNA modifications, such as histone acetylation, which may result in alterations in gene expression [54,55]. In the current study, two cancer cell lines, HepG2 and HT29, were investigated in vitro to determine the therapeutic effect of ALC on their cell viability, migration, morphology, and gene expression. The findings of this study demonstrate the significant anticancer action of ALC on HepG2 and HT29 cancer cell lines, consistent with previous research. In this study, we found that the IC50 of ALC on HepG2 cells was 43.12, 40.61, and 45.70 µM after 24, 48, and 72 h of treatment, respectively, whereas the IC50 of ALC on HT29 cell was 56.42, 54.71, and 56.28 µM after 24, 48, and 72 h of treatment, respectively. A previous in vitro investigation to elucidate the effects of ALC at concentrations of 0, 1, 10, and 100 μM on the proliferation of OVCAR-3 and SKOV-3 ovarian cancer lines using flow cytometry showed that there was a minor, but significant, reduction in the proliferation of the ovarian cancer cell line SKOV-3 when exposed to 10 µM and 100 µM ALC [44]. SW480 human colon cancer cell lines treated with 2 or 3 mM butyrate, with or without carnitine or ALC at a concentration of 5 mM for 48 h, caused a significant increase in the death of SW480 cells [42]. For prostate cancer cell lines PC3, DU145, LNCaP, and 22Rv1, ALC concentrations of 1 and 10 mM were determined to be effective in preventing cellular proliferation [45]. Human umbilical vein endothelial cells (HUVEC) were exposed to 1 or 10 mM ALC for 24 h and the results showed that ALC had a dose-dependent effect on HUVEC proliferation [43]. The effect of ALC on the longevity and proliferation of other human cell lines, including MRC5 and peripheral blood mononuclear cells collected from healthy people, has been previously investigated. At the maximum dose (10 mM), ALC had no influence on the proliferation of healthy cells [43]. Cell migration is an important approach, in which cells must be able to move and reach their correct place within a particular environment in order to carry out their activity [56]. In this study, we found that ALC reduces migration in both cell lines significantly when compared to untreated cells. This is in line with previous research which showed that ALC prevents prostate cancer cell lines from adhering, migrating, and invading [29]. In this study, we examined the cells under a microscope to detect morphological changes that could lead to apoptosis and cell death, and performed a DNA fragmentation assay. The results obtained showed that ALC induced apoptosis when compared to untreated cells. Other studies have evaluated the effect of carnitines on cancer cells in vitro and found that both apoptosis and DNA fragmentation are promoted by carnitine in malignant cells, in line with the present study finding [23,57,58]. Another study evaluated the anti-angiogenic and chemopreventive effects of ALC on four types of prostate cancer cell lines, and found that ALC significantly decreased cell division, promoted apoptosis, and inhibited the synthesis of pro-inflammatory cytokines and chemokines. They also found that ALC reduces cell migration, adhesion, and invasion properties through the downregulation of MMP-9, C-X-C chemokine receptor type 4 (CXCR-4), and the chemokine (C-C motif) ligand 12 (CCL12) pathway, which also induces the inhibition of the angiogenesis pathway via VEGF and C-X-C motif chemokine ligand 8 (CXCL8) [29]. Research in mouse models of cancer cachexia found that the anti-inflammatory cytokines TNFα and IL-6 were significantly inhibited, and the biochemical parameters were enhanced after the administration of oral L-carnitine. They concluded that L-carnitine, in conjunction with the PPAR-γ signaling pathway, exhibits beneficial impacts on cancer cachexia [59]. The impact of L-carnitine on the activities and mRNA expression levels of carnitine palmitoyltransferase (CPT) I and II in the livers of mice with cachectic cancer had been previously evaluated, with findings showing that significant elevations in the expression of CPT I and II were associated with a significant reduction in serum levels of TNF-α and IL-6 [60]. The ability of L-carnitine and ALC with or without the combination of curcumin to prevent cancer development via the 1,2-dimethylhydrazine-stimulated colon tumor mouse model was evaluated for 20 weeks and it was found that both L-carnitine and ALC inhibited the formation of neoplastic lesions as effectively as curcumin alone, if not more so [61]. The combination of palmitoylcarnitine and carnitine promotes oxidative stress and apoptosis in HT29 cells by boosting the efficiency of mitochondrial respiration [57]. Another study found that L-carnitine or palmitoylcarnitine alone could boost caspase-3 and DNA fragmentation, but when administered simultaneously apoptosis was induced [58]. The combination of butyrate and carnitine inhibited the proliferation of human colon cancer Caco-2 cells and induced apoptosis by upregulating proapoptotic proteins (BAX and BAK) and downregulating anti-apoptotic proteins (BCL); however, treatment with carnitine alone did not affect the expression of BAX and BAK, despite the fact that apoptotic effects were observed [23]. The mechanisms behind the antitumor effect of carnitine and ALC on the response of SW480 colon cancer cells to butyrate were investigated, and the combination of butyrate and ALC was found to enhance the death rate of cells. Additionally, carnitine and ALC boosted apoptosis induction, with ALC alone generating a $20\%$ reduction in p21. There was no impact of carnitine or ALC on BCLXL expression. The conclusion was that butyrate and ALC exhibit high antitumor effects and inhibit the viability of colon cancer SW480 cells [42]. In contrast, however, another study reported that L-carnitine administration had no cytotoxic effect on HepG2 cells [62]. Our data showed that, at the molecular level, the incubation of HepG2 and HT29 cells with ALC could reduce MMP expression, with MMPs likely being involved in the invasive properties of cancer cells [63,64]. Our data show similar results to previous research, showing that ALC downregulates the expression of MMP9, which likely suppresses the in vitro hallmarks of tumor progression by inhibiting adhesion, migration, and invasion of four prostate cell lines [29]. In this study, ALC significantly downregulated the expression level of VEGF. VEGF has been shown to perform a crucial function in tumor angiogenesis [64,65,66,67]. The effect of ALC as an “angiopreventive” was examined previously in vivo and in vitro at the molecular level, and it was found that ALC is capable of reducing the expression of several pathways, including VEGF pathways. ALC inhibited the stimulation of NF-B and ICAM-1, hence decreasing the adherence of a monocyte cell line to endothelial cells, leading to the conclusion that ALC has anti-angiogenic and anti-inflammatory characteristics. This could make cancer angioprevention possible [43]. The main limitation of the present study is that multiple diverse colorectal cancer and liver cancer cell lines were not used. However, future research could include the validation of the effect of ALC on additional colorectal cancer and liver cancer cell lines. In conclusion, ALC has antitumor activity against HepG2 and HT29 cell lines. Based on the obtained results, it is most likely that the antitumor effect of ALC is mediated by decreasing adhesion, migration, and invasion. The results of the present study suggest that ALC might be used as a possible chemopreventive supplement for liver and colon cancer. Further in vivo and in vitro investigations are required to evaluate the efficacy of ALC as a chemopreventive drug for clinical use. ## References 1. Herszenyi L., Tulassay Z.. Epidemiology of gastrointestinal and liver tumors. Eur. Rev. Med. Pharmacol. Sci. (2010) 14 249-258. PMID: 20496531 2. Pinheiro P.S., Callahan K.E., Jones P.D., Morris C., Ransdell J.M., Kwon D., Brown C.P., Kobetz E.N.. Liver cancer: A leading cause of cancer death in the United States and the role of the 1945–1965 birth cohort by ethnicity. JHEP Rep. (2019) 1 162-169. DOI: 10.1016/j.jhepr.2019.05.008 3. Alqahtani A., Khan Z., Alloghbi A., Said Ahmed T.S., Ashraf M., Hammouda D.M.. Hepatocellular carcinoma: Molecular mechanisms and targeted therapies. Medicina (2019) 55. DOI: 10.3390/medicina55090526 4. Fitzmaurice C., Dicker D., Pain A., Hamavid H., Moradi-Lakeh M., Maclntyre M., Allen C., Hansen G., Woodbrook R., Wolfe C.. The global burden of cancer 2013 global burden of disease cancer collaboration. JAMA Oncol. (2015) 1 505-527. PMID: 26181261 5. Liu P., Xie S.-H., Hu S., Cheng X., Gao T., Zhang C., Song Z.. Age-specific sex difference in the incidence of hepatocellular carcinoma in the United States. Oncotarget (2017) 8 68131. DOI: 10.18632/oncotarget.19245 6. Carr B.I., Nagalla S., Geva R.. Medical therapy of HCC. Hepatocellular Carcinoma (2016) 489-512 7. McGlynn K.A., London W.T.. The global epidemiology of hepatocellular carcinoma: Present and future. Clin. Liver Dis. (2011) 15 223-243. DOI: 10.1016/j.cld.2011.03.006 8. Smith J.W., Kroker-Lobos M.F., Lazo M., Rivera-Andrade A., Egner P.A., Wedemeyer H., Torres O., Freedman N.D., McGlynn K.A., Guallar E.. Aflatoxin and viral hepatitis exposures in Guatemala: Molecular biomarkers reveal a unique profile of risk factors in a region of high liver cancer incidence. PLoS ONE (2017) 12. DOI: 10.1371/journal.pone.0189255 9. Parkin D.M., Bray F., Ferlay J., Pisani P.. Global cancer statistics, 2002. CA A Cancer J. Clin. (2005) 55 74-108. DOI: 10.3322/canjclin.55.2.74 10. Giovannucci E., Wu K.. Cancer epidemiology and prevention. Cancers of the Colon and Rectum (2006) 11. Botteri E., Iodice S., Bagnardi V., Raimondi S., Lowenfels A.B., Maisonneuve P.. Smoking and colorectal cancer: A meta-analysis. JAMA (2008) 300 2765-2778. DOI: 10.1001/jama.2008.839 12. Giovannucci E.. Modifiable risk factors for colon cancer. Gastroenterol. Clin. (2002) 31 925-943. DOI: 10.1016/S0889-8553(02)00057-2 13. Romero-Gómez M., Zelber-Sagi S., Trenell M.. Treatment of NAFLD with diet, physical activity and exercise. J. Hepatol. (2017) 67 829-846. DOI: 10.1016/j.jhep.2017.05.016 14. Wattenberg L.W.. Inhibition of carcinogenesis by minor dietary constituents. Cancer Res. (1992) 52 2085s-2091s. PMID: 1544145 15. Kostova I.. Synthetic and natural coumarins as cytotoxic agents. Curr. Med. Chem.-Anti-Cancer Agents (2005) 5 29-46. DOI: 10.2174/1568011053352550 16. Longo N., Amat di San Filippo C., Pasquali M.. Disorders of carnitine transport and the carnitine cycle. Am. J. Med. Genet. Part C Semin. Med. Genet. (2006) 142C 77-85. DOI: 10.1002/ajmg.c.30087 17. İzgüt-Uysal V.N., Bülbül M., Tan R., Üstünel İ., Ağar A., Yargiçoğlu P.. Effect of chronic stress and L-carnitine on rat stomach. J. Physiol. Sci. (2007) 57 187-192. DOI: 10.2170/physiolsci.RP004707 18. Gülcin İ.. Antioxidant and antiradical activities of L-carnitine. Life Sci. (2006) 78 803-811. DOI: 10.1016/j.lfs.2005.05.103 19. Peluso G., Nicolai R., Reda E., Benatti P., Barbarisi A., Calvani M.. Cancer and anticancer therapy-induced modifications on metabolism mediated by carnitine system. J. Cell. Physiol. (2000) 182 339-350. DOI: 10.1002/(SICI)1097-4652(200003)182:3<339::AID-JCP4>3.0.CO;2-B 20. Wang M., Wang K., Liao X., Hu H., Chen L., Meng L., Gao W., Li Q.. Carnitine Palmitoyltransferase system: A new target for anti-inflammatory and anticancer therapy?. Front. Pharmacol. (2021) 12 760581. DOI: 10.3389/fphar.2021.760581 21. Radkhouy F., Soltanieh S., Solgi S., Ansari M., Abbasi B.. The effect of L-carnitine on colorectal cancer: A review on current evidence. Food Health (2018) 1 39-47 22. Flanagan J.L., Simmons P.A., Vehige J., Willcox M.D., Garrett Q.. Role of carnitine in disease. Nutr. Metab. (2010) 7 30. DOI: 10.1186/1743-7075-7-30 23. Roy M.-J., Dionne S., Marx G., Qureshi I., Sarma D., Levy E., Seidman E.G.. In vitro studies on the inhibition of colon cancer by butyrate and carnitine. Nutrition (2009) 25 1193-1201. DOI: 10.1016/j.nut.2009.04.008 24. Dionne S., Elimrani I., Roy M.-J., Qureshi I.A., Sarma D.R., Levy E., Seidman E.G.. Studies on the chemopreventive effect of carnitine on tumorigenesis in vivo, using two experimental murine models of colon cancer. Nutr. Cancer (2012) 64 1279-1287. DOI: 10.1080/01635581.2012.722247 25. Mota S.I., Pita I., Águas R., Tagorti S., Virmani A., Pereira F.C., Rego A.C.. Mechanistic perspectives on differential mitochondrial-based neuroprotective effects of several carnitine forms in Alzheimer’s disease in vitro model. Arch. Toxicol. (2021) 95 2769-2784. DOI: 10.1007/s00204-021-03104-1 26. Mazdeh M., Abolfathi P., Sabetghadam M., Mohammadi Y., Mehrpooya M.. Clinical evidence of acetyl-l-carnitine efficacy in the treatment of acute ischemic stroke: A pilot clinical trial. Oxidative Med. Cell. Longev. (2022) 2022 2493053. DOI: 10.1155/2022/2493053 27. Pettegrew J., Levine J., McClure R.. Acetyl-L-carnitine physical-chemical, metabolic, and therapeutic properties: Relevance for its mode of action in Alzheimer’s disease and geriatric depression. Mol. Psychiatry (2000) 5 616-632. DOI: 10.1038/sj.mp.4000805 28. Adeva-Andany M., Calvo-Castro I., Fernández-Fernández C., Donapetry-García C.. Pedre-Piñeiro A. Significance of l-carnitine for human health. IUBMB Life (2017) 69 578-594. DOI: 10.1002/iub.1646 29. Baci D., Bruno A., Cascini C., Gallazzi M., Mortara L., Sessa F., Pelosi G., Albini A., Noonan D.M.. Acetyl-L-Carnitine downregulates invasion (CXCR4/CXCL12, MMP-9) and angiogenesis (VEGF, CXCL8) pathways in prostate cancer cells: Rationale for prevention and interception strategies. J. Exp. Clin. Cancer Res. (2019) 38 464. DOI: 10.1186/s13046-019-1461-z 30. Pan T., Qian Y., Li T., Zhang Z., He Y., Wang J., Li L., Hu Y., Lin M.. Acetyl l-carnitine protects adipose-derived stem cells against serum-starvation: Regulation on the network composed of reactive oxygen species, autophagy, apoptosis and senescence. Cytotechnology (2022) 74 105-121. DOI: 10.1007/s10616-021-00514-y 31. Wang S., Xu J., Zheng J., Zhang X., Shao J., Zhao L., Hao J.. Anti-inflammatory and antioxidant effects of acetyl-L-carnitine on atherosclerotic rats. Med. Sci. Monit. Int. Med. J. Exp. Clin. Res. (2020) 26 e920250-e920251. DOI: 10.12659/MSM.920250 32. Tufekci O., Gunes D., Özoğul C., Kolatan E., Altun Z., Yılmaz O., Aktaş S., Erbayraktar Z., Kırkım G., Mutafoğlu K.. Evaluation of the effect of acetyl L-carnitine on experimental cisplatin nephrotoxicity. Chemotherapy (2009) 55 451-459. DOI: 10.1159/000240020 33. Traina G.. The neurobiology of acetyl-L-carnitine. Front. Biosci.-Landmark (2016) 21 1314-1329. DOI: 10.2741/4459 34. Yang S.P., Yang X.Z., Cao G.P.. Acetyl-l-carnitine prevents homocysteine-induced suppression of Nrf2/Keap1 mediated antioxidation in human lens epithelial cells. Mol. Med. Rep. (2015) 12 1145-1150. DOI: 10.3892/mmr.2015.3490 35. Sepand M.R., Razavi-Azarkhiavi K., Omidi A., Zirak M.R., Sabzevari S., Kazemi A.R., Sabzevari O.. Effect of acetyl-l-carnitine on antioxidant status, lipid peroxidation, and oxidative damage of arsenic in rat. Biol. Trace Elem. Res. (2016) 171 107-115. DOI: 10.1007/s12011-015-0436-y 36. Ames B.N.. Delaying the mitochondrial decay of aging. Ann. N. Y. Acad. Sci. (2004) 1019 406-411. DOI: 10.1196/annals.1297.073 37. Malaguarnera M., Gargante M.P., Cristaldi E., Vacante M., Risino C., Cammalleri L., Pennisi G., Rampello L.. Acetyl-L-carnitine treatment in minimal hepatic encephalopathy. Dig. Dis. Sci. (2008) 53 3018-3025. DOI: 10.1007/s10620-008-0238-6 38. Sima A.A.. Acetyl-L-Carnitine in Diabetic Polyneuropathy. CNS Drugs (2007) 21 13-23. DOI: 10.2165/00023210-200721001-00003 39. Traina G., Bernardi R., Cataldo E., Macchi M., Durante M., Brunelli M.. In the rat brain acetyl-L-carnitine treatment modulates the expression of genes involved in neuronal ceroid lipofuscinosis. Mol. Neurobiol. (2008) 38 146-152. DOI: 10.1007/s12035-008-8038-8 40. Liu J., Head E., Gharib A.M., Yuan W., Ingersoll R.T., Hagen T.M., Cotman C.W., Ames B.N.. Memory loss in old rats is associated with brain mitochondrial decay and RNA/DNA oxidation: Partial reversal by feeding acetyl-L-carnitine and/or R-α-lipoic acid. Proc. Natl. Acad. Sci. USA (2002) 99 2356-2361. DOI: 10.1073/pnas.261709299 41. Hagen T.M., Liu J., Lykkesfeldt J., Wehr C.M., Ingersoll R.T., Vinarsky V., Bartholomew J.C., Ames B.N.. Feeding acetyl-L-carnitine and lipoic acid to old rats significantly improves metabolic function while decreasing oxidative stress. Proc. Natl. Acad. Sci. USA (2002) 99 1870-1875. DOI: 10.1073/pnas.261708898 42. Elimrani I., Dionne S., Saragosti D., Qureshi I., Levy E., Delvin E., Seidman E.G.. Acetylcarnitine potentiates the anticarcinogenic effects of butyrate on SW480 colon cancer cells. Int. J. Oncol. (2015) 47 755-763. DOI: 10.3892/ijo.2015.3029 43. Baci D., Bruno A., Bassani B., Tramacere M., Mortara L., Albini A., Noonan D.M.. Acetyl-l-carnitine is an anti-angiogenic agent targeting the VEGFR2 and CXCR4 pathways. Cancer Lett. (2018) 429 100-116. DOI: 10.1016/j.canlet.2018.04.018 44. Engle D.B., Belisle J.A., Gubbels J.A., Petrie S.E., Hutson P.R., Kushner D.M., Patankar M.S.. Effect of acetyl-l-carnitine on ovarian cancer cells’ proliferation, nerve growth factor receptor (Trk-A and p75) expression, and the cytotoxic potential of paclitaxel and carboplatin. Gynecol. Oncol. (2009) 112 631-636. DOI: 10.1016/j.ygyno.2008.11.020 45. Gerharz M., Baranowsky A., Siebolts U., Eming S., Nischt R., Krieg T., Wickenhauser C.. Morphometric analysis of murine skin wound healing: Standardization of experimental procedures and impact of an advanced multitissue array technique. Wound Repair Regen. (2007) 15 105-112. DOI: 10.1111/j.1524-475X.2006.00191.x 46. Chu E., Sartorelli A.. Cancer chemotherapy. Lange’s Basic and Clinical Pharmacology (2018) 948-976 47. Carr C., Ng J., Wigmore T.. The side effects of chemotherapeutic agents. Curr. Anaesth. Crit. Care (2008) 19 70-79. DOI: 10.1016/j.cacc.2008.01.004 48. Wang J., Jiang Y.-F.. Natural compounds as anticancer agents: Experimental evidence. World J. Exp. Med. (2012) 2 45. DOI: 10.5493/wjem.v2.i3.45 49. Choudhari A.S., Mandave P.C., Deshpande M., Ranjekar P., Prakash O.. Phytochemicals in cancer treatment: From preclinical studies to clinical practice. Front. Pharmacol. (2020) 10 1614. DOI: 10.3389/fphar.2019.01614 50. Momenzadeh M., Aria A., Ghadimi K., Moghaddas A.. Acetyl-L-carnitine for the prevention of taxane-induced neuropathy in patients with breast cancer: A systematic review and meta-analysis. Res. Pharm. Sci. (2023) 18 112. PMID: 36873277 51. Pillich R.T., Scarsella G., Risuleo G.. Reduction of apoptosis through the mitochondrial pathway by the administration of acetyl-L-carnitine to mouse fibroblasts in culture. Exp. Cell Res. (2005) 306 1-8. DOI: 10.1016/j.yexcr.2005.01.019 52. Bodaghi-Namileh V., Sepand M.R., Omidi A., Aghsami M., Seyednejad S.A., Kasirzadeh S., Sabzevari O.. Acetyl-l-carnitine attenuates arsenic-induced liver injury by abrogation of mitochondrial dysfunction, inflammation, and apoptosis in rats. Environ. Toxicol. Pharmacol. (2018) 58 11-20. DOI: 10.1016/j.etap.2017.12.005 53. Zhang Z., Zhao M., Li Q., Zhao H., Wang J., Li Y.. Acetyl-l-carnitine inhibits TNF-α-induced insulin resistance via AMPK pathway in rat skeletal muscle cells. FEBS Lett. (2009) 583 470-474. DOI: 10.1016/j.febslet.2008.12.053 54. Pisano C., Pratesi G., Laccabue D., Zunino F., Lo Giudice P., Bellucci A., Pacifici L., Camerini B., Vesci L., Castorina M.. Paclitaxel and Cisplatin-induced neurotoxicity: A protective role of acetyl-L-carnitine. Clin. Cancer Res. (2003) 9 5756-5767. PMID: 14654561 55. Pomponi M.G., Neri G.. Butyrate and acetyl-carnitine inhibit the cytogenetic expression of the fragile X in vitro. Am. J. Med. Genet. (1994) 51 447-450. DOI: 10.1002/ajmg.1320510428 56. Te Boekhorst V., Preziosi L., Friedl P.. Plasticity of cell migration in vivo and in silico. Annu. Rev. Cell Dev. Biol. (2016) 32 491-526. DOI: 10.1146/annurev-cellbio-111315-125201 57. Wenzel U., Nickel A., Daniel H.. Increased mitochondrial palmitoylcarnitine/carnitine countertransport by flavone causes oxidative stress and apoptosis in colon cancer cells. Cell. Mol. Life Sci. CMLS (2005) 62 3100-3105. DOI: 10.1007/s00018-005-5378-7 58. Wenzel U., Nickel A., Daniel H.. Increased carnitine-dependent fatty acid uptake into mitochondria of human colon cancer cells induces apoptosis. J. Nutr. (2005) 135 1510-1514. DOI: 10.1093/jn/135.6.1510 59. Jiang F., Zhang Z., Zhang Y., Pan X., Yu L., Liu S.. L-Carnitine ameliorates cancer cachexia in mice partly via the carnitine palmitoyltransferase-associated PPAR-γ signaling pathway. Oncol. Res. Treat. (2015) 38 511-516. DOI: 10.1159/000439550 60. Liu S., Wu H.-J., Zhang Z.-Q., Chen Q., Liu B., Wu J.-P., Zhu L.. L-carnitine ameliorates cancer cachexia in mice by regulating the expression and activity of carnitine palmityl transferase. Cancer Biol. Ther. (2011) 12 125-130. DOI: 10.4161/cbt.12.2.15717 61. Roscilli G., Marra E., Mori F., Di Napoli A., Mancini R., Serlupi-Crescenzi O., Virmani A., Aurisicchio L., Ciliberto G.. Carnitines slow down tumor development of colon cancer in the DMH-chemical carcinogenesis mouse model. J. Cell. Biochem. (2013) 114 1665-1673. DOI: 10.1002/jcb.24508 62. Montesano A., Senesi P., Vacante F., Mollica G., Benedini S., Mariotti M., Luzi L., Terruzzi I.. L-Carnitine counteracts in vitro fructose-induced hepatic steatosis through targeting oxidative stress markers. J. Endocrinol. Investig. (2020) 43 493-503. DOI: 10.1007/s40618-019-01134-2 63. Leung E., McArthur D., Morris A., Williams N.. Cyclooxygenase-2 inhibition prevents migration of colorectal cancer cells to extracellular matrix by down-regulation of matrix metalloproteinase-2 expression. Dis. Colon Rectum (2008) 51 342-347. DOI: 10.1007/s10350-007-9120-4 64. Lai K.-C., Huang A.-C., Hsu S.-C., Kuo C.-L., Yang J.-S., Wu S.-H., Chung J.-G.. Benzyl isothiocyanate (BITC) inhibits migration and invasion of human colon cancer HT29 cells by inhibiting matrix metalloproteinase-2/-9 and urokinase plasminogen (uPA) through PKC and MAPK signaling pathway. J. Agric. Food Chem. (2010) 58 2935-2942. DOI: 10.1021/jf9036694 65. Kurebayashi J., Otsuki T., Kunisue H., Mikami Y., Tanaka K., Yamamoto S., Sonoo H.. Expression of vascular endothelial growth factor (VEGF) family members in breast cancer. Jpn. J. Cancer Res. (1999) 90 977-981. DOI: 10.1111/j.1349-7006.1999.tb00844.x 66. Ishigami S., Arii S., Furutani M., Niwano M., Harada T., Mizumoto M., Mori A., Onodera H., Imamura M.. Predictive value of vascular endothelial growth factor (VEGF) in metastasis and prognosis of human colorectal cancer. Br. J. Cancer (1998) 78 1379-1384. DOI: 10.1038/bjc.1998.688 67. Inai T., Mancuso M., Hashizume H., Baffert F., Haskell A., Baluk P., Hu-Lowe D.D., Shalinsky D.R., Thurston G., Yancopoulos G.D.. Inhibition of vascular endothelial growth factor (VEGF) signaling in cancer causes loss of endothelial fenestrations, regression of tumor vessels, and appearance of basement membrane ghosts. Am. J. Pathol. (2004) 165 35-52. DOI: 10.1016/S0002-9440(10)63273-7
--- title: Retinal Vessel Local Tortuosity under a Macula-to-Optic Disc Central-Framing Change authors: - Natalia Ramírez - Miquel Ralló - Maria S. Millan journal: Diagnostics year: 2023 pmcid: PMC10046985 doi: 10.3390/diagnostics13061030 license: CC BY 4.0 --- # Retinal Vessel Local Tortuosity under a Macula-to-Optic Disc Central-Framing Change ## Abstract Some ocular and cardiovascular diseases can be detected through the increased tortuosity of retinal blood vessels. Objective tortuosity measures can be obtained from digital image analysis of a retinography. This study tested a set of local tortuosity indices under a change in the frame center (macula, optic disc) of the eye fundus image. We illustrate the effects of such a change on 40 pairs of vessels evaluated with eight tortuosity indices. We show that the frame center change caused significant differences in the mean values of the vast majority of the tortuosity indices analyzed. The index defined as the ratio of the curvature to the arc length of a vessel segment proved to be the most robust in relation to a frame center change. Experimental results obtained from the analysis of clinical images are provided and discussed. ## 1. Introduction The retinal blood vessels are directly and noninvasively observable through the crystalline medium of the eye. They have been shown to be one of the first structures directly affected by arterial hypertension and vascular dysregulation [1]. Their thickness, tortuosity and degree of deformation increase in response to cardiovascular overexertion caused by increased blood pressure [2]. The tortuosity of the retinal vessels can be used as a biomarker of several types of abnormalities, such as diabetic retinopathy [3], retinopathy of prematurity, glaucoma, macular degeneration, diabetes mellitus, hypertension, ischemic heart disease, some types of genetic disorders that affect hypertension and coronary artery disease [1,4,5], and other complications that can cause damage to the cardiovascular system (diabetes, coronary heart disease, etc.) [ 6]. Moreover, since the retinal vessels are an extension of the vessels of the brain, their appearance can be also related to the presence of cerebrovascular disease [5,7]. The retinal vascular network consists of a set of vessels, namely, arteries and veins, arranged in a double branching structure emerging from the optic disc, which is commonly the brightest spot in a retinography (eye fundus image). The first classifications of the general appearance of retinal vessels were based on subjective visual grading. It was not until 1979 that the first objective assessments of vessel tortuosity based on eye fundus photographs were performed [8]. Later, digital image processing applied to fundus images allowed the introduction of local indices for an enhanced objective measurement of retinal vessel tortuosity [9,10,11,12,13]. The usual local tortuosity indices are defined from a few geometrical features of the curve described by a vessel segment: curve length, distance between curve endpoints, total curvature, and total squared curvature [9,13]. The lack of a standardized protocol for image acquisition is still a limiting factor that hinders the assessment and its practical applicability to early diagnosis, progression monitoring, and treatment efficacy. Local tortuosity indices are calculated from retinographies with frames indistinctly centered on the macula (M) or on the optic disk (D) (see, for instance, the figures contained in [9,10,11,14,15]), with no mention of the effect that a frame center change might have on the values of the local tortuosity indices. In this work, we studied whether the center setting of the retinographies, either on the macula (M-retinography) or on the optic disc (D-retinography), has significant effects on the measurements of local tortuosity indices. To this end, we compared the values of eight widely known local tortuosity indices, as defined in [9,13], measured in forty vessels segmented from pairs of retinographies centered on the macula and the optic disc. ## 2.1. Local Tortuosity Indices Local tortuosity indices measure the amount of twisting (ridges and valleys) of a vessel. They are commonly measured for a number of vessel segments presenting no bifurcation. These vessel segments can eventually contain bifurcation points, but not multiple branches. Local tortuosity metrics of vessel segments are commonly defined from the ratios of some geometric parameters [9,13], such as chord length (D), arc length (L), total curvature (TK), and total squared curvature (TSK). L, TK, and TSK can be defined from a regular parametrization Ct of the curve C traced by the vessel segment. Regularity entails a continuous differentiability of Ct and the fulfillment of the condition on its derivative C′t≠0→, meaning that it presents no cusps or backtracks on itself. Let Ct=xt,yt, with t0≤t≤t1, be a regular curve describing the centerline of a vessel segment in terms of the 2D coordinate space x,y, the definitions of TK and TSK are based on the curvature κt [1]κt=y″t·x′t−y′t·x″tx′t2+y′t232. Table 1 shows the definitions of the four geometric parameters (D, L,TK, TSK). L, TK, and TSK are line integrals expressed in terms of the parametrization Ct. Eight local tortuosity indices, beginning with the most widely used distance factor (DF), a variant thereof, T1, and the following T2…T7, have been defined from these geometric parameters [13], as shown in Table 2. Figure 1 illustrates further the DF index as “the relative length increase over a straight vessel” [13]. ## 2.2. Participants and Equipment To illustrate the problem, we analyzed 40 pairs of vessel segments extracted from the retinographies of two subjects {1, 2}. Two retinographies were acquired of each subject’s eye {left (L), right (R)}: one centered on the macula (M), and the other centered on the optic disc (D) (Figure 2). The set of eight images were labeled as reported in Table 3. All images were acquired at the Ophthalmology Department at Mataró Hospital (Consorci Sanitari del Maresme, Barcelona, Spain) by a single experienced optometrist with a TRC-NW400 non-mydriatic retinal camera (Topcon Healthcare, Tokyo, Japan), in TIF format, RGB color, and a resolution of 768 × 806 pixels. Both subjects were selected among those regularly attending the service for the range of tortuosity observed in their retinas and the clarity of the vessel tree. We considered that their eye fundus images were useful for the purpose of our study. All the acquired images were anonymized by the hospital team before release. Figure 2 displays one of those pairs of images. ## 2.3. Region of Interest (ROI) We cropped the original images for ROI selection. This way, the pixels were not altered—neither their RGB intensity values nor their geometrical features after extraction from each original retinography. Since we wanted to compare two different images of each individual vessel segment, it was essential to preserve the ROI content unaffected. Only vessel segments contained in both images of every M/D pair were interesting to explore the potential effect of changing the frame center on the tortuosity measurements. We developed an algorithm, based on binary masks, to roughly crop the common region of each M/D pair of images (Figure 3). Firstly, we cut each circular image of the retina by excluding the black corners of the square frame with an appropriate binary mask. Next, we located the optic disc center in each retinography from its brightest point. To determine such a position, we smoothed the RGB components of each image with a median filter. A window of 21 × 21 pixels (about 3 times bigger than the thickest vessel diameter) was used to remove impulse noise and fine details from the image (Figure 3a,b). The brightest point of the image was calculated from the midpoint of the brightest pixels of the R, G, B components (red cross-shaped points in Figure 3a,b). In the case of having more than one pixel with maximum value in some component, the particular midpoint would be calculated for such a component prior to calculating the midpoint of the three R, G, B maxima for optic disc center location. This procedure yielded two optic disc center points: one for the M-retinography and the other for the D-retinography (red square-shaped points in Figure 3c,d). An analogous procedure, but for the darkest point, allowed us to determine the macula center (green square-shaped point in Figure 3c) and hence, to calculate the distance between the optic disc and the macula center points in one of the images (arbitrarily chosen to be the D-retinography). We roughly estimated the radius of the optic disk as one fifth of such a distance [16]. We translated either image to make their optic disk center points coincide and clear outside the overlapping area (gray-shaded region in Figure 3e). Additionally, we used the estimated optic disc radius to clear it with a circular mask in both images. The resulting ROIs (DR2-ROI and MR2-ROI in Figure 3f and Figure 3g, respectively) provided us with two image versions of each vessel segment whose tortuosity was to be evaluated. ## 2.4. Vessel Segmentation and Parametrization Figure 4 illustrates the segmentation process starting from DR2-ROI (Figure 4a). We developed a Matlab (Mathworks, Natick, MA, USA) tool to assist in the manual selection of the 40 vessel segments included in this study. The Matlab tool allowed the user to roughly draw a free-hand line on a vessel path (blue line in Figure 4b), between two formerly selected endpoints. The purpose of this line was to create a specific ROI for the vessel segment by padding the line with a surrounding area of 20 pixels (let us recall that it means about three times bigger than the thickest vessel diameter), in both the X and the Y directions (Figure 4c). The endpoints’ coordinates were recorded as tentative values and later used to refine the final endpoint positions of the segment. To make it more accurate, we considered endpoints clearly identifiable in both retinographies, so they were limited to bifurcation points, crossing points, the optic disk edge, the general ROI edge, or branch endpoints. This condition for the endpoints aimed to reduce subjective inaccuracies in the double vessel-ROI selection of each segment. The part of the vessel tree contained in that vessel-ROI was binarized (Figure 4d) following the method described in [17]. This is an unsupervised binarization method that overcomes the common problem of non-uniform illumination of eye fundus images. The method follows with an iterative algorithm that starts with a seed and adds, at each iteration, a new vessel segment connected to the previously segmented part. The result preserves the connectivity as a distinct feature of the retinal vessel tree. For the current work, the described method [17] provided the segmentation of the part of the tree contained in the vessel-ROI and, hence, the vessel segment of interest. To smooth small irregularities, the result was improved with a morphological closing operation (Figure 4d). A circular structural element with a radius of 7 pixels (the thickest vessel width) was used for that closing. Next, we skeletonized the segmented part of the vessel tree (Figure 4e). Endpoints, bifurcation points, and crossing points were recognized in the skeletonized element (Figure 5a) using the following criteria: pixels with only one neighbor pixel were labeled as endpoints (in yellow in Figure 5b–d), whereas pixels with three or more neighbor pixels were labeled as tree-branching pixels (either bifurcation or crossing). This set of labeled pixels should contain the two principal endpoints of the vessel segment, which could differ slightly from the previously recorded as tentative endpoints. From all the pixels labelled as end, bifurcation, and crossing point pixels in the skeletonized vessel, we eventually identified the two principal endpoints of the vessel segment as those being the closest to the tentative ones. Needle branches were removed by an iterative procedure: secondary endpoints, that is, endpoints other than the principal, were removed from the skeleton; next, new tree secondary endpoints of the remaining skeleton were found and further removed. The procedure was repeated until no endpoints other than the principal ones remained. As a result, we obtained a curve, 1-pixel wide of connected pixels, running along the central line of the selected vessel segment (Figure 5e). We proceeded to the parametrization of the resulting line of the vessel segment. We built a string with their pixel coordinates. The first element of the string was arbitrarily assigned to the principal endpoint closest to the [0, 0] pixel of the original image. It was followed by the next neighbor pixel of the line and so on. Note that the parameter runs from $t = 1$ to the total number of pixels of the vessel segment. A final smoothing operation with a 3-term moving average completed the parametrization of the vessel segment. Figure 6 displays the 10 vessels selected from the retinography DR2. ## 2.5. Vessel Curvature Besides the chord (D) and arc (L) lengths (Table 1 and Figure 1), the definition of other tortuosity indices (Table 2) involves total curvature (TK) and total squared curvature (TSK). Curvature is defined from the first and second derivatives of the curve coordinates, and their direct calculation as differences between consecutive terms usually displays a noisy behavior. Smoothing is frequently used to handle this issue, but it can alter the geometry of the curve when is applied excessively. We estimated the first and second derivatives following a robust method against small perturbations of the curve, described in [18]. The method uses the second-order Taylor expansion of the coordinate functions and the weighted least squares method. For each point of a string xm,ym and for p=±4,±8,±12,±16,±20, we obtained two sets of 10 equations from the second-order Taylor expansion (the number of equations decreased down to 5 for points close to the principal endpoints):[2]xm+p=xm+x’m·p+12 x’’m·p2+εxm,p, p=±4,±8,±12,±16,±20, [3]ym+p=ym+y’m·p+12 y’’m·p2+εym,p, p=±4,±8,±12,±16,±20. We set p values taking into account that 4 pixels corresponded roughly to half the width of the thickest vessel. The terms εxm,p y εym,p represent higher order contributions in Equations [2] and [3]. Both systems of equations were solved using weighted least squares with weights equal to 1p, which assigned lower weight to equations corresponding to farther neighbor pixels. Equation [2] was solved for x’m and x’’m, and Equation [3] for y’m and y’’m. The sequences x’m, x’’m, y’m, and y’’m were finally smoothed using a 3-pixel moving average operator. The resulting derivatives were used to calculate the curvature κt (Equation [1]). ## 3. Results We applied the method described in the previous section to the set of 40 pairs of vessel segments selected from the D- and M- retinographies of both (R, L) eyes of subjects 1 and 2. To illustrate the results, Figure 7 shows the individual values of the indexes DF, T3, T5, and T7 for the ten vessel segments selected from DR2. They appear colored according to the grade of tortuosity (tortuosity increases from dark magenta to light blue). For the vessels in Figure 7, the four indices showed different values. Three of them (T3, T5, and T7) coincided in pointing vessel 3 as the most tortuous, while the DF index indicated vessel 4. The T5 and T7 indices very similar (only vessels 4 and 6 appeared swapped in second and third positions from the highest tortuosity). All four indices agreed in marking vessel 1 as the least tortuous of the group. Table 4 contains the mean and standard deviation values of the tortuosity indices computed for the set of vessel segments in either D- or M- retinography. From the table, it stands out that the indices were expressed in very different scales, even orders of magnitude apart; therefore, their direct comparison is not meaningful. For each index, their statistical values seemed to be quite similar for the D- and M- frame centers, but a detailed analysis of the individual differences using Bland–Altman plots (Figure 8) and paired t-tests (Table 5) revealed significant differences for most of them. We also analyzed any possible effect of the specific eye under examination on the individual differences of the tortuosity indices. All the points corresponding to a given eye share the same color in Figure 8. For each tortuosity index, a one-factor ANOVA showed no significant differences between the eyes. One-factor ANOVA tests compared the four means of the tortuosity differences between pairs of vessels, a mean value for each eye. These tests are suited to detect heterogeneity across eyes. For each tortuosity index, the Bland–Altman plot showed systematic differences caused by a D-to-M frame center change. The X-axis accounts for the amount of tortuosity (mean value), and the Y-axis for the individual differences. In each plot, the central line represents the mean value of the individual differences (second column in Table 5) and denotes the systematic difference (bias) caused by the frame center change: all the tortuosity indices had a positive bias, that is, reached higher values, on average, when they were evaluated in D- rather than in M-retinographies. The upper and lower lines are the limits of concordance (mean of the differences ± 1.96 * standard deviation of the differences, in the fourth and fifth columns of Table 5). All systematic differences were statistically significant (i.e., different from 0, with p-values < 0.05 in Table 5), except for T4. In other words, the mean tortuosity values resulting from either M- or D-retinographies were significantly different for all tortuosity indices, excluding T4. The higher p-values (0.040, 0.112, 0.038) correspond to tortuosity indices based not on κ2, but on linear κ (T2, T4, and T6). From the definition of the concordance limits in the Bland–Alman plot, about $5\%$ of the differences caused by the frame center change should lie outside the limits of agreement presented in Table 5. Three points fell beyond the limits in the T5 and T7 plots, very close to the two ($5\%$) expected points. The DF and T1 plots are identical, as a consequence of the T1 definition (T1=DF−1, see Table 2). The plots showed higher differences for higher tortuosity values, although this behavior was mild with T4. Therefore, the variability increased with the magnitude of the tortuosity measure. The points corresponding to 18, 14, 23, and 36 pairs of vessels, were fairly beyond the limits of concordance in some plots of Figure 8. They corresponded to moderate to high tortuosity values. The corresponding pairs of vessels appear redrawn in magenta (vessel from M-retinography) and cyan (vessel from D-retinography) in Figure 9. The cyan lines seem to have more hairpin turns than the magenta ones; therefore, the cyan lines computed higher curvature values. This was confirmed by the geometrical features TK and TSK of those vessels, listed in Table 6. Moreover, in D-retinographies, their lengths L (except for the 23) appeared to be greater, while the chords D appeared to be shorter. These two facts, longer L and shorter D chord, led to more twisted lines. Since the values reached by the set of indices in Table 4 were not straightforwardly comparable, we analyzed the consistence of the indices through the Spearman rank correlation coefficient (ρ). For consistence, it is meant that a vessel characterized by a high value of a specific tortuosity index would also present a high value of the other tortuosity indices. In other words, the list of vessels ordered by their tortuosity using a specific index should be equal or very similar to the list obtained considering other indices. Table 7 contains the Spearman rank correlation coefficient and the Pearson correlation coefficient (r) for pairs of tortuosity indices. Let us recall that r assesses the linear relationship between two tortuosity indices. As expected from their definition, an exact coincidence of tortuosity rank orders was found for DF and T1 (ρ=1, $r = 1$), closely followed by T5 and T7 (ρ=0.998, $r = 0.946$). The lowest values of both the Spearman and the Pearson correlation coefficients were found for T4 and DF or T1 (ρ=0.775, $r = 0.758$). Finally, a dendrogram (Figure 10) summarizes the similarities among the set of tortuosity indices. The tree diagram displays the groups arising from an iterative clustering of the tortuosity indices. The dissimilarity level (1−r) according to Pearson’s correlation coefficient is represented on the vertical axis, and the indices are listed on the horizontal axis. The lowest correlation was found between the cluster of indices based on the vessel length (DF, T1) and the cluster of indices based on the vessel curvature (T2 … T7). Within this second cluster, the biggest dissimilarity was observed between the subgroup of total curvature measurements (T2, T3) and the subgroup of relative curvature measurements (T4 … T7), concerning either TK (T4, T6) or TSK (T5, T7). ## 4. Discussion and Conclusions The tortuosity of a retinal vessel tree can be analyzed on retinographies with a frame center either on the macula (M) or on the optic disk (D). No recommendation or standard protocol has been found to use one or another for image acquisition. We have analyzed the effect of a frame center change on the tortuosity values measured through eight local tortuosity indices already introduced in the field and widely used in related literature [13]. To illustrate the issue with examples, we selected 40 vessel segments from the clinical fundus images of two subjects’ eyes, ten segments per eye. Two separate retinographies, M- and D-centered, of each eye, provided a pair of parametric descriptions for the centerline of each vessel segment. Vessel segments were selected to have easily identifiable endpoints in both retinographies and also to exemplify a varied grade of tortuosity. Our results showed that a frame center change affected significantly the tortuosity measures of almost all indices of the set. The tortuosity indices reached higher values when they were evaluated through a D-retinography than through an M-retinography. The differences were statistically significant for all the indices tested, except for the index T4, which is based on the total curvature (TK) divided by the arc length (L). The Bland–Altman plots also showed unwanted behaviors of the individual differences. For all the indices, the standard deviation tended to increase across the tortuosity magnitude value, producing inverted funnel-shaped plots. The effect was mild with T4. However, the inverted funnel was clear with the DF (and its variant T1) indices, as well as with the indices based on the squared curvature (TSK) (T3, T5, and T7). For them, we also analyzed the cases of extreme differences, meaning those points beyond the tortuosity limits of concordance in the Bland–Altman plots. They corresponded to vessels with higher total curvatures (for most of them, longer arc lengths too) and shorter chord lengths in the D-retinographies than in the M-retinographies. The tortuosity index T4 was the most robust when performing a frame center change (M- to D-) among the eight indices analyzed. Based on the Spearman correlation coefficient (ρ), the rank order obtained with T4 was very similar to those derived from the other curvature-based indices T3 (0.921), T5 (0.983), T6 (0.995), and T7 (0.982); however, the rank ordered obtained with T4 showed the maximum difference with respect to that derived from DF (and T1) (0.775). When analyzing retinographies with either D or M frame center, the tortuosity index T4 appeared to compensate better for the tiny effects of the perspective change on the parametric description of the vessels and, hence, on their local tortuosity measures. The DF index (and T1), not based on the vessel curvature, showed significant differences when evaluated in either a D- or an M-retinography. This fact needs to be taken into account, since DF and its variant T1 are conceptually simple and the most widely used tortuosity measures [9,13,19]. Moreover, its use has been objected as it may underestimate vessel tortuosity, as reported by Kalitzeos et al. [ 13] and formerly by Aslam et al. [ 20]. The rest of the tortuosity indices (T2, T3, T5,... T7), though based on the curvature (TK or TSK) and, some of them, also divided by either the arc (T5) or the chord (T6, T7) length (Table 1 and Table 2), do not capture the same representation of a vessel tortuosity with the change of the frame center. In the dendrogram (Figure 10), DF and T1 appear as redundant. T4 and T6, on the one hand, and T5 and T7, on the other hand, are very similar, as they only differ in the sort of length magnitude used in denominator (arc or chord length). On the following level, the curvature-based indices (T2, T4, T6) are dissimilar to the square curvature-based indices (T3, T5, and T7), and yet, those absolute curvature-based indices (T2, T3) in a higher level of dissimilarity from the indices with curvature relative to a length magnitude (T4, T5, T6, T7). In the highest level of dissimilarity, we find the DF (T1) index separated from the rest of curvature-based indices (T2 … T7). This work has two obvious limitations. Firstly, although we found significant differences in most of the indices used to evaluate retinal vessel tortuosity when changing the fundus frame center (macula, optic disc), we ignored the impact such a difference may have in the clinical practice. We think this fact deserves to be investigated. Secondly, we analyzed just 40 pairs of vessel segments to provide an example that may illustrate the issue and motivate further research. The investigation should be extended to cover a variety of cases, in particular, those with clinical interest. Finally, and most importantly, these results should lead to establish standard protocols for eye fundus acquisition relative to the evaluation of retinal vessel tortuosity. ## References 1. Witt N., Wong T.Y., Hughes A.D., Chaturvedi N., Klein B.E., Evans R., McNamara M., Thom S.A.M., Klein R.. Abnormalities of retinal microvascular structure and risk of mortality from ischemic heart disease and stroke. Hypertension (2006) 47 975-981. DOI: 10.1161/01.HYP.0000216717.72048.6c 2. Cheung C.Y., Zheng Y., Hsu W., Lee M.L., Bcomp Q.P.L., Mitchell P., Mmed J.J.W., Klein R., Wong T.Y.. Retinal vascular tortuosity, blood pressure, and cardiovascular risk factors. Ophthalmology (2011) 118 812-818. DOI: 10.1016/j.ophtha.2010.08.045 3. Sun Z., Yang D., Tang Z., Ng D.S., Cheung C.Y.. Optical coherence tomography angiography in diabetic retinopathy: An updated review. Eye Lond (2021) 35 149-161. DOI: 10.1038/s41433-020-01233-y 4. Veluchamy A., Ballerini L., Vitart V., Schraut K.E., Kirin M., Campbell H., Joshi P.K., Relan D., Harris S., Brown E.. Novel Genetic Locus Influencing Retinal Venular Tortuosity Is Also Associated With Risk of Coronary Artery Disease. Arterioscler. Thromb. Vasc. Biol. (2019) 39 2542-2552. DOI: 10.1161/ATVBAHA.119.312552 5. van de Kreeke J.A., Nguyen H.T., Konijnenberg E., Tomassen J., den Braber A., Kate M., Sudre C.H., Barkhof F., Boomsma D.I., Tan H.S.. Retinal and Cerebral Microvasculopathy: Relationships and Their Genetic Contributions. Investig. Ophthalmol. Vis. Sci. (2018) 59 5025-5031. DOI: 10.1167/iovs.18-25341 6. McClintic B.R., McClintic J.I., Bisognano J.D., Block R.C.. The relationship between retinal microvascular abnormalities and coronary heart disease: A review. Am. J. Med. (2010) 123 374.e1-374.e7. DOI: 10.1016/j.amjmed.2009.05.030 7. Stefánsson E., Olafsdottir O.B., Einarsdottir A.B., Eliasdottir T.S., Eysteinsson T., Vehmeijer W., Vandewalle E., Bek T., Hardarson S.H.. Retinal Oximetry Discovers Novel Biomarkers in Retinal and Brain Diseases. Invest. Ophthalmol. Vis. Sci. (2017) 58 BIO227-BIO233. DOI: 10.1167/iovs.17-21776 8. Lotmar W., Freiburghaus A., Bracher D.. Measurement of vessel tortuosity on fundus photographs. Albrecht Graefes Arch. Klin. Exp. Ophthalmol. (1979) 211 49-57. DOI: 10.1007/BF00414653 9. Hart W.E., Goldbaum M., Côté B., Kube P., Nelson M.R.. Measurement and classification of retinal vascular tortuosity. Int. J. Med. Inform. (1999) 53 239-252. DOI: 10.1016/S1386-5056(98)00163-4 10. Grisan E., Foracchia M., Ruggeri A.. A novel method for the automatic grading of retinal vessel tortuosity. IEEE Trans. Med. Imaging (2008) 27 310-319. DOI: 10.1109/TMI.2007.904657 11. Trucco E., Azegrouz H., Dhillon B.. Modeling the tortuosity of retinal vessels: Does caliber play a role?. IEEE Trans. Biomed. Eng. (2010) 57 2239-2247. DOI: 10.1109/TBME.2010.2050771 12. Onkaew D., Turior R., Uyyanonvara B., Akinori N., Sinthanayothin C.. Automatic Retinal Vessel Tortuosity Measurement Using Curvature of Improved Chain Code (2011) 183-186 13. Kalitzeos A.A., Lip G.Y., Heitmar R.. Retinal vessel tortuosity measures and their applications. Exp. Eye Res. (2013) 106 40-46. DOI: 10.1016/j.exer.2012.10.015 14. Mapayi T., Tapamo J.R., Viriri S., Adio A.O.. Automatic Retinal Vessel Detection and Tortuosity Measurement. Image Anal. Stereol. (2016) 35 117-135. DOI: 10.5566/ias.1421 15. Ramos L., Novo J., Rouco J., Romeo S., Álvarez M.D., Ortega M.. Computational assessment of the retinal vascular tortuosity integrating domain-related information. Sci. Rep. (2019) 9 19940. DOI: 10.1038/s41598-019-56507-7 16. Mok K.H., Lee V.W.. Disc-to-macula distance to disc-diameter ratio for optic disc size estimation. J. Glaucoma (2002) 11 392-395. DOI: 10.1097/00061198-200210000-00004 17. Ralló M., Millán M.S.. Connectivity-based segmentation of retinal vessels in eye fundus images. Opt. Pura Apl. (2017) 50 359-368. DOI: 10.7149/OPA.50.4.49070 18. Lewiner T., Gomes J.D., Lopes H., Craizer M.. Curvature and torsion estimators based on parametric curve fitting. Comp. Grap. (2005) 29 641-655. DOI: 10.1016/j.cag.2005.08.004 19. Ramos L., Novo J., Rouco J., Romeo S., Álvarez M.D., Ortega M.. Retinal vascular tortuosity assessment: Inter-intra expert analysis and correlation with computational measurements. BMC Med. Res. Methodol. (2018) 18. DOI: 10.1186/s12874-018-0598-3 20. Aslam T., Fleck B., Patton N., Trucco M., Azegrouz H.. Digital image analysis of plus disease in retinopathy of prematurity. Acta Ophthalmol. (2009) 87 368-377. DOI: 10.1111/j.1755-3768.2008.01448.x
--- title: Prediction of Coronary Artery Disease Using Machine Learning Techniques with Iris Analysis authors: - Ferdi Özbilgin - Çetin Kurnaz - Ertan Aydın journal: Diagnostics year: 2023 pmcid: PMC10046987 doi: 10.3390/diagnostics13061081 license: CC BY 4.0 --- # Prediction of Coronary Artery Disease Using Machine Learning Techniques with Iris Analysis ## Abstract Coronary Artery Disease (CAD) occurs when the coronary vessels become hardened and narrowed, limiting blood flow to the heart muscles. It is the most common type of heart disease and has the highest mortality rate. Early diagnosis of CAD can prevent the disease from progressing and can make treatment easier. Optimal treatment, in addition to the early detection of CAD, can improve the prognosis for these patients. This study proposes a new method for non-invasive diagnosis of CAD using iris images. In this study, iridology, a method of analyzing the iris to diagnose health conditions, was combined with image processing techniques to detect the disease in a total of 198 volunteers, 94 with CAD and 104 without. The iris was transformed into a rectangular format using the integral differential operator and the rubber sheet methods, and the heart region was cropped according to the iris map. Features were extracted using wavelet transform, first-order statistical analysis, a Gray-Level Co-Occurrence Matrix (GLCM), and a Gray Level Run Length Matrix (GLRLM). The model’s performance was evaluated based on accuracy, sensitivity, specificity, precision, score, mean, and Area Under the Curve (AUC) metrics. The proposed model has a $93\%$ accuracy rate for predicting CAD using the Support Vector Machine (SVM) classifier. With the proposed method, coronary artery disease can be preliminarily diagnosed by iris analysis without needing electrocardiography, echocardiography, and effort tests. Additionally, the proposed method can be easily used to support telediagnosis applications for coronary artery disease in integrated telemedicine systems. ## 1. Introduction Approximately 17.9 million people die annually due to cardiovascular disease, about $30\%$ of global deaths [1]. The American Heart Association reports that about half of American adults are affected by heart disease. If precautions are not taken, then by 2030, the global death toll is projected to rise to 22 million [2]. Coronary Artery Disease (CAD) has the highest mortality rate among cardiovascular diseases [3]. Coronary arteries are the arteries on the surface of the heart that supply the heart with blood. The blood pumped by the heart first carries oxygen to the heart muscles through the coronary arteries. Three main coronary arteries exist: the left anterior descending artery, the left circumflex artery, and the right coronary artery. CAD occurs due to the decrease or complete cessation of blood flow to the heart muscle caused by the hardening of these coronary arteries [4,5]. The main cause of hardening (plaque formation) in the vessels is the accumulation of fatty or fibrous materials on the inner walls of the vessels, also called atherosclerosis. Plaques are mostly composed of lipids, cholesterol, and apoptosis residues which reduce blood flow, increasing the risk of blood clot formation and embolization [6]. This study defines patients with CAD as patients who are 18 years or older and have at least one clinical scenario of a chronic coronary syndrome (CCS) based on the 2019 European Society of Cardiology (ESC) guidelines for the diagnosis and treatment of CCS. The clinical scenarios for CAD include: (i) patients with suspected CAD and stable anginal symptoms and/or dyspnea, (ii) patients with newly onset heart failure (HF) or left ventricular (LV) dysfunction and suspected CAD, (iii) asymptomatic and symptomatic patients, or recently revascularized patients with stable symptoms less than one year after ACS, (iv) asymptomatic and symptomatic patients more than one year after the diagnosis of CCS or revascularization, (v) patients with angina and suspected vasospastic or microvascular disease, and (vi) asymptomatic participants detected to have CAD during routine screenings. A series of laboratory tests and imaging methods are used to diagnose CAD. The diagnosis is made by evaluating the patient’s complaints, family history, risk factors, and the results of physical examination findings. To diagnose CAD, blood tests, electrocardiography (ECG), effort tests, Holter tests, and echocardiography (ECHO) are commonly used tests [7,8]. The onset of symptoms in patients with CAD can range from simple nonspecific chest pain to a direct heart attack or even death. Neglected findings can lead to a heart attack; even if the patient does not die, severe damage to the heart muscle can occur. Therefore, early diagnosis is very important in CAD [4]. In recent years, the iris, which contains many nerve endings, has been used for the early diagnosis of diseases. The iris contains approximately 28,000 neural networks communicating between the brain and organs [9]. If an organ is not functioning properly, information is sent to the brain about this situation, which is reflected in the iris as a change in pattern, color, or characteristic feature. Iridology is the study of the changes in pattern, texture, color, and structure that occur in the special regions of the iris and their relationship with various diseases. As a result of various studies within the field of iridology, iris maps were created that show the regions in the iris that are related to specific organs and tissues. Bernard Jensen finalized the Iris map, which consists of 166 sections, 80 on the right and 86 on the left [10,11]. ## 1.1. Related Work on IRIS When reviewing the literature, iridology studies investigate the anatomical changes in specific areas of the iris, which are typically caused by functional changes in a particular organ or tissue [12]. Ma et al. discovered with significant accuracy that diseases can be diagnosed using geometric features such as the size of the pupil, shape, and shape of the iris [13]. Samant and Agarwal conducted a study to diagnose diabetes using various machine-learning techniques by analyzing the texture of the iris pancreatic region. The study found an accuracy rate of around $90\%$ [14]. Similarly, many other models for diagnosing diabetes have been proposed by researchers in recent years [15,16,17]. Rehman et al. proposed an iridology-based approach for diagnosing chronic liver disease [18]. They found that iris analysis combined with the ensemble learning method had an accuracy rate of approximately $98\%$. In the literature, there are studies on diseases of organs such as the kidney [19] and brain [20] using iridology, and there are various studies on cholesterol values in the blood [12,21,22,23]. In line with these studies, iridology has been shown to be effective in the non-invasive early diagnosis of diseases. However, there is a limited amount of research on the use of iris analysis for the early diagnosis of heart diseases. Various researchers around the world have made significant discoveries in non-invasive image processing and artificial intelligence-based diagnosis by using iris images related to the heart, which is a vital organ for maintaining life functions. Gunawan et al. [ 24] proposed a method for detecting coronary artery disease using the Support Vector Machines (SVM) classifier with five Gray-Level Co-Occurrence Matrix (GLCM) features. In their study involving 250 volunteers, the features of 100 volunteers were used as test data, and the Gaussian kernel SVM classifier achieved $91\%$ accuracy in detecting coronary artery disease. Putra et al. [ 25] developed a system with 90 volunteers utilizing iris analysis to detect cardiac issues. They employed the Principal Component Analysis (PCA) and Gray-Level Co-Occurrence Matrix (GLCM) methods to extract features in the system they developed, and they performed the classification process using neural networks. They achieved a classification accuracy of $77.5\%$ for the test data using GLCM features, and they achieved $90\%$ accuracy using PCA features. The PCA feature extraction method and SVM classifier were utilized in the method proposed by Permatasari et al. [ 26]. The highest accuracy achieved was reported to be $80\%$. Kusuma et al. [ 27] proposed a model for detecting cardiac abnormalities by acquiring and using iris images with a mobile-based system. The ratio of black and white pixels obtained after converting the analysis region to black and white format was used as a feature. The accuracy performance value for the test data, as classified by the thresholding method, was measured at $83.3\%$. These studies demonstrate the effectiveness of using iridology for the diagnosis of CAD. ## 1.2. Research Gaps of Previous Work on IRIS/CAD When studies in the literature are examined, it is seen that various methods are used to diagnose heart diseases via the iris early. However, it appears that no specific heart disease has been evaluated in depth. These studies follow a standard procedure, including finding the iris positions, performing the rectangular transformation, determining the analysis region, extracting the features from the analysis region, and classification. The differences in the studies begin after the determination of the analysis region. When the studies are examined at this stage, it is seen that the sub-components were formed by applying the wavelet transform to the analysis region, and although successful results were obtained in the studies conducted for the diagnosis of diabetes, this method has not been tested for heart diseases. In this study, more comprehensive and qualified results were obtained compared to the existing studies for the diagnostics of CAD by increasing the number of features to be extracted using the wavelet transform and the number of classifiers. ## 1.3. Contribution of This Paper In this study, a new diagnostic approach is proposed using iris images for the non-invasive detection of CAD. The data used in the study were collected from 198 volunteers, including 94 individuals with CAD and 104 control individuals, from the Cardiology Polyclinic of Giresun University Health Practice and Research Hospital. The study includes a feature selection method based on wavelet transform, resulting in 136 features, including statistical, GLCM, and Gray-Level Run Length Matrix (GLRLM) features. According to their rank values, the best 25, 50, and 75 features were selected using the Relieff method. A total of 22 classifiers belonging to the Decision Trees (DT), Naive Bayes (NB), Support Vector Machines (SVM), k-Nearest Neighbor (kNN), and Neural Networks (NN) families, which are commonly used for classification, were applied. The performance metrics calculated in the study indicate that the proposed model is more successful in detecting CAD than existing models. Detailed comparisons and evaluations are provided in the Results and Discussion section. The main contributions of this study can be summarized as follows:A novel diagnostic approach is proposed for the non-invasive detection of CAD using iris images. The Relieff feature selection method based on wavelet transform is introduced, resulting in 136 features including statistical, GLCM, and GLRLM features. A comparison is made between different classifiers, such as DT, NB, SVM, kNN, and NN, and the best-performing classifier is identified. The proposed model was compared with existing models and was more successful in detecting CAD. ## 2. Materials and Methods In the study methodology, a standard design was carried out to diagnose CAD through a non-invasive procedure. The flow chart of the study is shown in Figure 1. ## 2.1. Subject Selection for Data Acquisition In this study, the dataset was created by collecting iris images from 198 subjects with the volunteers’ consent and with the assistance of relevant doctors from the Giresun University Health Practice and Research Hospital Cardiology Polyclinic. Ethics committee approval was obtained for data collection per the decision of Samsun University Clinical Research Ethics Committee, numbered SUKAEK-2022 $\frac{12}{21}$, dated 23 November 2022. Out of the 198 volunteers, 94 were diagnosed with CAD, while 104 were healthy individuals without the disease. The incidence of CAD varies according to gender, with it being more common in men [1]. As a result, the proportion of men among the volunteers included in the study is higher than that of women. Of the volunteers aged between 19 and 86 who participated in the study, 156 were men and 42 were women. Table 1 and Figure 2 provide detailed information about the age, gender, and health status of the volunteers. ## 2.2. Eye Image Acquisition Left eye images of the subjects labeled as having CAD and of those labeled as healthy by their respective doctors were collected. Eye images were taken using a Nikon D3300 DSLR camera with a Nikon AF-S DX Micro Nikkor 85 mm F/3.5G VR lens and with macro ring flash illumination. The resulting images were 6000 × 4000 in size and had a resolution of 24 megapixels. Example images for both healthy and CAD volunteers are provided in Figure 3. ## 2.3. Eye Image Pre-Processing After obtaining the eye images, they needed to go through several pre-processing steps to prepare them for analysis. Algorithm 1 and Figure 4 illustrate the eye image pre-processing process step-by-step. Algorithm 1 Eye image pre-processing algorithm[1] Input: Eye image[2] Iris localization from the eye image (a) Localization pupil using Daugman’s Integral Differential Operator (b) Localization iris using Daugman’s Integral Differential Operator[3] Iris normalization using Daugman’s rubber sheet Technique - Normalized iris becomes a fixed size: 360 × 720[4] ROI cropped according to the iris map in Figure 4 - The ROI size is 190 × 120[5] ROI enhancement using the CLAHE method[6] Output: ROI image Figure 4: An example of the pre-processing process used in the study. The techniques used for the image pre-processing process are as follows: ## 2.3.1. Iris Localization At this stage, the pupil and iris positions were determined from the image. The iris positions in the image converted to the gray format were determined using the integral differential operator (IDO) [28]. The IDO method can accurately determine the inner and outer borders of the iris by using different values of pupil and sclera color. The mathematical expression of the method is provided in the equation below. [ 1]maxr,x0,y0⁡Gσr∂∂r∮r,x0,y0I(x,y)2πrds Here, the expression I(x, y) denotes the color value of the (x, y) position in the image I. x0 and y0 represent the coordinates of the potential center point, and the symbol r represents the distance to the potential center point. Gσ represents the Gaussian function with σ standard deviation. ## 2.3.2. Iris Normalization The normalization process was the next step after determining the iris’s inner and outer positions. The iris was transformed into a rectangular format in the normalization process, standardizing it and making it easier to analyze. As a result of the normalization process, the rectangular iris image was resized to a fixed size of 360 × 720. Daugman’s rubber sheet method, as shown in Figure 5, is one of the most commonly used normalization methods, and it was used in this study. The remapping of the iris image from the I(x, y) cartesian coordinates to the polar representation can be expressed as the following equation. [ 2]I(xr,θ,yr,θ)→I(r,θ) where [3]xr,θ=1−rxpθ+rxl(θ) [4]yr,θ=1−rypθ+ryl(θ) Here, the I(x, y) is the iris region, (x, y) represents the Cartesian coordinates, (r, θ) represents the normalized polar coordinates, and xp, yp and xl, yl are expressions that denote the pupil and iris boundary coordinates in the θ direction. ## 2.3.3. Region of Interest (ROI) After completing the normalization process, the Region of Interest (ROI) was cropped according to the heart region in the left iris in the iris map shown in Figure 6. The heart region is located in the left iris between the 2 and 4 o’clock positions. After converting the circular iris image to a fixed-size rectangle, the heart region in the iris was cropped. ## 2.3.4. Enhancement of ROI Histogram equalization is a commonly used image enhancement technique due to its high performance and simplicity. It redistributes the probabilities of the occurrence of gray-levels so that the histogram of the output image is closer to a uniform distribution. Although the method generally gives good results, it may not achieve the desired improvement in images with darker or lighter colored pixels than other pixel values. To address this limitation, instead of using the whole image for equalization, the image was divided into certain regions, and the histogram equalization of the regions increased image improvement performance. The Contrast Limited Adaptive Histogram Equalization (CLAHE) method [29] was used for this purpose. In this study, the CLAHE method was used for ROI correction. ## 2.4. Iris Feature Extraction Because the iris contains many blood vessels and nerves, it has a very rich structural pattern. Many researchers have extracted features from the iris using various methods such as the Gabor Filter, Hilbert Transform, and Discrete Wavelet Transform (DWT). In this study, DWT transformation was used for feature extraction. The process of feature extraction is outlined in Algorithm 2. Algorithm 2 Feature extraction process[1] Input: ROI Image[2] Perform 1 Level 2D-DWT to ROI image - Four sub-bands occur (cA, cV, cD, cH)[3] Extract features from sub-bands (a) Extract 5 first-order statistical features as shown in Table 2 (b) Extract 22 GLCM-based features as shown in Table 3 - Formation of the 8 × 8 GLC matrix using θ = [00, 450, 900, 1350] with $d = 1.$ Values for each direction are found and averaged (c) Extract 7 GLRLM-based features as shown in Table 4 - Formation of the GLRL matrix using θ = [00, 450, 900, 1350] with quantize level = 16. Values for each direction are found and averaged[4] Fusion of features (5 statistical + 22 GLCM + 7 GLRLM = 34 features for each sub-band)[5] Output: feature vector with 136 features DWT decomposes an image into four subsampled images, as shown in Figure 7, namely the approximation (LL), horizontal (HL), vertical (LH), and diagonal (HH) images. The input image of size N × N is divided into four sub-images, each of size N/2 × N/2. Each sub-image contains information from different frequency components [30]. In Figure 7, the LL sub-band was obtained by applying low-pass filtering to both rows and columns, resulting in an image with less noise than the other sub-bands. The HH band was obtained by applying high-pass filtering in both directions, and it contains higher frequency components than the other bands. The HL and LH sub-bands were obtained by using low-pass filtering in one direction and high-pass filtering in the other. The LH sub-band mostly contains vertical detail information corresponding to horizontal edges, while the HL sub-band contains horizontal detail information corresponding to vertical edges. The HL, LH, and HH sub-bands add high-frequency detail to the approximate image. The image is typically decomposed multiple times using the DWT, usually starting with the LL band [31]. A block diagram of the feature extraction process is shown in Figure 8. In Figure 8, cA describes the approximation coefficients matrix, and cH, cV, and cD describe the detail coefficients’ matrices (horizontal, vertical, and diagonal, respectively). A total of 34 features were extracted for each of the four coefficients’ matrices (cA, cH, cV, cD). These features included five statistical features, 22 GLCM (Gray Level Co-occurrence Matrix) features, and 7 GLRLM (Gray Level Run Length Matrix) features. At the end of the feature extraction process, 136 feature vectors (34 for each region) were obtained. This study used a 1-level DWT decomposition to analyze the ROI image. Statistical features and features obtained using GLCM and GLRLM were extracted for each sub-band. Figure 9 provides an example of extracting features for a sample image. The attributes of the extracted features are described in the following headings. ## 2.4.1. Statistical Features The study calculated and used the ROI’s five first-order statistical features: mean density, standard deviation, entropy, skewness, and kurtosis. The mathematical expressions for these parameters obtained from the gray-level ROI are provided in Table 2. Five statistical features were obtained for each sub-band. Table 2 \| Feature Name \| Formula \| Feature Name.1 \| Formula.1 \| \| --- \| --- \| --- \| --- \| \| Mean intensity \| 1N∑i=1NX(i) \| Skewness \| 1N∑i=1N(Xi−X−)31N∑1N(Xi−X−)23 \| \| Standard deviation \| 1N−1∑i=1N(Xi−X−)21/2 \| Kurtosis \| 1N∑i=1N(Xi−X−)41N∑1N(Xi−X−)22 \| \| Entropy \| ∑i=1N1Pi.log2⁡P(i) \| Kurtosis \| 1N∑i=1N(Xi−X−)41N∑1N(Xi−X−)22 \| ## 2.4.2. Gray-Level Co-Occurrence Matrix (GLCM) Features Using only first-order statistical approaches is insufficient for detecting and grading textures or patterns in an image. These features provide information about the intensity distribution but do not reveal the relationship between pixels. To gain information about neighboring pixels, GLCM and related features offered by Haralick et al. [ 32] can be used. GLCM is a gray-level matrix that characterizes, quantifies, and explores the distribution of gray-level intensities. Direction and neighborhood information is used when calculating GLCM. As shown in Figure 10, the 0°, 45°, 90°, and 135° directions were used. When creating the GLCM, the grayscale value of each pixel in the image was calculated as given in Equation [5]. [ 5]Pi,j=P(i,j,d,θ)∑$i = 1$∑$j = 1$P(i,j,d,θ) After the GLCM of the image was created, the textural features of the image were extracted from this matrix. This study used 22 parameters [32,33,34] to extract features using GLCM. The names, mathematical expressions, and definitions of these parameters are provided in Table 3. Table 3 \| Feature Name \| Formula \| Feature Name.1 \| Formula.1 \| \| --- \| --- \| --- \| --- \| \| Auto correlation \| ∑i=1N∑j=1Ni.jp(i,j) \| Information measure of correlation 1 \| HXY−HXY1max⁡(HX,HY) \| \| Cluster prominence \| ∑i=1N∑j=1Ni+j−2u3p(i,j) \| Information measure of correlation 2 \| 1−exp⁡[−2HXY2−HXY] \| \| Cluster shade \| ∑i=1N∑j=1Ni+j−2u4p(i,j) \| Inverse difference moment \| ∑i=1N∑j=1Np(i,j)1+i−j \| \| Contrast \| ∑i=1N∑j=1Ni−j2p(i,j) \| Maximum probability \| maxi,jp(i,j) \| \| Correlation \| ∑i=1N∑j=1Ni−µxσxj−µyσyp(i,j) \| Sum average \| ∑k=22Nkpx+y(k) \| \| Difference entropy \| −∑k=0N−1px−yklog px−y(k) \| Sum entropy \| −∑k=22Npx+yklog⁡px+y(k) \| \| Difference variance \| ∑k=0N−1(k−µx−y)2px−y(k) \| Sum of squares \| ∑i=1N∑j=1Npi−µ2p(i,j) \| \| Dissimilarity \| ∑i=1N∑j=1Ni−j.p(i,j) \| Sum variance \| ∑k=22Nk−µx+y2px+y(k) \| \| Energy \| ∑i=1N∑j=1Np(i,j)2 \| Maximal correlation coefficient \| λ2(Qi,j) \| \| Entropy \| −∑i=1N∑j=1Np(i,j)log p(i,j) \| Inverse difference normalized \| ∑i=0N−1∑j=0N−111+i−j2p(i,j) \| \| Homogeneity \| ∑i=1N∑j=1Np(i,j)1+(i−j)2 \| Inverse difference moment normalized \| ∑i=0N−1∑j=0N−1p(i,j)1+i−jN2 \| The features listed in Table 3 were calculated for the four sub-bands obtained after the wavelet transform. For each wavelet component, the features calculated by considering pixels in four directions and one neighbor distance were averaged. This resulted in the creation of 22 GLCM attributes for each region. ## 2.4.3. Gray-Level Run Length (GLRL) Matrix Features The Gray-Level Running Length Matrix (GLRLM) method is based on calculating the number of different gray-level lengths [32]. It is a way of extracting higher-order statistical texture features. A gray-level run is a linear array of adjacent image points with the same gray-level value. The gray-level run length is the number of image points in the array. GLRLM is a two-dimensional matrix and is used for texture feature extraction. In this study, seven attributes, along with their names, mathematical equations, and descriptions, are provided in Table 4, which were used when using GLRLM. Table 4 \| Feature Name \| Formula \| Feature Name.1 \| Formula.1 \| \| --- \| --- \| --- \| --- \| \| Short Run Emphasis (SRE) \| ∑i=1G∑j=1Rp(i,jθ)j2/∑i=1G∑j=1Rp(i,jθ)1 \| Run Length Non-Uniformity (RLN) \| ∑j=1R∑i=1Gp(i,jθ)2/∑i=1G∑j=1Rp(i,jθ) \| \| Long Run Emphasis (LRE) \| ∑i=1G∑j=1Rj2×p(i,jθ)/∑j=1Rp(i,jθ) \| Low Gray-Level Run Emphasis (LGRE) \| ∑i=1G∑j=1Rp(i,jθ)i2/∑i=1G∑j=1Rp(i,jθ) \| \| Gray-Level Non-Uniformity (GLN) \| ∑i=1G∑j=1Rp(i,jθ)2/∑i=1G∑j=1Rp(i,jθ) \| High Gray-Level Run Emphasis (HGRE) \| ∑i=1G∑j=1Ri2×p(i,jθ)/∑i=1G∑j=1Rp(i,jθ) \| \| Run Percentage (RP) \| 1N∑i=1G∑j=1Rp(i,jθ) \| High Gray-Level Run Emphasis (HGRE) \| ∑i=1G∑j=1Ri2×p(i,jθ)/∑i=1G∑j=1Rp(i,jθ) \| ## 2.5. Feature Selection Feature selection is an important step in reducing complexity and saving time in machine learning methods for classification problems. It makes classification more reliable by eliminating unnecessary data. Relieff, a widely used filter-based feature selection method, was preferred in this study. The algorithm developed by Kira et al. performs the selection process by weighting the parameters according to their relationship [35]. Kononenko created this algorithm, as the method did not give successful results in datasets with multiple classes [36]. The method selects a sample from the dataset and performs feature selection by creating a model based on the proximity of the sample to other samples in its class and based on its distance from different classes. In this study, the best 25, 50, and 75 features were selected among 136 features obtained from ROI. There were four sub-band images, each containing 34 features. Choosing specific features from each sub-band and including different feature groups can be beneficial in more effectively determining the impact of sub-bands and methods on performance. This approach helps to accurately identify the performance effects of sub-bands and methods. ## 2.6. Classification In classification, there are two main types: supervised and unsupervised. In supervised classification, the model performance is determined by the test data in models created using labeled data. In this study, 22 classifiers from 5 different classifier families, which are commonly used in literature, were employed. Although the classifiers mentioned above are commonly used in various fields, the MATLAB Classification Learner application, which includes standard parameters, was used in this study to avoid bias that may occur from manual selection of the parameters. The training and test data were divided into five groups using the fivefold cross-validation technique for the classification process. The performance values were obtained by taking the average of the parameters calculated five times. ## 2.7. Performance Evaluation Various evaluation metrics were used to determine the success of the models created during the classification process. These metrics are based on a table called the confusion matrix [37]. Each row of the matrix represents the actual values, and each column represents the predicted values. A two-class confusion matrix and the values it will take are shown in Figure 11. In Figure 11, TP refers to true positive results, FN refers to false negative results, FP refers to false positive results, and TP refers to true negative results. The metrics used in this study to determine the classification performance using the confusion matrix are listed in Table 5. Accuracy is the ratio of correct guesses to the total number of values. A high value indicates high accuracy. Specificity is the ratio of correct negative predictions to the total number of negatives. Precision is the ratio of correctly predicted positive observations to the total predicted positive observations, and it measures the accuracy of predictions for positive class. Sensitivity is the ratio of correctly predicted positive observations to all observations in the actual positive class. The F1-score is the harmonic mean of the ratio of true positive values (sensitivity) and precision. It is a measure of how well the classifier is performing. The geometric mean is a metric that measures the balance in classification between majority and minority classes. A low value indicates poor performance in the classification of positive cases, even if it correctly classified negative cases [38,39]. In addition to these metrics, the Receiver Operating Characteristic (ROC) curve was also used to measure performance. The ROC curve is a graphical representation of the performance of a classifier over all possible threshold values. It has False Positive Rate (FPR) on the x-axis and True Positive Rate (TPR) on the y-axis. The Area Under Curve (AUC) is the area under the ROC curve. The AUC value ranges from 0 to 1, and the closer the value is to 1, the better the model’s performance [40]. ## 3. Results and Discussion In this study, iris images of 198 volunteers were analyzed to detect coronary artery disease. The relationship between the 136 features obtained from the iris images and the target variable was first investigated. Then, the performance evaluations obtained from the classification process using the best 25, 50, and 75 features determined by the Relieff feature selection method were presented. ## 3.1. Feature Analysis The correlation coefficient values showing the relationship of the 136 features obtained from the wavelet transform with the target variable are illustrated in Figure 10. In the ROI, which is divided into four components after the wavelet transform, 34 features, five statistical, 22 GLCM, and seven GLRLM features were extracted for each component. The four components were labeled cA, cH, cV, and cD. In Figure 12, the components and attributes are presented in this order. The highest correlation value of 0.6734 belonged to the 134th feature, RP, which is a GLRLM attribute of the cD sub-band. There were 13 features in total with a correlation value above 0.6, three features with values between 0.5 and 0.6, two features between 0.4 and 0.5, 24 features between 0.3 and 0.4, and 27 features between 0.2 and 0.3. Among the 10 features with the highest correlation coefficients, there were three features in the cA component, two in the cH component, two in the cV component, and three in the cD component. Nine of these features belonged to GLRLM features, and one of them belonged to a GLCM feature not among the 10 features with the highest 1st-order statistical feature coefficients. Out of the nine GLRLM attributes, LRE 4, LGRE 3, and RP were included twice. RP was the two best attributes. The GLCM attribute also had the highest correlation coefficient. From the high correlation coefficients of the features, it can be seen that the features were evenly distributed among the components obtained from the wavelet transform. It can be observed that the statistical features had lower correlation coefficients compared to the other feature groups, and the highest coefficients were in the GLRLM and GLCM features. ## 3.2. Results after Feature Selection Before the classification process, the feature selection process was applied. Using the Relieff algorithm, the best 25, 50, and 75 features were determined according to their rank values. The first 25 (Group 1), second 25 (Group 2), and third 25 (Group 3) attribute groups with the highest rank are listed in Table 6. The metrics obtained from the classification process using the attributes in Group 1 in Table 6 are listed in Table 7. In total, the accuracy values ranged from 0.64 to 0.90 for the 22 classifiers. The Fine Gaussian SVM method had the lowest accuracy, whereas the Narrow Neural Network had the highest accuracy. The sensitivity value was 0.96, and the recall value was 0.96, with the highest from the Kernel Naive Bayes method. While the Decision Tree performed well in specificity and precision values, the Narrow Neural Network performed better in Fscore and Gmean metrics. Medium and Coarse Gaussian SVM were the best classifiers for the AUC value. The performance evaluation values, as a result of the analysis in which the best 50 attributes obtained by combining the attributes in Group 1 and Group 2 in Table 6, were used as the inputs listed in Table 8. There are four methods in the table with an accuracy value of 0.9. Although three of these methods were included in the SVM methods, one of them is from the Neural Network family. It can be said that the SVM method’s classifiers gave better performance metrics results than other methods. The specificity, precision, recall, Fscore, and Gmean values were 1.00, 1.00, 0.96, 0.91, and 0.92, respectively. The highest AUC value was seen in the classifier Naive Bayes. The values in Table 9 were obtained when all of the features in Groups 1, 2, and 3 were included in the analysis. The highest accuracy value was obtained by combining these three groups. The Medium Gaussian SVM method had the highest accuracy value for this feature group, with a value of 0.93. This value was also the highest value among all analyses. The medium Gaussian SVM classifier was the best classifier according to the sensitivity, recall, Fscore, Gmean, AUC, and accuracy values. The highest precision value was seen in Gaussian Naive Bayes, whereas the highest specificity value of 0.94 was seen in Fine Gaussian. As the number of features used in the analysis increased, the cost and the performance values of many classifiers increased. Although the values of the metrics obtained as a result of Naive Bayes, SVM, and kNN analyses increased close to a linear increase with the increase in the number of features, it was seen that there was an increase in some of the Decision Tree and Neural Network classifiers and a decrease in others. Nevertheless, it can be said that the classifiers included in the study achieved high success in detecting coronary artery disease. ## 3.3. Comparison with Studies in the Literature The comparative values of the findings in Table 7, Table 8 and Table 9 and the studies on the diagnosis of heart disease from iris images in the literature are listed in Table 10. The table includes the feature extraction methods, classifier names, and evaluation metrics used in existing studies. Among existing studies, Gunawan et al. [ 24] obtained $91\%$ accuracy using the SVM classifier with GLCM features. Putra et al. [ 25] reached an accuracy value of 0.78 by using the Neural Network with the same feature extraction method and also achieved $90\%$ success with the PCA method. Kusuma et al. [ 27] and Permatasari et al. [ 26] used the Black and White Ratio and PCA methods for feature extraction, respectively, and performed classification with the Thresholding and SVM methods, respectively. These studies did not include performance metrics other than accuracy. In this study, using wavelet transform-based statistical, GLCM, and GLRLM features and five different classifiers, a higher accuracy value of $93\%$ was obtained with the SVM classifier compared to other studies. In addition, the second highest value was obtained in the NN classifier, with an accuracy value of $92\%$. In this study, unlike other studies, performance measurements such as sensitivity, specificity, precision, Fscore, Gmean, and AUC were carried out in addition to accuracy. These values indicate that the analysis successfully detected coronary artery disease. ## 4. Conclusions This study proposes a non-invasive method for detecting coronary artery disease (CAD), as verified in an experiment that used the iris images of 198 volunteers. After the iris pre-processing processes, a total of 136 statistical, GLCM, and GLRLM features were extracted from the four subcomponents obtained by applying wavelet transform to the heart region in the iris. The Relieff feature selection process was used to determine the best 25, 50, and 75 features before classification. The classification phase was carried out using 22 classifiers from five main classifier families. Accuracy, sensitivity, specificity, precision, Fscore, Gmean, and AUC metrics were used to evaluate performance. The SVM Medium Gaussian classifier achieved the highest accuracy value at $93\%$. According to the results of the other classifiers, it can be said that the CAD classification of the values of accuracy and other metrics yielded successful results. It can be stated that the proposed method for the detection of CAD from the iris is quite successful. The proposed method can be used to support telediagnostic applications for coronary artery disease in telemedicine systems. Thus, information about the patient’s CAD can be obtained by using the patient’s iris images in order to make a preliminary assessment before performing daily clinical practice. This study provides a reference for detecting CAD from iris images. In future studies, the relationship of various heart diseases, such as heart failure, with iris analysis can be examined. Performance improvement can be made by trying different feature extraction and machine learning methods and by detecting various diseases using convolutional neural networks. ## References 1. Virani S.S., Alonso A., Benjamin E.J., Bittencourt M.S., Callaway C.W., Carson A.P., Chamberlain A.M., Chang A.R., Cheng S., Delling F.N.. Heart disease and stroke statistics—2020 update: A report from the American Heart Association. Circulation (2020.0) 141 e139-e596. DOI: 10.1161/CIR.0000000000000757 2. Benjamin E.J., Muntner P., Alonso A., Bittencourt M.S., Callaway C.W., Carson A.P., Chamberlain A.M., Chang A.R., Cheng S., Das S.R.. Heart disease and stroke statistics—2019 update: A report from the American Heart Association. Circulation (2019.0) 139 e56-e528. DOI: 10.1161/CIR.0000000000000659 3. Nowbar A.N., Gitto M., Howard J.P., Francis D.P., Al-Lamee R.. Mortality from ischemic heart disease: Analysis of data from the World Health Organization and coronary artery disease risk factors From NCD Risk Factor Collaboration. Circ. Cardiovasc. Qual. Outcomes (2019.0) 12 e005375. DOI: 10.1161/CIRCOUTCOMES.118.005375 4. Malakar A.K., Choudhury D., Halder B., Paul P., Uddin A., Chakraborty S.. A review on coronary artery disease, its risk factors, and therapeutics. J. Cell. Physiol. (2019.0) 234 16812-16823. DOI: 10.1002/jcp.28350 5. Bauersachs R., Zeymer U., Brière J.-B., Marre C., Bowrin K., Huelsebeck M.. Burden of coronary artery disease and peripheral artery disease: A literature review. Cardiovasc. Ther. (2019.0) 2019 8295054. DOI: 10.1155/2019/8295054 6. Novak R., Hrkac S., Salai G., Bilandzic J., Mitar L., Grgurevic L.. The role of ADAMTS-4 in atherosclerosis and vessel wall abnormalities. J. Vasc. Res. (2022.0) 59 69-77. DOI: 10.1159/000521498 7. Ghiasi M.M., Zendehboudi S., Mohsenipour A.A.. Decision tree-based diagnosis of coronary artery disease: CART model. Comput. Methods Programs Biomed. (2020.0) 192 105400. DOI: 10.1016/j.cmpb.2020.105400 8. Alizadehsani R., Zangooei M.H., Hosseini M.J., Habibi J., Khosravi A., Roshanzamir M., Khozeimeh F., Sarrafzadegan N., Nahavandi S.. Coronary artery disease detection using computational intelligence methods. Knowl.-Based Syst. (2016.0) 109 187-197. DOI: 10.1016/j.knosys.2016.07.004 9. Fausett L.V.. Fundamentals of Neural Networks: Architectures, Algorithms and Applications (2006.0) 10. Jensen B.. Iridology Simplified (2012.0) 11. Sivasankar K., Sujaritha M., Pasupathi P., Muthukumar S.. FCM based iris image analysis for tissue imbalance stage identification. Proceedings of the 2012 International Conference on Emerging Trends in Science, Engineering and Technology (INCOSET) 210-215 12. Kurnaz Ç., Gül B.K.. Determination of the relationship between sodium ring width on iris and cholesterol level. J. Fac. Eng. Archit. Gazi Univ. (2018.0) 33 1557-1568 13. Ma L., Zhang D., Li N., Cai Y., Zuo W., Wang K.. Iris-based medical analysis by geometric deformation features. IEEE J. Biomed. Health Inform. (2012.0) 17 223-231. PMID: 23144041 14. Samant P., Agarwal R.. Machine learning techniques for medical diagnosis of diabetes using iris images. Comput. Methods Programs Biomed. (2018.0) 157 121-128. DOI: 10.1016/j.cmpb.2018.01.004 15. Samant P., Agarwal R.. Analysis of computational techniques for diabetes diagnosis using the combination of iris-based features and physiological parameters. Neural Comput. Appl. (2019.0) 31 8441-8453. DOI: 10.1007/s00521-019-04551-9 16. Bansal A., Agarwal R., Sharma R.. Determining diabetes using iris recognition system. Int. J. Diabetes Dev. Ctries. (2015.0) 35 432-438. DOI: 10.1007/s13410-015-0296-1 17. Önal M.N., Güraksin G.E., Duman R.. Convolutional neural network-based diabetes diagnostic system via iridology technique. Multimed. Tools Appl. (2022.0) 82 173-194. DOI: 10.1007/s11042-022-13291-3 18. Rehman M.U., Najam S., Khalid S., Shafique A., Alqahtani F., Baothman F., Shah S.Y., Abbasi Q.H., Imran M.A., Ahmad J.. Infrared sensing based non-invasive initial diagnosis of chronic liver disease using ensemble learning. IEEE Sens. J. (2021.0) 21 19395-19406. DOI: 10.1109/JSEN.2021.3091471 19. Muzamil S., Hussain T., Haider A., Waraich U., Ashiq U., Ayguadé E.. An intelligent iris based chronic kidney identification system. Symmetry (2020.0) 12. DOI: 10.3390/sym12122066 20. Hernández F., Vega R., Tapia F., Morocho D., Fuertes W.. Early detection of Alzheimer’s using digital image processing through iridology, an alternative method. Proceedings of the 2018 13th Iberian Conference on Information Systems and Technologies (CISTI) 1-7 21. Ozbilgin F., Kurnaz C.. An alternative approach for determining the cholesterol level: Iris analysis. Int. J. Imaging Syst. Technol. (2022.0) 32 1159-1171. DOI: 10.1002/ima.22689 22. Özbilgin F.. Determination of Iris Symptoms of Systemic Diseases by Iris Analysis Method. Master’s Thesis (2019.0) 23. Ramlee R., Ranjit S.. Using iris recognition algorithm, detecting cholesterol presence. Proceedings of the 2009 International Conference on Information Management and Engineering 714-717 24. Gunawan V.A., Putra L.S.A., Imansyah F., Kusumawardhani E.. Identification of Coronary Heart Disease through Iris using Gray Level Co-occurrence Matrix and Support Vector Machine Classification. Int. J. Adv. Comput. Sci. Appl. (2022.0) 13. DOI: 10.14569/IJACSA.2022.0130177 25. Putra L.S.A., Isnanto R.R., Triwiyatno A., Gunawan V.A.. Identification of Heart Disease with Iridology Using Backpropagation Neural Network. Proceedings of the 2018 2nd Borneo International Conference on Applied Mathematics and Engineering (BICAME) 138-142 26. Permatasari L.I., Novianty A., Purboyo T.W.. Heart disorder detection based on computerized iridology using support vector machine. Proceedings of the 2016 International Conference on Control, Electronics, Renewable Energy and Communications (ICCEREC) 157-161 27. Kusuma F.D., Kusumaningtyas E.M., Barakbah A.R., Hermawan A.A.. Heart abnormalities detection through iris based on mobile. Proceedings of the 2018 International Electronics Symposium on Knowledge Creation and Intelligent Computing (IES-KCIC) 152-157 28. Daugman J.. How iris recognition works. The Essential Guide to Image Processing (2009.0) 715-739 29. Reza A.M.. Realization of the contrast limited adaptive histogram equalization (CLAHE) for real-time image enhancement. J. VLSI Signal Process. Syst. Signal Image Video Technol. (2004.0) 38 35-44. DOI: 10.1023/B:VLSI.0000028532.53893.82 30. Diwakar M., Tripathi A., Joshi K., Sharma A., Singh P., Memoria M.. A comparative review: Medical image fusion using SWT and DWT. Mater. Today Proc. (2021.0) 37 3411-3416. DOI: 10.1016/j.matpr.2020.09.278 31. Kumar S., Singh B.K.. DWT based color image watermarking using maximum entropy. Multimed. Tools Appl. (2021.0) 80 15487-15510. DOI: 10.1007/s11042-020-10322-9 32. Haralick R.M., Shanmugam K., Dinstein I.H.. Textural features for image classification. IEEE Trans. Syst. Man Cybern. (1973.0) 6 610-621. DOI: 10.1109/TSMC.1973.4309314 33. Soh L.-K., Tsatsoulis C.. Texture analysis of SAR sea ice imagery using gray level co-occurrence matrices. IEEE Trans. Geosci. Remote Sens. (1999.0) 37 780-795. DOI: 10.1109/36.752194 34. Clausi D.A.. An analysis of co-occurrence texture statistics as a function of grey level quantization. Can. J. Remote Sens. (2002.0) 28 45-62. DOI: 10.5589/m02-004 35. Kira K., Rendell L.A.. A practical approach to feature selection. Machine Learning Proceedings 1992 (1992.0) 249-256 36. Kononenko I.. Estimating attributes: Analysis and extensions of RELIEF. Proceedings of the European Conference on Machine Learning 171-182 37. Luque A., Carrasco A., Martín A., de Las Heras A.. The impact of class imbalance in classification performance metrics based on the binary confusion matrix. Pattern Recognit. (2019.0) 91 216-231. DOI: 10.1016/j.patcog.2019.02.023 38. Chicco D., Tötsch N., Jurman G.. The Matthews correlation coefficient (MCC) is more reliable than balanced accuracy, bookmaker informedness, and markedness in two-class confusion matrix evaluation. BioData Min. (2021.0) 14 1-22. DOI: 10.1186/s13040-021-00244-z 39. Room C.. Confusion Matrix. Mach. Learn (2019.0) 6 27 40. Hasan M.K., Alam M.A., Das D., Hossain E., Hasan M.. Diabetes prediction using ensembling of different machine learning classifiers. IEEE Access (2020.0) 8 76516-76531. DOI: 10.1109/ACCESS.2020.2989857
--- title: 'The Possible Effect of the Long-Term Use of Glucagon-like Peptide-1 Receptor Agonists (GLP-1RA) on Hba1c and Lipid Profile in Type 2 Diabetes Mellitus: A Retrospective Study in KAUH, Jeddah, Saudi Arabia' authors: - Ghada M. A. Ajabnoor - Kamal Talat Hashim - Mohammed Meshari Alzahrani - Abdullah Zeid Alsuheili - Abdullah Fahad Alharbi - Amani Matook Alhozali - Sumia Enani - Basmah Eldakhakhny - Ayman Elsamanoudy journal: Diseases year: 2023 pmcid: PMC10046996 doi: 10.3390/diseases11010050 license: CC BY 4.0 --- # The Possible Effect of the Long-Term Use of Glucagon-like Peptide-1 Receptor Agonists (GLP-1RA) on Hba1c and Lipid Profile in Type 2 Diabetes Mellitus: A Retrospective Study in KAUH, Jeddah, Saudi Arabia ## Abstract [1] Background: Type 2 diabetes (T2DM) is a chronic metabolic disease with serious health complications. T2DM is associated with many chronic illnesses, including kidney failure, cardiovascular diseases (CVD), vision loss, and other related diseases. Obesity is one of the major factors associated with insulin resistance and dyslipidemia. Recently, the development of GLP-1 Receptor agonist (GLP-1RA) showed great therapeutic potential for T2DM. Aim: To retrospectively investigate the association of the long-term use of GLP-1RA therapy in T2DM patients with HbA1c levels and dyslipidemia. [ 2] Methods: Retrospective data collection and analysis of demographic, clinical records, and biochemical parameters were carried out for 72 T2DM taking GLP-1RA treatments for six months. [ 3] Results: A total of 72 T2DM patients with a mean age = 55 (28 male and 44 female) were divided into two groups. Group 1 received statins ($$n = 63$$), and group 2 did not receive statins ($$n = 9$$). The GLP-1RA effect on BMI was significantly decreased in group 1 ($p \leq 0.01$). A significant effect was observed for HbA1c in both groups for six months of treatment duration ($p \leq 0.05$). The AST levels significantly decreased in group 2 from 25.2 to 19.4 U\L ($$p \leq 0.011$$). [ 4] Conclusions: GLP-1RA treatments were associated with weight reduction and improved glycemic control for T2DM patients. Moreover, it is suggested that it has anti-inflammatory and hepatoprotective effects. However, no direct association was found with the lipid profile in all groups of T2DM. ## 1. Introduction Type 2 diabetes mellitus (T2DM) is a chronic metabolic disease causing an increase in morbidity and mortality [1]. Globally, around 400 million people have diabetes, and it is expected that by 2040, diabetes incidence will increase to 640 million [2]. In 2019, the Global Burden of Diseases Study reported that diabetes is the direct cause of 1.5 million deaths [3]. The World Health Organization (WHO) reported that Saudi Arabia ranks second in the middle east with a high incidence of type 2 diabetes mellitus [4]. Consequently, type 2 diabetes mellitus is associated with many other chronic illnesses, including kidney failure (>$50\%$), cardiovascular diseases (CVD) ($70\%$), vision loss, and other related diseases [3]. The major complication of type 2 diabetes mellitus is the progression of macrovascular and microvascular complications [5]. Hence, type 2 diabetes mellitus patients are more susceptible to atherosclerosis and stroke than people without type 2 diabetes mellitus disease [1,6]. One of the significant risk factors for type 2 diabetes mellitus is obesity and its associated dyslipidemia and insulin resistance [7,8]. Dyslipidemia is an abnormal circulating blood level of lipids, including cholesterol, low/very low-density lipoprotein, high-density lipoprotein, and triglycerides [8]. These can lead to the development of atherosclerosis and CVD [8]. The increase in insulin levels leads to the rise of circulating free fatty acids, which is the main defect of lipid profiles in diabetic patients [6]. Many therapeutic strategies for diabetic patients have been used in the last decades. These strategies include lifestyle modification, weight loss, physical activity, and non-insulin and insulin medications [9,10]. The therapeutic management of type 2 diabetes mellitus includes metformin, sulfonylureas, insulin therapy, and thiazolidinediones [10]. Additionally, a recent development involves a new therapy that targets diabetic dyslipidemia, such as glucagon-like peptide receptor agonists (GLP-1RA) [6,10]. GLP-1 is a peptide hormone related to the glucagon superfamily [11] and shares a significant amino acid sequence with glucagon [12]. The glucagon superfamily peptides are secreted from the small intestine, pancreas, brain, and peripheral nerves [12]. They can act as hormones or neurotransmitters [6]. The proglucagon gene encodes the glucagon superfamily peptides, which generate a single precursor protein. It undergoes co-translation and post-translation processing to rise to different active peptides [12]. GLP-1 is one of these peptides secreted from the small intestine [9]. GLP-1 can also be expressed in tissues such as the endocrine pancreas, lower brain regions, and the large intestine [13]. GLP-1 secretion into the bloodstream is stimulated by food consumption or glucose; the active GLP-1 peptide disappears within 3 h [14]. GLP-1 acts as one of the primary regulators of blood glucose levels and stimulates insulin secretion in response to hyperglycemia [15]. GLP-1 acts as an incretin hormone by augmenting insulin release and inhibiting glucagon secretion, thereby regulating the postprandial glucose level [16]. A study by Drucker et al. found that GLP-1 showed an effective glucose-dependent insulin secretion in normal and diabetic animals and human tissues [9]. These findings, followed rapidly by another study, demonstrated that GLP-1 inhibited glucagon secretion, food intake, and gastric emptying [9]. Furthermore, GLP-1 controls intestinal motility and decreases gastric motility [17]. It also has an effect of satiety, which may be attributed to its effect on the gut, but it also has a direct effect on the hypothalamic feeding centres [17]. Consequently, these observations have led to diabetes therapy development in the form of GLP-1 receptor (GLP-1R) agonists for the treatment of type 2 diabetes mellitus and, subsequently, for obesity [9]. The establishment of GLP-1 receptor agnostic (GLP-1 RA) therapy has been used to improve insulin sensitivity in obese and diabetic patients [11]. GLP-1RA inhibits glucagon secretion from the pancreatic alpha cells when blood sugar levels are high and can decrease pancreatic beta cell apoptosis in proliferation [18,19,20]. Moreover, GLP-1RA showed an effect on satiety improvement and body weight loss. Thus, it can also be used for weight loss and dyslipidemia therapy [21]. Therefore, the GLP-1RA can be suitable for reducing plasma glucose in type 2 diabetes mellitus patients with dyslipidemia [11]. GLP-1RAs augment glucose-stimulated insulin secretion (GSIS) and suppress glucagon secretion at hyperglycemic or euglycemic conditions. Moreover, GLP-1RAs regulate cardiovascular and endothelial cell functions through an anti-inflammatory, antioxidant, and vasodilator effect on endothelial cells [22,23]. Thus, the reported benefits of GLP-1RA use in patients with CVD include a reduction in the risk of cardiovascular events [24] Until now, there is no reported data regarding the long-term effects of the therapeutic use of GLP-1RA on diabetes associated with dyslipidemia in Saudi Arabia. The current study aims to retrospectively assess the possible association of the long-term use of GLP-1RA effect in Saudi type 2 diabetes mellitus patients with Hb A1c and dyslipidemia. ## 2. Experimental Design An observational retrospective study included patients with type 2 diabetes mellitus ($$n = 72$$) adults older than 18 years who had been prescribed their first-time GLP-1 RA, such as liraglutide and/or semaglutide treatments. The study was conducted between January 2019 and December 2021 at King Abdulaziz University Hospital (KAUH), Jeddah, Saudi Arabia. Type 2 diabetes mellitus patients were following up regularly with their diabetology clinic at KAUH. The Research Ethics Committee at the Faculty of Medicine—King Abdulaziz University, Jeddah, Saudi Arabia, approved the study, reference no. 231-22. The electronic clinical database for patients and follow-up records were all obtained from the KAUH phoenix database system (phoenix). The demographic characteristics database for patients was collected at the starting date of GLP-1RA treatments and included the following: age, nationality, gender, height, history of statin medication, and comorbidities. The therapeutic protocol of liraglutide and semaglutide was as follows: liraglutide 0.6 mg subcutaneous injection daily for one week; weekly increase until it reached 3 mg. For Semaglutide, 0.25 mg subcutaneously once a week for four weeks, then progressively increased until it reached 1 mg; six patients received combined starting with liraglutide and shifted to semaglutide [25]. Additionally, patients selected for the study had pre- and post-medication records. The BMI, weight, and biochemical parameters were recorded twice over an estimated six-month period based on the pre-and post-index of each. The biochemical parameters included are cholesterol, triglycerides (TG), low-density lipoproteins (LDL), high-density lipoprotein (HDL), aspartate aminotransferase (AST), alanine aminotransferase (ALT), high sensitivity c-reactive protein (hs-CRP) and Hb-A1C. The low-density lipoprotein (VLDL) values were obtained using lipid panel values by Friedewald’s equation calculation [26]. The inclusion criteria included subjects with type II DM with dyslipidemia who received GLP-1RA for six months or more. The type of dyslipidemia (hypertriglyceridemia, hypercholesterolemia, and disturbed LDL/HDL ratio) in those patients is of the nonfamilial and non-genetic types. Exclusion criteria included patients with less than six months of GLP-1RA use, patients with familial hypercholesteremia or other genetic dyslipidemia, type 1 diabetes, patients with abnormal thyroid function tests, and pregnant females, as well as patients who had undergone bariatric surgery or taken weight-loss medications, such as orlistat. A flowchart of the selection criteria, including inclusion and exclusion criteria, is given in Figure 1. ## Statistical Analysis Data analysis was performed using IBM SPSS statistics version 23.0 for Windows. Baseline characteristics were expressed as the mean and standard error of the mean (SEM). An independent t-test analysis was used to compare factors between the two groups (statin vs. no statin), while paired t-test analysis was used to compare baseline and six months. The Pearson correlation coefficient (r) was used to study the correlation between the dose of GLP-1 receptor agonist and different parameters. ## 3. Results A summary of demographic and anthropometrics results presents in Table 1. In total, 72 diabetic patient records (28 male and 44 female) were analyzed. The average age was 55 years old. The average height was 171 cm in males and 156 cm in females, and the average BMI was 36.24 Kg/m2 in males and 38 Kg/m2 in females. Furthermore, 46 ($64\%$) of the patients were Saudis. In total, 29 patients were also diagnosed with hypertension and 8 with heart failure. Moreover, 63 patients (22 males and 41 females) were prescribed cholesterol-lowering medication in the form of high-intensity statins, rosuvastatin [36], and atorvastatin [27]. Accordingly, the study sample was divided into two groups, group 1 patients who received statin medication ($$n = 63$$) and group 2 patients without statin medication ($$n = 9$$). The patient diabetic drug history showed that 39 ($54\%$) were taking insulin, $76\%$ were taking metformin, $18\%$ were taking gliclazide, $8\%$ were taking repaglinide, and $2\%$ were taking glimepiride; their distributions between the statin and the non-statin group is presented in Table 2. Patients received different doses of GLP-1RA as subcutaneous injections for at least six months. Other parameters and biochemical tests were recorded at baseline (before receiving GLP-1RA) and during follow-up (after six months). The most significant finding can be seen in BMI and HbA1c, detailed in Table 3 and Figure 2. The BMI values were significantly changed from 37.3 kg/m2 to 35.6 kg/m2 with a p-value < 0.001 in the statin group. However, no statistically significant differences showed in the no-statin group. Furthermore, HbA1c significantly decreased in both groups, from $9.2\%$ to $7.9\%$ p-value < 0.001 in the statin group and from $6.9\%$ to $5.8\%$ p-value < 0.03 in the no statin group. Lipid profiles were recorded at baseline and after 6 months of GLP-1RA treatment. Some minor decreases could be seen in TG and VLDL levels in the no-statin group from 2.1 mmol/L to 1.7 mmol/L and 38.7 mg/dL to 30.37 mg/dL, respectively, but the changes were not statistically significant. On the other hand, GLP-1RA treatments showed no remarkable changes in cholesterol, HDL or LDL levels, as seen in Table 4. Results for liver enzymes and hs-CRP levels are presented in Table 5. ALT levels were decreased in both groups from 25.1 to 22.9 U\L in the statin group, with a p-value of 0.09 trending toward significance, and from 27.2 to 25.2 U\L in the no statin group. AST levels decreased only in the no statin group from 25.2 to 19.4 U\L and a p-value of 0.011. However, no apparent changes in the hs-CRP levels were found. Finally, patients received different doses of GLP-1 receptors agonist ranging from 0.25–3 mg\week. Thus, the correlation analysis was conducted to examine the relationship between the dosage and changes observed during the six months in any parameter shown in Table 6. There was a negative correlation between the dose and changes in LDL r = −0.326, with a p-value of 0.024. Conversely, the changes in cholesterol also negatively correlate with the dose, with the p-value trending toward a significant 0.08. ## 4. Discussion GLP-1RAs have been used as an important established therapy for diabetic patients throughout the last decade. Other diabetic therapies, including metformin, are still the standard first-line treatment for type 2 diabetes mellitus, according to the American Diabetes Association guidelines. Nevertheless, GLP-1RAs, in addition to type 2 diabetes mellitus patients therapy, are also considered for patients with metformin intolerance and HbA1c levels higher than $1.5\%$ of the normal target range [27,28]. In the current study, we investigated the association between the long-term use of GLP-1RAs with glycemic control and lipid profile in diabetic patients at King Abdulaziz University Hospital in Jeddah. Alanazi et al. group studied the effect of GLP-1RAs on body weight, body mass index, and HbA1c before and after the six-month use of GLP-1RA in patients with type II diabetes mellitus [29]. Their results revealed a significant reduction in BMI with no significant difference in HbA1c [29]. In the present study, we revealed a significant reduction in BMI in GLP-1RA-treated subjects who were treated with statin (lipid-lowering drugs). This is consistent with other previous studies that showed the effect of GLP-1RAs on reducing body weight [29,30,31,32,33,34,35]. Thus, GLP-1RAs lower the appetite and, consequently, decrease food intake by increasing satiety and a feeling of stomach fullness [25]. It also decreases gastrointestinal motility and reduces calorie intake [9,23]. This action occurs via the GLP-1RAs mechanism which involves stimulating insulin secretion and reducing glucagon secretion [9,22]. Furthermore, GLP-1RAs have an anorectic effect as they activate GLP-1Rs in the arcuate nucleus of the brain. Hence, GLP-1RAs act centrally on the brain–adipcyte axis. This mechanism could explain the anti-obesity effect of GLP-1RAs [25,36]. Regarding the relationship between GLP-1RAs and glycemic control, our findings show a significant decrease in HbA1c levels. It is found in the GLP-1RAs treated subjects with and without lipids-lowering drug therapy. This finding proves its role in managing glycemic parameters, consistent with previous studies [37,38,39,40]. Thus, it has been reported that GLP-1RAs enhance beta cell functions, which improve insulin sensitivity and reduce glucagon secretion to the lowest basal level [37]. Therefore, GLP-1RAs are effectively unique in lowering glycemic levels and reducing body weight. They also promote a decrease in glucosuria levels [25]. An increase in Homeostasis Model Assessment-2 B (HOMA2-B) indices and a decrease in proinsulin levels, proinsulin/C-peptide ratios, and proinsulin/insulin ratios were also reported to be a direct effect of GLP-1RAs [41,42]. Moreover, GLP-1RAs therapy can be associated with increased adiponectin gene expression and serum adiponectin level [43]. Adiponectin enhances the GLUT4-mediated glucose uptake in the white adipocytes, decreases hepatic lipid accumulation, and minimizes visceral fat depots [44]. Furthermore, GLP-1RAs improve endothelial cell function through their anti-inflammatory effect [45]. The anti-inflammatory effects of GLP-1RAs on macrophages protect against the development of atherosclerosis. Moreover, it is confirmed that GLP-1RAs produce a protective effect against hepatotoxicity and hepatic steatosis [46]. A study by Ohki et al. observed that GLP-1RA directly decreases liver fibrosis and steatosis in vivo. This effect is the principle of potential GLP-1RA therapeutic use for nonalcoholic fatty liver diseases (NAFLD) [47,48]. Consequently, the relationship between GLP-1RAs and AST serum level was also reported by Hartman et al. [ 46]. One of the apparent observations of the current study is the decrease in AST serum level, with no change observed regarding ALT. The improvement of one of the vital liver functions could be explained by the anti-inflammatory effect of GLP-1RAs [45] and the protection against lipotoxicity [44]. However, the direct effect of GLP-1RAs on reducing liver enzyme levels is not fully understood, requiring further investigation [47]. Surprisingly, the results of the current study did not demonstrate an association between GLP-1RAs treatment and lipid profile. No consistent relationship was observed with individual lipid profile components, including total cholesterol, triglyceride, LDL-C, HDL-C, and VLDL levels in subjects, either with or without lipids-lowering drugs(statin) therapy. Only a significant negative correlation was observed between GLP-1RAs dose and LDL-C level variations between pre-and post-treatment with GLP-1RAs. These findings were in accordance with the previous reports that showed an indirect, no direct association of GLP-1RA treatment effect with lipid profile levels [47,48,49]. However, Sun et al. reported a negative association between GLP-1RA monotherapy without statin therapy and LDL-C, total cholesterol, and triglycerides but no considerable improvement in HDL-C level [41]. In addition, Hasegawa et al. reported that it reduced LDL-C in patients with type II diabetes mellitus treated with statins in their study in Japan [50]. Collectively, the present study is one of few studies that assess the long-term association effect of GLP-1RAs therapy in patients with type II diabetes Mellitus in Saudi Arabia. In 2020, Alanazi and Ghoraba retrospectively investigated the association between the GLP-1RA treatment of type II diabetes mellitus and body mass index [21]. They concluded that liraglutide (one of the most widely used GLP-1RA therapeutic products in the King Abdulaziz University Hospital) has a crucial effect on weight loss and reduction in body mass index. One of the recent studies investigated the sodium-glucose cotransporter-2 inhibitors (SGLT2i) as a co-therapy with GLP-1 RAs for treating type II diabetes mellitus [51]. Both studies did not examine the potential effect of GLP-1 RA treatment on the serum lipid profile [29,51]. Our findings showed GLP-1RAs treatments are associated with weight reduction and improved glycemic control for patients of type II diabetes mellitus but have no direct effect on lipid profile. However, the current study has a few limitations, including data extracted from a single medical centre (King Abdulaziz University Hospital) and the limited sample number. Therefore, obtaining data from multiple medical centres could improve the findings regarding the effects of GLP-1RA therapy, with or without other lipid-lowering drugs. The limited number of the analyzed data after the exclusion of the results did not meet the inclusion criteria of our study. Moreover, the difficulty in obtaining access to the selected cases’ personal, other medical, and therapeutic data. A degree of individual compliance is necessary in order to use the drug. Further randomized multicentric studies in Saudi Arabia are needed to confirm the results of the current study and to fairly assess the drug’s efficacy among GLP-1RA receivers for either the weight reduction or treatment of diabetes. Additionally, investigating other possible favorable or unfavorable effects is important. ## 5. Conclusions This retrospective study concludes that GLP-1RAs treatment is associated with weight reduction and better glycemic control in type 2 diabetes within a six-month period. However, no direct association was observed regarding its relation to serum lipid profile except a negative correlation between GLP-1RAs with LDL-C. ## References 1. Pyörälä K., Laakso M., Uusitupa M.. Diabetes and atherosclerosis: An epidemiologic view. Diabetes/Metab. Res. Rev. (1987) 3 463-524. DOI: 10.1002/dmr.5610030206 2. Swami O.C., Sharma S.K., Panneerselvam A., Singh K.P., Parmar G., Gadge P.. Teneligliptin in management of type 2 diabetes mellitus. Diabetes Metab. Syndr. Obes. Targets Ther. (2016) 9 251-260. DOI: 10.2147/DMSO.S106133 3. Global Burden of Disease Study 2019; Latest GBD Results: 2019. (2020) 4. Al Dawish M.A., Robert A.A., Braham R., Al Hayek A.A., Al Saeed A., Ahmed R.A., Al Sabaan F.S.. Diabetes Mellitus in Saudi Arabia: A Review of the Recent Literature. Curr. Diabetes Rev. (2016) 12 359-368. DOI: 10.2174/1573399811666150724095130 5. Almeda-Valdes P., Cuevas-Ramos D., Mehta R., Gomez-Perez F., Aguilar-Salinas C.. UKPDS Risk Engine, Decode and Diabetes PHD Models for the Estimation of Cardiovascular Risk in Patients with Diabetes. Curr. Diabetes Rev. (2010) 6 1-8. DOI: 10.2174/157339910790442646 6. Patel V., Joharapurkar A., Shah G., Jain M.. Effect of GLP-1 Based Therapies on Diabetic Dyslipidemia. Curr. Diabetes Rev. (2014) 10 238-250. DOI: 10.2174/1573399810666140707092506 7. Hemmer A., Maiter D., Buysschaert M., Preumont V.. Long-term effects of GLP-1 receptor agonists in type 2 diabetic patients: A retrospective real-life study in 131 patients. Diabetes Metab. Syndr. Clin. Res. Rev. (2018) 13 332-336. DOI: 10.1016/j.dsx.2018.09.007 8. Ormazabal V., Nair S., Elfeky O., Aguayo C., Salomon C., Zuñiga F.A.. Association between insulin resistance and the development of cardiovascular disease. Cardiovasc. Diabetol. (2018) 17 122. DOI: 10.1186/s12933-018-0762-4 9. Drucker D.J.. Mechanisms of Action and Therapeutic Application of Glucagon-like Peptide-1. Cell Metab. (2018) 27 740-756. DOI: 10.1016/j.cmet.2018.03.001 10. Chaudhury A., Duvoor C., Reddy Dendi V.S., Kraleti S., Chada A., Ravilla R., Marco A., Shekhawat N.S., Montales M.T., Kuriakose K.. Clinical Review of Antidiabetic Drugs: Implications for Type 2 Diabetes Mellitus Management. Front. Endocrinol. (2017) 8 6. DOI: 10.3389/fendo.2017.00006 11. Nauck M.A., Meier J.J.. GLP-1 receptor agonists and SGLT2 inhibitors: A couple at last?. Lancet Diabetes Endocrinol. (2016) 4 963-964. DOI: 10.1016/S2213-8587(16)30263-7 12. Seon M.J., Hwang S.Y., Son Y., Song J., Kim O.Y.. Circulating GLP-1 Levels as a Potential Indicator of Metabolic Syndrome Risk in Adult Women. Nutrients (2021) 13. DOI: 10.3390/nu13030865 13. Lee C.-H., Jeon S.J., Cho K.S., Moon E., Sapkota A., Jun H.S., Ryu J.H., Choi J.W.. Activation of Glucagon-Like Peptide-1 Receptor Promotes Neuroprotection in Experimental Autoimmune Encephalomyelitis by Reducing Neuroinflammatory Responses. Mol. Neurobiol. (2017) 55 3007-3020. DOI: 10.1007/s12035-017-0550-2 14. D’Alessio D.. IsGLP-1 a hormone: Whether and When?. J. Diabetes Investig. (2016) 7 50-55. DOI: 10.1111/jdi.12466 15. Sandoval D.A., D’Alessio D.A.. Physiology of Proglucagon Peptides: Role of Glucagon and GLP-1 in Health and Disease. Physiol. Rev. (2015) 95 513-548. DOI: 10.1152/physrev.00013.2014 16. Holst J.J.. The Physiology of Glucagon-like Peptide 1. Physiol. Rev. (2007) 87 1409-1439. DOI: 10.1152/physrev.00034.2006 17. MacDonald P.E., El-Kholy W., Riedel M.J., Salapatek A.M.F., Light P.E., Wheeler M.B.. The Multiple Actions of GLP-1 on the Process of Glucose-Stimulated Insulin Secretion. Diabetes (2002) 51 S434-S442. DOI: 10.2337/diabetes.51.2007.S434 18. Garber A.J.. Long-Acting Glucagon-Like Peptide 1 Receptor Agonists. Diabetes Care (2011) 34 S279-S284. DOI: 10.2337/dc11-s231 19. Gallwitz B.. GLP-1 Agonists and Dipeptidyl-Peptidase IV Inhibitors. Diabetes—Perspectives in Drug Therapy (2011) 53-74. DOI: 10.1007/978-3-642-17214-4_3 20. Okerson T., Chilton R.J.. The Cardiovascular Effects of GLP-1 Receptor Agonists. Cardiovasc. Ther. (2010) 30 e146-e155. DOI: 10.1111/j.1755-5922.2010.00256.x 21. Cena H., Calder P.C.. Defining a Healthy Diet: Evidence for the Role of Contemporary Dietary Patterns in Health and Disease. Nutrients (2020) 12. DOI: 10.3390/nu12020334 22. Andreadi A., Bellia A., Di Daniele N., Meloni M., Lauro R., Della-Morte D., Lauro D.. The molecular link between oxidative stress, insulin resistance, and type 2 diabetes: A target for new therapies against cardiovascular diseases. Curr. Opin. Pharmacol. (2021) 62 85-96. DOI: 10.1016/j.coph.2021.11.010 23. Bray J.J.H., Foster-Davies H., Salem A., Hoole A.L., Obaid D.R., Halcox J.P.J., Stephens J.W.. Glucagon-like peptide-1 receptor agonists improve biomarkers of inflammation and oxidative stress: A systematic review and meta-analysis of randomised controlled trials. Diabetes Obes. Metab. (2021) 23 1806-1822. DOI: 10.1111/dom.14399 24. Bethel M.A., Patel R.A., Merrill P., Lokhnygina Y., Buse J.B., Mentz R.J., Pagidipati N.J., Chan J.C., Gustavson S.M., Iqbal N.. Cardiovascular outcomes with glucagon-like peptide-1 receptor agonists in patients with type 2 diabetes: A meta-analysis. Lancet Diabetes Endocrinol. (2018) 6 105-113. DOI: 10.1016/S2213-8587(17)30412-6 25. Nauck M.A., Quast D.R., Wefers J., Meier J.J.. GLP-1 receptor agonists in the treatment of type 2 diabetes—State-of-the-art. Mol. Metab. (2020) 46 101102. DOI: 10.1016/j.molmet.2020.101102 26. Wilson P.W., Zech L.A., Gregg R.E., Schaefer E.J., Hoeg J.M., Sprecher D.L., Brewer H.. Estimation of VLDL cholesterol in hyperlipidemia. Clin. Chim. Acta (1985) 151 285-291. DOI: 10.1016/0009-8981(85)90091-9 27. Hunt B., Malkin S.J.P., Moes R.G.J., Huisman E.L., Vandebrouck T., Wolffenbuttel B.H.R.. Once-weekly semaglutide for patients with type 2 diabetes: A cost-effectiveness analysis in the Netherlands. BMJ Open Diabetes Res. Care (2019) 7 e000705. DOI: 10.1136/bmjdrc-2019-000705 28. Burcelin R., Gourdy P.. Harnessing glucagon-like peptide-1 receptor agonists for the pharmacological treatment of overweight and obesity. Obes. Rev. (2016) 18 86-98. DOI: 10.1111/obr.12465 29. Alanazi N.K., Ghoraba M.A.. Effect of Glucagon-like peptide-1 agonist (liriglutide) on weight and glycemic control among adults with type 2 diabetes mellitus attending primary care center at security forces hospital in Riyadh, Saudi Arabia. J. Fam. Med. Prim. Care (2020) 9 3933-3936. DOI: 10.4103/jfmpc.jfmpc_361_20 30. Cena H., Chiovato L., Nappi R.E.. Obesity, Polycystic Ovary Syndrome, and Infertility: A New Avenue for GLP-1 Receptor Agonists. J. Clin. Endocrinol. Metab. (2020) 105 e2695-e2709. DOI: 10.1210/clinem/dgaa285 31. Sattar N., Lee M.M.Y., Kristensen S.L., Branch K.R.H., Del Prato S., Khurmi N.S., Lam C.S.P., Lopes R.D., McMurray J.J.V., Pratley R.E.. Cardiovascular, mortality, and kidney outcomes with GLP-1 receptor agonists in patients with type 2 diabetes: A systematic review and meta-analysis of randomised trials. Lancet Diabetes Endocrinol. (2021) 9 653-662. DOI: 10.1016/S2213-8587(21)00203-5 32. Ryan P.M., Seltzer S., Hayward N.E., Rodriguez D.A., Sless R.T., Hawkes C.P.. Safety and Efficacy of Glucagon-Like Peptide-1 Receptor Agonists in Children and Adolescents with Obesity: A Meta-Analysis. J. Pediatr. (2021) 236 137-147.e13. DOI: 10.1016/j.jpeds.2021.05.009 33. Guo N., Sun J., Chen H., Zhang H., Zhang Z., Cai D.. Liraglutide prevents diabetes progression in prediabetic OLETF rats. Endocr. J. (2013) 60 15-28. DOI: 10.1507/endocrj.EJ12-0094 34. Yousef C.C., Thomas A., Al Matar M., Ghandoura L., Aldossary I., Almuhanna S.M., Alhussain F., AL Bisher F.B., Aljohani R.M., Balubaid A.N.. Liraglutide effects on glycemic control and weight in patients with type 2 diabetes Mellitus: A real-world, observational study and brief narrative review. Diabetes Res. Clin. Pract. (2021) 177 108871. DOI: 10.1016/j.diabres.2021.108871 35. Alrowais S.S.. Relationship between exposure to Liraglutide and weight loss: A cross-sectional study in Riyadh, Saudi Arabia. Int. J. Clin. Exp. Med. (2021) 14 2435-2445 36. Guo X., Zhou Z., Lyu X., Xu H., Zhu H., Pan H., Wang L., Yang H., Gong F.. The Antiobesity Effect and Safety of GLP-1 Receptor Agonist in Overweight/Obese Patients Without Diabetes: A Systematic Review and Meta-Analysis. Horm. Metab. Res. (2022) 54 458-471. DOI: 10.1055/a-1844-1176 37. Thomas M.K., Nikooienejad A., Bray R., Cui X., Wilson J., Duffin K., Milicevic Z., Haupt A., Robins D.A.. Dual GIP and GLP-1 Receptor Agonist Tirzepatide Improves Beta-cell Function and Insulin Sensitivity in Type 2 Diabetes. J. Clin. Endocrinol. Metab. (2020) 106 388-396. DOI: 10.1210/clinem/dgaa863 38. Bethel M.A., Diaz R., Castellana N., Bhattacharya I., Gerstein H.C., Lakshmanan M.C.. HbA1c Change and Diabetic Retinopathy During GLP-1 Receptor Agonist Cardiovascular Outcome Trials: A Meta-analysis and Meta-regression. Diabetes Care (2020) 44 290-296. DOI: 10.2337/dc20-1815 39. Ryan D.H.. Next Generation Antiobesity Medications: Setmelanotide, Semaglutide, Tirzepatide and Bimagrumab: What do They Mean for Clinical Practice?. J. Obes. Metab. Syndr. (2021) 30 196-208. DOI: 10.7570/jomes21033 40. Karagiannis T., Avgerinos I., Liakos A., Del Prato S., Matthews D.R., Tsapas A., Bekiari E.. Management of type 2 diabetes with the dual GIP/GLP-1 receptor agonist tirzepatide: A systematic review and meta-analysis. Diabetologia (2022) 65 1251-1261. DOI: 10.1007/s00125-022-05715-4 41. Sun F., Wu S., Wang J., Guo S., Chai S., Yang Z., Li L., Zhang Y., Ji L., Zhan S.. Effect of Glucagon-like Peptide-1 Receptor Agonists on Lipid Profiles Among Type 2 Diabetes: A Systematic Review and Network Meta-analysis. Clin. Ther. (2015) 37 225-241.e8. DOI: 10.1016/j.clinthera.2014.11.008 42. Sun J., Cui J., He Q., Chen Z., Arvan P., Liu M.. Proinsulin misfolding and endoplasmic reticulum stress during the development and progression of diabetes☆. Mol. Asp. Med. (2015) 42 105-118. DOI: 10.1016/j.mam.2015.01.001 43. Ceperuelo-Mallafré V., Duran X., Pachón G., Roche K., Garrido-Sánchez L., Vilarrasa N., Tinahones F.J., Vicente V., Pujol J., Vendrell J.. Disruption of GIP/GIPR Axis in Human Adipose Tissue Is Linked to Obesity and Insulin Resistance. J. Clin. Endocrinol. Metab. (2014) 99 E908-E919. DOI: 10.1210/jc.2013-3350 44. Stern J.H., Rutkowski J.M., Scherer P.E.. Adiponectin, Leptin, and Fatty Acids in the Maintenance of Metabolic Homeostasis through Adipose Tissue Crosstalk. Cell Metab. (2016) 23 770-784. DOI: 10.1016/j.cmet.2016.04.011 45. Ma X., Liu Z., Ilyas I., Little P.J., Kamato D., Sahebka A., Chen Z., Luo S., Zheng X., Weng J.. GLP-1 receptor agonists (GLP-1RAs): Cardiovascular actions and therapeutic potential. Int. J. Biol. Sci. (2021) 17 2050-2068. DOI: 10.7150/ijbs.59965 46. Hartman M.L., Sanyal A.J., Loomba R., Wilson J.M., Nikooienejad A., Bray R., Karanikas C.A., Duffin K.L., Robins D.A., Haupt A.. Effects of Novel Dual GIP and GLP-1 Receptor Agonist Tirzepatide on Biomarkers of Nonalcoholic Steatohepatitis in Patients With Type 2 Diabetes. Diabetes Care (2020) 43 1352-1355. DOI: 10.2337/dc19-1892 47. Ohki T., Isogawa A., Iwamoto M., Ohsugi M., Yoshida H., Toda N., Tagawa K., Omata M., Koike K.. The Effectiveness of Liraglutide in Nonalcoholic Fatty Liver Disease Patients with Type 2 Diabetes Mellitus Compared to Sitagliptin and Pioglitazone. Sci. World J. (2012) 2012 496453. DOI: 10.1100/2012/496453 48. Rezaei S., Tabrizi R., Nowrouzi-Sohrabi P., Jalali M., Atkin S.L., Al-Rasadi K., Jamialahmadi T., Sahebkar A.. GLP-1 Receptor Agonist Effects on Lipid and Liver Profiles in Patients with Nonalcoholic Fatty Liver Disease: Systematic Review and Meta-Analysis. Can. J. Gastroenterol. Hepatol. (2021) 2021 8936865. DOI: 10.1155/2021/8936865 49. Aoki K., Kamiyama H., Takihata M., Taguri M., Shibata E., Shinoda K., Yoshii T., Nakajima S., Terauchi Y.. Effect of liraglutide on lipids in patients with type 2 diabetes: A pilot study. Endocr. J. (2020) 67 957-962. DOI: 10.1507/endocrj.EJ19-0464 50. Hasegawa Y., Hori M., Nakagami T., Harada-Shiba M., Uchigata Y.. Glucagon-like peptide-1 receptor agonists reduced the low-density lipoprotein cholesterol in Japanese patients with type 2 diabetes mellitus treated with statins. J. Clin. Lipidol. (2018) 12 62-69.e1. DOI: 10.1016/j.jacl.2017.11.006 51. Korayem G.B., Alshaya O.A., Alghamdi A.A., Alanazi S.S., Almutib R.T., Alsaileek M., Alrashidi A., Aldosari N., Bin Sheraim N., Al Yami M.S.. The prescribing pattern of sodium-glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists in patient with type two diabetes mellitus: A two-center retrospective cross-sectional study. Front. Public Health (2022) 10 4092. DOI: 10.3389/fpubh.2022.1031306
--- title: FGF-2 Differentially Regulates Lens Epithelial Cell Behaviour during TGF-β-Induced EMT authors: - Mary Flokis - Frank J. Lovicu journal: Cells year: 2023 pmcid: PMC10046997 doi: 10.3390/cells12060827 license: CC BY 4.0 --- # FGF-2 Differentially Regulates Lens Epithelial Cell Behaviour during TGF-β-Induced EMT ## Abstract Fibroblast growth factor (FGF) and transforming growth factor-beta (TGF-β) can regulate and/or dysregulate lens epithelial cell (LEC) behaviour, including proliferation, fibre differentiation, and epithelial–mesenchymal transition (EMT). Earlier studies have investigated the crosstalk between FGF and TGF-β in dictating lens cell fate, that appears to be dose dependent. Here, we tested the hypothesis that a fibre-differentiating dose of FGF differentially regulates the behaviour of lens epithelial cells undergoing TGF-β-induced EMT. Postnatal 21-day-old rat lens epithelial explants were treated with a fibre-differentiating dose of FGF-2 (200 ng/mL) and/or TGF-β2 (50 pg/mL) over a 7-day culture period. We compared central LECs (CLECs) and peripheral LECs (PLECs) using immunolabelling for changes in markers for EMT (α-SMA), lens fibre differentiation (β-crystallin), epithelial cell adhesion (β-catenin), and the cytoskeleton (alpha-tropomyosin), as well as Smad$\frac{2}{3}$- and MAPK/ERK$\frac{1}{2}$-signalling. Lens epithelial explants cotreated with FGF-2 and TGF-β2 exhibited a differential response, with CLECs undergoing EMT while PLECs favoured more of a lens fibre differentiation response, compared to the TGF-β-only-treated explants where all cells in the explants underwent EMT. The CLECs cotreated with FGF and TGF-β immunolabelled for α-SMA, with minimal β-crystallin, whereas the PLECs demonstrated strong β-crystallin reactivity and little α-SMA. Interestingly, compared to the TGF-β-only-treated explants, α-SMA was significantly decreased in the CLECs cotreated with FGF/TGF-β. Smad-dependent and independent signalling was increased in the FGF-2/TGF-β2 co-treated CLECs, that had a heightened number of cells with nuclear localisation of Smad$\frac{2}{3}$ compared to the PLECs, that in contrast had more pronounced ERK$\frac{1}{2}$-signalling over Smad$\frac{2}{3}$ activation. The current study has confirmed that FGF-2 is influential in differentially regulating the behaviour of LECs during TGF-β-induced EMT, leading to a heterogenous cell population, typical of that observed in the development of post-surgical, posterior capsular opacification (PCO). This highlights the cooperative relationship between FGF and TGF-β leading to lens pathology, providing a different perspective when considering preventative measures for controlling PCO. ## 1. Introduction The ocular lens is a transparent, avascular tissue responsible for transmitting light onto the retina. It contains two cell types: cuboidal epithelial cells and adjacent elongate fibre cells, both comprised of specialized molecular (e.g., crystallins) and cytoskeletal (e.g., intermediate filaments) properties to facilitate vision [1]. Ocular growth factors, such as fibroblast growth factor (FGF) and transforming growth factor-beta (TGF-β), are key regulators of different cellular processes in the lens, including epithelial cell proliferation [2,3,4], fibre differentiation [1,5,6,7,8,9,10], and epithelial–mesenchymal transition (EMT) that lead to lens pathology [11,12,13,14,15,16,17]. In situ, FGF is thought to be required for regulating normal lens cell processes in a spatially dependent manner, as previously reviewed [1]. TGF-β can regulate and/or concurrently dysregulate normal lens homeostasis, cell growth, and survival, by altering lens epithelial cell (LEC) morphology [17,18,19]. The dysregulation of lens epithelial cell architecture induced by TGF-β is characterized by EMT, a phenomenon that has been widely reviewed [20,21,22], with normal cuboidal LECs transitioning to become aberrant migratory, contractile myofibroblastic cells. These myofibroblastic cells can aggregate to form a fibrotic plaque leading to cataracts [23,24,25]. To date, cataracts, that have been extensively reviewed and studied, are still considered the most common form of blindness worldwide [26,27,28], with the only form of treatment being surgical intervention. Despite the effectiveness of surgery, posterior capsular opacification (PCO), known also as a secondary cataract, may result post surgery, requiring further intervention [29,30,31]. PCO results from the aberrant behaviour of LECs left after surgery, with these cells either undergoing EMT to form a posterior subcapsular plaque (fibrotic PCO) [11,24,25,32], or differentiating into aberrant fibre cells leading to Elschnig’s pearls and Soemmerring’s ring (regenerative, pearl PCO) [33], as previously reviewed [34,35]. While these two different spatially distinct epithelial PCO pathologies are well characterised [36,37], the underlying molecular mechanisms regarding their formation are poorly understood. Numerous models have established TGF-β-induced lens EMT responses in humans [14], embryonic chicks [13,38], and murine cell lines and explant models [13,15,25,32,39]. In dissociated embryonic chick lens epithelial cells treated with TGF-β, we see a heterogenous response, with some cells undergoing fibre differentiation, while others undergo EMT [13,40]. In in vitro studies using mammalian lens epithelia, exogenous treatment of LECs with TGF-β results in a homogenous EMT response [17,41,42,43]. In situ, however, anterior subcapsular cataracts (ASCs) develop in transgenic mice in response to elevated activity of ocular TGF-β [44]; the subcapsular plaques are comprised a heterogenous population of aberrant lens fibre cells and myofibroblastic cells, similar to those seen in human cataracts [24]. The in situ transgenic mouse model ideally replicates the human clinical pathology of fibrotic cataracts, that is attributed to the endogenous ocular milieu of different growth factors and cytokines, that do not act in isolation, unlike what we have in vitro. Since two disparate lens epithelial phenotypes contribute to ASC and PCO, it is important to better understand how they are derived, and the putative interplay of the different ocular factors involved. While FGF is well established in regulating lens epithelial cell proliferation and fibre differentiation, it has also previously been shown to influence TGF-β-induced EMT and aberrant cell behaviour, promoting wound healing, repair, and fibrogenesis [5,16,41,45]. For example, different relatively low doses of FGF-2 (2.5–20 ng/mL) can exacerbate TGF-β2 (0.5–3 ng/mL)-induced lens opacification in intact rat lenses, with the higher dose combinations exhibiting the most pronounced response, resulting in dense cellular plaques and elevated deposition of ECM [16]. In contrast, other studies have shown that FGF can counteract and antagonise EMT in rodent LECs [14,15]. Rat lens epithelial cell explants cotreated with a relatively low dose of TGF-β2 (50 pg/mL) and a lower dose of FGF-2 (10 ng/mL) still formed spindle-like cells typical of EMT; however, with minimal cell loss compared to explants treated with TGF-β2 alone [15]. This increased cell survival was unique to FGF as other regulatory ocular growth factors (e.g., EGF, IGF, HGF, or PDGF) could not block the hallmark features of TGF-β-induced EMT, including lens capsular wrinkling, apoptosis, and cell loss [15,46]. The influence of FGF regulating TGF-β-induced EMT may be attributed to the putative crosstalk between various downstream intracellular signalling pathways; the TGF-β-canonical Smad$\frac{2}{3}$-dependent proteins, and non-canonical mitogen-activated protein kinases (MAPK), such as extracellular signal-regulated kinase (ERK$\frac{1}{2}$) [14,38,47,48,49]. Studies using mouse LEC lines (MLECs) showed that cotreatment of cells with FGF-2 (10 ng/mL) and TGF-β2 (10 ng/mL) resulted in elongated fibroblastic-like cells and enhanced cell migratory mechanisms, with elevated ERK$\frac{1}{2}$-signalling [14]. Interestingly, in human lens epithelial cells (HLECs) from this same study, cotreated with the same doses of FGF-2/TGF-β2, they report on the antagonistic behaviour of FGF-2 with a reduction in cytoskeletal markers involved in stress fibre formation [14]. It is clear from these studies that there is no consistency in cell responsiveness to both FGF/TGF-β across different species. In the current study, we characterized the influence that FGF has on TGF-β-induced cell behaviour in rat lens explants to best model the conditions needed to promote a heterogenous cell population typical of fibrotic cataracts as seen in situ. We demonstrate that a high fibre-differentiating dose of FGF is protective of TGF-β-induced EMT in peripheral lens epithelia; however, this is not evident in central lens epithelia induced by TGF-β. This emulates the spatial phenotypic response of lens cells seen in human PCO and may serve as a model to better understand the mechanisms leading to this post-surgical pathology. ## 2.1. Animals and Tissue Culture All procedures conducted abided by the Australian Code for animal care and usage for scientific purposes and the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals for Ophthalmic and Vision biomedical research (USA). The experiments were approved by the Animal Ethics Committee of The University of Sydney, NSW, Australia (AEC# $\frac{2021}{1913}$). Wistar rats (rattus norvegicus) at 21-days of age (P21 ± 1 day) were humanely euthanized with CO2 followed by cervical dislocation. ## 2.2. Lens Epithelial Explants All collected primary rat ocular tissues were kept in medium 199 with Earle’s Salts (M199) (11825015, GibcoTM, Thermo Fisher Scientific, Sydney, NSW, Australia) in 35 mm Nunc™ culture dishes (NUN150460, Thermo Fisher Scientific). The media was supplemented with 2.5 μg/mL Amphotericin B (15290-018, GibcoTM, Thermo Fisher Scientific), $0.1\%$ bovine serum albumin (BSA) (9048-46-8, Sigma-Aldrich Corp., St. Louis, MO, USA), 0.1 μg/mL L-glutamine (200 mM) (25030081, GibcoTM, Life Technologies, Carlsbad, CA, USA), and penicillin (100 IU/mL)/streptomycin (100 μg/mL) (15140-122, GibcoTM, Life Technologies). The collected eyes were placed under a dissecting microscope to remove the lenses. The posterior capsule of the lens was torn using fine forceps and the remaining intact anterior capsule containing a sheet of lens epithelial cells (LECs) was pinned to the base of the culture dish using the gentle pressure of the forceps, as previously described [5]. Explants were maintained in a humidified incubator (37 °C, $5\%$ CO2). Different doses of recombinant human TGF-β2 (302-B2-002, R&D systems, Minneapolis, MN, USA) were used to induce EMT in the lens explants, as previously described [39]. A lower dose of TGF-β2 (50 pg/mL) gave a more regulated EMT response over 7 days, while a higher dose (200 pg/mL) was used to induce a more rapid EMT response in the lens explants over this same time period. To determine the impact of FGF-2 on TGF-β2-induced lens EMT, we cotreated TGF-β2-treated LECs with either a low proliferating dose of recombinant human FGF-2 (5 ng/mL: 233-FB, R&D systems) or a high fibre-differentiating dose of FGF-2 (200 ng/mL) [10,50]. Control explants had no growth factors added to the media. ## 2.3. Assessment of Cell Morphology and Immunofluorescence Cultured LEC explants were monitored and photographed daily over 7 days using phase contrast microscopy (Leica FireCam imaging, Leica Microsystems, Version 1.5, 2007). To examine the extent of how transdifferentiated cells modulated the underlying lens capsule, some treated explants were rinsed with filtered Milli-Q H2O to debride all cells from the explant to completely expose the underlying lens capsule. Phase contrast images were captured before and after rinsing. Following the different growth factor treatments, at set time points, the explants were fixed in $10\%$ neutral buffered formalin (NBF; HT501320-9.5L, Sigma-Aldrich Corp) for 10 min, followed by 3 × 5 min rinses in phosphate-buffered saline (PBS) supplemented with BSA ($0.1\%$, v/w; PBS/BSA). The cells were permeabilized using PBS/BSA supplemented with Tween-20 ($0.05\%$, v/v; 3 × 5 min), followed by subsequent rinses in PBS/BSA (2 × 5 min). The explants were then incubated at room temperature for 30 min with $3\%$ normal goat serum (NGS diluted in PBS/BSA, w/v), before adding the primary antibodies; anti-mouse α-SMA (A2547, Sigma-Aldrich Corp.), anti-alpha tropomyosin (Tpm; α/9d; provided by Prof. Gunning, University of New South Wales, Sydney, NSW, Australia), anti-rabbit β-catenin (ab6302, Abcam, Fremont, CA, USA), anti-β-crystallin, and anti-total-Smad$\frac{2}{3}$ (t-Smad$\frac{2}{3}$: 8685, Cell Signaling Tech., Danvers, MA, USA), all diluted 1:100 in NGS/PBS/BSA. The explants were incubated overnight at 4 °C, followed by rinsing in PBS/BSA (3 × 5 min). The respective secondary antibodies were then applied for a 2 h incubation at room temperature: goat anti-rabbit IgG Alexa-Fluor® 488 (ab150077, Abcam), and goat anti-mouse Alexa-Fluor® 594 (ab150116, Abcam), all diluted 1:1000 in PBS/BSA. Three 5 min rinses in PBS/BSA were followed before a 5 min application of 3 μg/mL bisbenzimide (H33342 trihydrochloride, Hoechst counterstain, B2261, Sigma-Aldrich) diluted in PBS/BSA. The explants were rinsed again before mounting with $10\%$ glycerol in PBS and imaged using epifluorescence microscopy (Leica DMLB 100S with DFC-450C camera, Leica Application Suite, Version 4.8, 2021). ## 2.4. SDS-Page and Western Blotting Cultured lens epithelial explants at set time points were rinsed in cold PBS. The central and peripheral regions of the explants were isolated using a scalpel blade to delineate each region. A central square of tissue, no more than a third of the explant diameter (central LECs, CLECs), and the remaining surrounding peripheral LECs (PLECs) were isolated separately. CLEC and PLEC protein was harvested, pooled into allocated Eppendorf tubes, and lysed with cold radioimmunoprecipitation assay (RIPA) lysis buffer containing 150 mM NaCl, $0.5\%$ sodium deoxycholate, $0.1\%$ Sodium dodecyl sulphate (SDS), 1 mM sodium orthovanadate, 1 mM NaF, 50 mM Tris-HCl (pH 7.5), $0.1\%$ Triton X-100, phosphatase (PhosSTOPTM), and protease (cOmpleteTM) EASYpacks inhibitor tablets (04906837001 and 05892970001; Roche Applied Science, Basel, Switzerland). LECs were homogenized and centrifuged for 10 min at 4 °C (14,400× g) for lysate/supernatant separation. Quantification of the total lens protein of each supernatant sample was conducted using a PierceTM Micro bicinchoninic acid (BCA) protein assay reagent kit (23235; Thermo Fisher Scientific). LEC protein sample lysates were prepared using $5\%$ 2-mercaptoethanol (M6250, Sigma-Aldrich) combined with 2× Laemmli sample buffer at a 1:1 (v/v) ratio (1610737, Bio-Rad Laboratories, NSW, Australia). For electrophoresis, 10 μg of protein lysates were loaded onto $12\%$ SDS-PAGE gels for 20 min at 70 V followed by 2 h at 120 V. LEC protein was then transferred onto an immobilon®-PSQ polyvinylidene fluoride (PVDF) membrane (ISEQ00010, Merck Millipore, Rahway, NJ, USA) for 1 h at 100 V. PVDF membranes were incubated in $2.5\%$ BSA blocking buffer diluted in tris-buffered saline with $0.1\%$ Tween-20 (TBST) and incubated for 1 h with agitation at room temperature. Primary antibodies were added to the membranes and left overnight to incubate (at 4 °C): anti-mouse α-SMA, anti-GAPDH (G8795, Sigma-Aldrich Corp.), anti-tropomyosin alpha, and anti-β-crystallin, t-Smad$\frac{2}{3}$, phospho-Smad$\frac{2}{3}$ (p-Smad$\frac{2}{3}$, 8828, Cell Signalling Tech., Danvers, MA, USA), phospho-ERK$\frac{1}{2}$ (p-ERK$\frac{1}{2}$, 9101, Cell Signalling Tech.), and total-ERK$\frac{1}{2}$ (t-ERK$\frac{1}{2}$, 9102, Cell Signalling Tech.), all diluted in blocking buffer/TBST at 1:1000, apart from α-SMA and GAPDH (1:2000). Following overnight incubation, the membranes were rinsed in TBST (3 × 5 min) and incubated with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies for 2 h at room temperature: HRP-conjugated goat anti-rabbit IgG (7074, Cell Signalling Tech.) and horse anti-mouse IgG (7076, Cell Signalling Tech.), diluted in TBST at 1:5000. The membranes were rinsed in TBST (3 × 10 min) followed by the application of an immobilon chemiluminescent HRP substrate for 3–5 min (WBKLS0500, Merck Millipore). Protein chemiluminescent signals were imaged using Bio-Rad ChemiDocTM MP imaging. Following immunolabeling, PVDF membranes were stripped for 10 min in stripping buffer ($10\%$ SDS, 0.5 M Tris HCl pH 6.8, Milli-Q H2O, and $0.8\%$ 2-mercaptoethanol) with gentle agitation. The membranes were then washed in TBST (3 × 5 min) and re-blocked in blocking buffer/TBST for 1 h. Following blocking, the membranes were probed for loading control GAPDH (1:2000, 1 h) and incubated with an HRP-conjugated horse anti-mouse secondary antibody for 1 h prior to chemiluminescent signalling analysis. Protein densitometry was carried out using Bio-Rad ImageLab software (Version 6.1.0, 2019). ## 2.5. Statistical Analysis For each experimental analysis, three independent experiments were carried out. For every experiment, a minimum of three individual replicates ($$n = 3$$) per treatment group (different treatment of explants) were used. For Western blotting, each group contained up to eight explants to isolate central and peripheral lens cells that were randomly obtained from different P21 rats. For measuring changes in protein expression, we used densitometry to calculate the selected protein intensity relative to the loading control (GAPDH). For Western blot experiments examining differences in Smad-dependent (Smad$\frac{2}{3}$) and Smad-independent (MAPK/ERK$\frac{1}{2}$) activity, phosphorylated protein expression was calculated using the following ratio: relative phosphorylated density per total protein. Prior to the use of one-way analysis of variance (ANOVA), several assumptions were tested and confirmed; we assumed equal standard deviation (SD) and residuals appeared normally distributed. Based on these confirmed assumptions, we compared the differences among the means of all treatment groups using one-way ANOVA, followed by Tukey’s multiple comparisons post-hoc test. All data acquired were plotted appropriately using GraphPad Prism software version 9.0 (GraphPad Software Inc., San Diego, CA, USA). To quantify the spatial differences in t-Smad$\frac{2}{3}$ immunoreactivity, six separate images of central and peripheral regions were captured per explant across three randomized explants per treatment group, over three individual experiments. Nuclear and cytoplasmic localisation of t-Smad$\frac{2}{3}$ was manually counted using ImageJ’s Cell Counter plugin. The mean percentage of t-Smad$\frac{2}{3}$ nuclear and cytoplasmic fluorescence was calculated and statistically analysed using GraphPad Prism. Tabled data were represented as the standard error of the mean (±SEM) and probability values, where $p \leq 0.05$ was considered statistically significant. ## 3.1. FGF-2 Promotes a Spatially Dependent TGF-β2-Induced EMT Response in Lens Epithelial Explants We examined the efficacy of different doses of FGF-2 in modulating the effect of TGF-β2 on lens epithelial cells induced to undergo EMT. Using phase contrast microscopy, control LECs without FGF-2 or TGF-β2 treatment (Figure 1A,E), as well as explants treated with only a low dose of FGF-2, demonstrated no significant morphological changes and retained their epithelial phenotype over the culture period. When the lens epithelial explants were treated with a low dose of TGF-β2 (50 pg/mL), this promoted an EMT response across the entire explant, similar to a higher dose (200 pg/mL) of TGF-β2, albeit at a slower rate, consistent with previous studies [42]. With different dose combinations of FGF-2 and TGF-β2, most cells in the explants underwent a uniform EMT response over 5 days, with the exception of cells in the explants cotreated with a relatively high dose of FGF-2 (200 ng/mL) and the lower dose of TGF-β2 (Figure 1), where we observed a differential response between CLECs and PLECs (Table 1, Supplementary Figure S1). The cells in the lens epithelial explants treated with the high a fibre-differentiating dose of FGF-2 elongated over 5 days (Figure 1B,F), compared to the control LEC explants (no growth factor treatment; Figure 1A,E). This FGF-induced fibre differentiation response was more pronounced in PLECs compared to CLECs (Figure 1B,F). LECs in explants treated with a low dose of TGF-β2 (Figure 1C,G) displayed prominent EMT by day 5, with the LECs losing their uniform packing and adhesion as they transdifferentiated into myofibroblastic cells. TGF-β2 treatment also led to increased cellular blebbing (refractile bodies) and apoptotic cell loss, evident by areas of bare lens capsule that displayed prominent signs of capsular modification in the form of wrinkling throughout the explant. When the explants were cotreated with TGF-β2 and FGF-2, CLECs underwent similar morphological transformations by day 5 (Figure 1D) to that seen with TGF-β2-treatment alone (Figure 1C,G). In contrast, PLECs in the explants cotreated with FGF-2/TGF-β2 showed no evidence of EMT (Figure 1H), instead demonstrating morphological changes more consistent with that observed in the explants treated with FGF-2 alone (Figure 1B,F). PLECs in cotreated explants exhibited changes in the LEC phenotype as early as day 3 (Supplementary Figure S2H). Debridement of all cells at this time revealed the underlying lens capsule with no apparent capsular modulation (no folds or wrinkles) in either control (Supplementary Figure S3A,D) or FGF-2 treated (Supplementary Figure S3B,E) explants. In the TGF-β2-treated explants, increased capsular modulation was apparent with wrinkling and folds in the central explant regions (Supplementary Figure S3C) and in the peripheral regions (Supplementary Figure S3F). Consistent with the differential cell response in the central and peripheral regions of the F/t-cotreated explants (Figure 2A,A1,A3), the explants exhibited capsular modulation only in the central explant region (Figure 2A2), with no capsular wrinkling observed in the peripheral region (Figure 2A4). We observed that with ongoing culture (up to 7 days), regardless of the explant region or treatment, all of the cells exposed to TGF-β2 (200 pg/mL) are lost by 7 days (Supplementary Figure S4A); however, in the cotreated explants (F/t), with continual supplementation of the media with FGF-2 (200 ng/mL) after day 3 of culture, this promoted cell survival, whereby we continue to observe many myofibroblastic cells in the central region of the explants (Supplementary Figure S4B) and, similarly, relatively normal lens cells at the periphery of the explants are also maintained (Supplementary Figure S4C). ## 3.2.1. Immunofluorescent Labelling We used immunofluorescence to characterise the different cell types in explants treated with TGF-β2 and/or FGF-2 over 3 days, labeling for lens fibre differentiation markers, β-crystallin, and alpha-tropomyosin (α/9d), as well as the EMT marker, α-SMA (Figure 3). Isotype controls for all of the antibodies show little to no specific labelling. Control LECs throughout the explant exhibited no reactivity for β-crystallin and/or α-SMA after 3 days of culture (Figure 3A), labelling only for α/9d (Figure 3B). FGF-2-treated LECs exhibited strong reactivity for β-crystallin throughout the explant (Figure 3C,I), with stronger labelling in PLECs (Figure 3I). Treatment with FGF-2 did not promote α-SMA reactivity in any cultured lens epithelia. FGF-2-treated CLECs presented diffuse α/9d-reactivity (Figure 3F), while PLECs had a more defined reactivity for α/9d, highlighting actin filaments in the elongating, differentiating fibre cells (Figure 3L). LECs treated with only TGF-β2 displayed clear evidence of an EMT response, with strong reactivity for α-SMA, with no β-crystallin observed throughout the explant (Figure 3D,J). TGF-β2-treated CLECs had a highly specific localisation of α/9d to actin stress fibres (Figure 3G), which were also very prominent in PLECs (Figure 3M). Unlike cells treated with only FGF-2 or only TGF-β2, that had a relatively uniform label for the different markers across the entire explant, in the FGF-2/TGF-β2 cotreated explants, we observed distinct spatial differences in the labelling of the markers, consistent with our earlier morphological observations. The CLECs in the TGF-β2/FGF-2-treated explants predominantly labelled for α-SMA with little to no β-crystallin reactivity at day 3 (Figure 3E), similar to the explants treated with only TGF-β2 (Figure 3D,J). In contrast, the PLECs in these same cotreated explants displayed the inverse label, with strong reactivity primarily for β-crystallin in elongated cells, with few neighboring smaller cells immunolabelling for α-SMA (Figure 3K). This differential β-crystallin and α-SMA reactivity in the cotreated explants was sustained up to 5 days of culture (Supplementary Figure S5). Stronger labelling for α/9d was also observed throughout the cotreated explants (Figure 3H,N), highlighting the marked elongation of peripheral fibre-like cells (Figure 3K,N), as well as central myofibroblastic cells (Figure 3E,H). ## 3.2.2. Western Blotting Alpha-Smooth Muscle Actin. We quantified protein changes in the treated explants using Western blotting. CLECs had a significant increase in α-SMA when treated with TGF-β2, compared to the relatively lower levels in the control (NT) and FGF-2-treated explants ($p \leq 0.0001$) (Figure 4A,C). FGF-2 treatment did not impact α-SMA levels in the CLECs compared to the control cells ($$p \leq 0.3429$$). In the FGF-2/TGF-β2 cotreated explants, there was a significant reduction in α-SMA levels in the CLECs relative to the TGF-β2 alone CLECs ($p \leq 0.0001$). In fact, these CLECs in the cotreated explants displayed no significant difference in levels of α-SMA compared to the CLECs of the control ($p \leq 0.9999$) or FGF-2 alone ($$p \leq 0.3249$$) explants. In the PLECs of the FGF-2/TGF-β2 cotreated explants, consistent with the reduced EMT response, there were reduced α-SMA levels when compared to the CLECs, comparable to the lower α-SMA levels seen in the PLECs of the control ($$p \leq 0.7371$$), FGF-2 ($p \leq 0.9999$)-, and TGF-β2-treated explants ($$p \leq 0.0053$$) (Figure 4B,D). The PLECs in the explants treated with FGF-2 alone did not have increased α-SMA levels when compared to the control cells ($$p \leq 7366$$); however, the PLECs in the explants treated with TGF-β2 alone had significantly increased α-SMA levels, compared to the control ($$p \leq 0.0203$$) and the FGF-2-treated ($$p \leq 0.0053$$) explants. β-crystallin. When compared to the control cells, there was no significant difference in the levels of β-crystallin in the CLECs of the explants treated with FGF-2 ($$p \leq 0.8742$$) (Figure 4A,C). Treatment with TGF-β2 did not significantly increase levels of β-crystallin in the CLECs compared to the control ($$p \leq 0.5844$$), FGF-2 ($$p \leq 0.2260$$) or the cotreated explants ($$p \leq 0.1459$$). We did observe a significant decrease in β-crystallin in the CLECs of the cotreated explants, relative to the control ($$p \leq 0.0160$$) and FGF-2 ($$p \leq 0.0044$$)-treated explants (Figure 4A,C). Treatment of the explants with FGF-2 significantly increased β-crystallin levels in the PLECs when compared to the PLECs of the control ($$p \leq 0.0374$$) and the TGF-β2-treated explants ($$p \leq 0.0003$$) (Figure 4B,D). The PLECs in the explants treated with TGF-β2 alone had reduced β-crystallin levels when compared to the control PLECs ($$p \leq 0.0222$$). The PLECs of the explants cotreated with FGF-2/TGF-β2 had slightly elevated β-crystallin levels in comparison to the PLECs of the control ($$p \leq 0.6396$$) or the TGF-β2-treated ($$p \leq 0.0056$$) explants (Figure 4B,D). Alpha-Tropomyosin. α/9d levels were significantly elevated only in the CLECs and PLECs of the TGF-β2-treated explants, when compared to the corresponding cells of all other treatment groups (Figure 4A–D). For the CLECs, levels of α/9d in the control explants were reduced in both the FGF-2 ($$p \leq 0.0924$$)- and FGF-2/TGF-β2-treated explants ($$p \leq 0.0959$$) and were significantly reduced when compared to the elevated α/9d levels found in the CLECs of the TGF-β2-treated explants (control vs. TGF-β2, $$p \leq 0.0034$$; FGF-2 vs. TGF-β2, $$p \leq 0.0002$$) (Figure 4C). In the PLECs, there was no obvious difference in the levels of α/9d across all of the treatment groups (Figure 4B), except for elevated levels in the PLECs of the TGF-β2-treated explants as mentioned (control vs. TGF-β2, $$p \leq 0.0103$$; FGF-2 vs. TGF-β2, $$p \leq 0.0039$$; TGF-β2 vs. FGF-2/TGF-β2, $$p \leq 0.0249$$) (Figure 4D). ## 3.3.1. Nuclear Translocation of Smad2/3 Given that FGF-2 can differentially regulate TGF-β2-mediated LEC behaviour, we tested its impact on TGF-β2 mediated Smad$\frac{2}{3}$-signalling. Active TGF-β2-signalling is evident with the nuclear translocation of phosphorylated Smad$\frac{2}{3}$ (Figure 5). After 2 h of culture, in both the control (Figure 5A,E) and the FGF-2 (Figure 5B,F)-treated explants, we do not see any Smad$\frac{2}{3}$ nuclear localisation: 0.45–$2.81\%$ nuclear labelling (Table 2, Figure 5I,J). In contrast, distinct nuclear localisation of Smad$\frac{2}{3}$ was evident throughout the TGF-β2-treated explants (Figure 5C,G): 86.44–$88.8\%$ nuclear labelling. In the lens epithelial explants cotreated with FGF-2/TGF-β2, we observed prominent nuclear translocation of Smad$\frac{2}{3}$ in the CLECs (Figure 5D,I): $44.87\%$ nuclear labelling; however, in the PLECs the Smad$\frac{2}{3}$-labelling was primarily cytosolic with significantly reduced nuclear labelling: $19.85\%$ (Table 2, Figure 5H,J). ## 3.3.2. Smad2/3-Signalling Treatment of the explants with FGF-2 did not impact p-Smad$\frac{2}{3}$ levels in CLECs when compared to similar levels in the control CLECs ($$p \leq 0.9768$$, Figure 6A,B) or the PLECs ($$p \leq 0.9310$$, Figure 6D,E) after 6 h of culture. Consistent with our immunofluorescent nuclear localisation of Smad$\frac{2}{3}$ (Figure 5), TGF-β2 significantly elevated p-Smad$\frac{2}{3}$ levels in the CLECs compared to the CLECs of the control explants ($$p \leq 0.0061$$) and the FGF-2-treated explants ($$p \leq 0.0101$$) (Figure 6A,B). In the CLECs of the explants cotreated with FGF/TGF-β2, there was no significant difference in p-Smad$\frac{2}{3}$ levels when compared to the CLECs of the TGF-β2 ($$p \leq 0.5944$$)- and the FGF-2-treated explants ($$p \leq 0.0585$$); however, p-Smad$\frac{2}{3}$ levels significantly increased in the cotreated CLECs compared to the control explants ($$p \leq 0.0334$$) (Figure 6A,B). In the PLECs, the TGF-β2 treated explants exhibited elevated p-Smad$\frac{2}{3}$ levels in comparison to the control ($$p \leq 0.0178$$), FGF-2 alone ($$p \leq 0.0403$$), and cotreated PLEC explants ($$p \leq 0.6496$$, Figure 5D,E) (Figure 6D,E). Compared to the control- and FGF-2-treated PLEC explants, cotreatment with FGF-2/TGF-β2 increased p-Smad$\frac{2}{3}$ levels ($$p \leq 0.0933$$ for the control, $$p \leq 0.2122$$ for FGF-2) (Figure 6D,E). ## 3.3.3. MAPK/ERK1/2-Signalling Levels of phosphorylated ERK$\frac{1}{2}$ (p-ERK$\frac{1}{2}$) remained constant in the CLECs of the control and FGF-2-treated ($$p \leq 0.7703$$, Figure 6A,C) explants after 6 h but were elevated in the PLECs of the FGF-2-treated explants, compared to the control PLECs ($$p \leq 0.0140$$, Figure 6D,F). TGF-β2 treatment of the explants slightly increased p-ERK$\frac{1}{2}$ activity in the CLECs compared to the levels in the CLECs of the control ($$p \leq 0.0227$$) and FGF-2 treated explants ($$p \leq 0.0880$$) (Figure 6A,C). In contrast, the PLECs of the TGF-β2-treated explants had decreased p-ERK$\frac{1}{2}$ levels compared to the PLECs of the control ($$p \leq 0.6711$$) and FGF-2-treated explants ($$p \leq 0.0033$$) (Figure 6D,F). The CLECs in the explants cotreated with FGF-2/TGF-β2 had reduced p-ERK$\frac{1}{2}$ levels in comparison to the CLECs in the TGF-β2-treated ($$p \leq 0.0174$$), FGF-2-treated ($$p \leq 0.6641$$), and control explants ($$p \leq 0.9972$$) (Figure 6A,C). Interestingly, the PLECs of the cotreated explants demonstrated a significant increase in their p-ERK$\frac{1}{2}$ levels in contrast to the low levels in the PLECs of the control ($$p \leq 0.0195$$) and TGF- β2 treated explants ($$p \leq 0.0242$$) (Figure 6D,F). ## 4. Discussion The present study has demonstrated the impact of FGF-2 on the behaviour of lens epithelial cells induced to undergo EMT in response to TGF-β. A lens-fibre-differentiating dose of FGF-2 was able to block TGF-β2-induced lens EMT in only the peripheral LECs in explants (equivalent to the germinative region of the intact lens) and not in the central (more anterior) lens epithelia. As seen in previous wholemount rat lens epithelial cell explant models, we have demonstrated that CLECs and PLECs exposed to TGF-β2 alone undergo an EMT response, with no evidence of lens fibre differentiation [24,41,51]; however, in combination with FGF-2, FGF-2 potentiates this TGF-β2-induced activity, with elevation of canonical Smad$\frac{2}{3}$ signalling activity, as well as EMT-associated markers, more so in the CLECs. For our lens epithelial explant model, we used relatively low doses of TGF-β2 (50 and 200 pg/mL) to induce an EMT response across a short culture period [48,52]. This dose is physiologically representative of concentrations of TGF-β2 in its mature (approx. 100 pg/mL) and total (>3000 pg/mL) forms observed in situ [53]. In addition, it is comparable to active forms of TGF-β2 (approx. 100–400 pg/mL) found in cataractous patients [53,54,55,56,57]. Our use of a lower TGF-β2 dose contrasts to other studies that have used much higher doses (0.5–1.5 ng/mL) to elicit an EMT response in rodent lens cells [13,14,16], which could potentially lead to off-target growth factor signalling activity. Exogenous addition of FGF-2 at a high dose encourages all lens epithelial cells (both CLECs and PLECs) to undergo a change in cell morphology typical of fibre differentiation [3,4,58,59]. In explants cotreated with TGF-β2 and FGF-2, FGF-2 appeared to protect PLECs from TGF-β-induced EMT by promoting a fibre differentiation response. We showed that the PLECs in these FGF-2/TGF-β2 cotreated explants had prominent elongated fibres, reminiscent of many earlier studies from our laboratory [59]. An elevated dose of TGF-β2, was able to prevent any fibre differentiation in the PLECs, leading to an enhanced EMT response in both central and peripheral cells. In addition, we demonstrated that the PLECs in the cotreated explants did not exhibit contractile properties as evidenced by the lack of capsular wrinkling in this region, unlike the region of the CLECs undergoing EMT. The inhibition of lens epithelial cell contraction by FGF despite the presence of TGF-β has been shown in other fibrotic models to be dose dependent, such as in bovine LECs cultured in collagen I gel [60] and valvular interstitial cells (VICs) modelling valvular fibrosis [61], which is also correlated with reduced α-SMA expression. We not only report how TGF-β can impact FGF-induced lens cell responsiveness but how FGF in turn influences TGF-β-induced responses in LECs, the main focus of our study. When we examined for changes in cytoskeletal and stress-fibre associated proteins, the CLECs in TGF-β/FGF-cotreated lens explants exhibited predominant α-SMA localisation and little to no β-crystallin, suggesting that these cells cannot resist the EMT process, despite the presence of a high differentiating dose of FGF-2. The co-influence of FGF-2 and TGF-β on fibre differentiation, epithelial, and EMT-associated marker expression has previously been reported in other models, including human lung epithelial cells and rat alveolar epithelial-like cells [62], as well as E10 chick lens epithelial cells [13], and the lenses of postnatal mice [10]. Despite the CLECs in the TGF-β/FGF-cotreated explants undergoing a prominent EMT response, we noted reduced α-SMA and α/9d levels, compared to the TGF-β-alone-treated explants, suggesting that FGF-2 may be potentially compromising Tpm activity (a recruiter for actin assembly) and attenuating α-SMA stress fibre association. It has been previously shown that FGF may influence Tpm activity and expression, as well as cell biomechanics, in the presence of TGF-β [14]. For example, when murine LECs (MLECs) are cotreated with FGF-2/TGF-β2, the loss of Tpm1 corresponded with decreased α-SMA reactivity [14]. This same study also confirmed FGF-2 modulation of Tpm in HLECs, when cotreated with TGF-β2, with a significant reduction in both Tpm1 and Tpm2 levels [14]. We localized Tpm (α/9d) in LECs undergoing different phenotypic changes in fibre differentiation, but more compellingly in cells undergoing EMT, where it was associated with the α-SMA-reactive stress fibres of myofibroblasts. This may be attributed to the fact that the α/9d antibody we used specifically targets several isoform splice variant products of the αTm gene (TPM1), including Tpm1.4, Tpm1.6–1.9, and Tpm2.1, with some cross-reactivity also for Tpm3.1 [63]. Tpm1.6, Tpm2.1, and Tpm3.1 have all previously been characterized as being stress-fibre associated and are suggested to play a role in TGF-β induced EMT [64,65,66]. FGF has shown a role in propagating stress-induced EMT in conjunction with TGF-β in other pathologies, such as wound healing in mice skin keratinocytes [45] and in the tumor stromal cell microenvironment of prostate fibroblasts [47]. Consistent with our findings, Koike et al. [ 2020] [45] found that FGF-2 could not solely induce EMT in mice keratinocytes; however, in keratinocytes cotreated with FGF-2 and TGF-β1, there was a significant upregulation of cell migratory/motility and EMT-associated genes (e.g., VIM and SNAI2), similar to keratinocytes with only TGF-β1 stimulation. In a non-transformed mouse mammary gland epithelial cell line (NMuMG), TGF-β modulated FGF receptor activation, increased FGF-2 cell sensitivity, and promoted an EMT response through activation of ERK$\frac{1}{2}$ signalling [49], highlighting the synergistic signalling role of these two growth factors. To determine how FGF-2 was modulating and antagonizing TGF-β2-induced EMT in PLECs of cotreated LEC explants, we explored changes in their signalling activity, namely changes to Smad$\frac{2}{3}$ and ERK$\frac{1}{2.}$ In cotreated lens epithelial explants, we saw stronger signalling for the respective pathways in different regions; CLECs undergoing EMT had more pronounced p-Smad$\frac{2}{3}$ activity, while PLECs undergoing fiber differentiation had more pronounced p-ERK$\frac{1}{2}$-signalling. FGF is a well-known regulator of ERK$\frac{1}{2}$ within the lens, with its marked phosphorylation evident in lens cells within minutes post treatment [3,4,67,68]. While ERK$\frac{1}{2}$ has been shown to be required for lens epithelial cell proliferation, it is also very important for lens fiber differentiation [4,67,68,69,70]. This differs from TGF-β2-induced EMT, where we found that while ERK is also involved in this EMT process, blocking ERK$\frac{1}{2}$ does not completely block TGF-β2-mediated EMT progression in lens epithelia [39,48,52]. In fact, canonical Smad$\frac{2}{3}$-signalling is most evident in EMT, as shown here in our cotreated CLECs, and in many earlier studies examining TGF-β2-induced lens EMT [11,13,38,52,71,72]. While we and others have shown FGF-2 is not able to promote Smad$\frac{2}{3}$-signalling in LECs [10], FGF-2 was shown to impede nuclear localisation of Smad$\frac{2}{3}$ in PLECs in explants cotreated with TGF-β2; however, in CLECs of these same explants, FGF-2 appeared to have less of an impact on TGF-β2-induced Smad$\frac{2}{3}$-activity. How FGF-2 directly blocks Smad$\frac{2}{3}$ activity in PLECs is not clear but given the strong ERK$\frac{1}{2}$ activation in these cells, this may favour lens fibre differentiation and cell survival, as we see here and has been shown by others [2,3,67,73]. Conversely, FGF-mediated ERK$\frac{1}{2}$ signalling can correlate with the upregulation of TGF-β activity, as seen in other fibrosis models [14,49,74,75,76], as well as the current study where TGF-β-induced CLECs are associated with elevated ERK$\frac{1}{2}$-signalling. Similar to the current study, in valvular interstitial cells (VICs) modelling valvular fibrosis, it was shown that inhibition of this fibrosis was dependent on FGF-2-mediated MAPK signalling when cotreated with TGF-β1 [61]. This study demonstrated that FGF (10 ng/mL) prevented Smad3 nuclear localisation in VICs cotreated with TGF-β1 (5 ng/mL), and at higher doses (100 ng/mL), it was able to perturb TGF-β1-mediated α-SMA expression [61], highlighting the ability of FGF to modulate canonical TGF-β signalling activity and downstream gene expression. Although not completely understood, crosstalk between FGF and TGF-β signalling has proven influential in mediating various fibrotic disorders and carcinoma progression. For example, a study implementing mouse tumor-associated endothelial cells (TECs) demonstrated how FGF can promote a differential cell response by reducing TGF-β-induced contractile and myofibroblastic properties, while concurrently promoting cell proliferation and motility [75]. A similar finding was observed in primary human dermal fibroblasts (HDFs), whereby FGF-2 with TGF-β1 cotreatment, both positively and negatively regulated fibroblast transition into cancer-associated fibroblasts (CAFs) [77]. This same study also showed how this FGF-2/TGF-β1 treatment of HDFs can downregulate common CAF-activated and EMT-associated markers (e.g., ACTA2, ITGA11, and COL1A1) as well as upregulating cell motility and morphogenetic genes (e.g., HGF and BMP2) [77]. Research into the mechanisms surrounding differential types of PCO involving lens fibre cell types is ongoing and it is believed to be due to FGF/TGF-β interactions during EMT induction [13,15,16,38,74,78]. As FGF is a major factor influencing normal lens fibre differentiation, it is important to understand what promotes aberrant fibre differentiation during pearl PCO development at the lens equator [13,24,35,79]. In situ, for ASC and for post-operative PCO, more anterior lens epithelial cells are likely exposed to a high insult of TGF-β, and relatively low levels of FGF are normally found in the aqueous humour. At the lens equator, however, epithelial cells in the posterior chamber are regularly exposed to elevated levels of FGF, and regardless of any increased TGF-β levels, the cells here likely undergo aberrant fibre differentiation, leading to pearl PCO. This may result from the heightened sensitivity to FGF of these peripheral LECs, namely due to their elevated levels of high-affinity FGF receptor tyrosine kinase (RTK) receptors, compared to the central lens epithelia [6,58,80,81]. In situ, during lens fibrosis, we do not see EMT resulting in cell death, likely due to survival growth factors present within the ocular media. Given the findings from the current study, we propose that FGF is a putative survival factor in situ, maintaining fiber cells at the lens equator and the myofibroblastic phenotype leading to fibrotic PCO. Further studies investigating differences/changes in levels of FGF and TGF-β receptors, between central and peripheral lens cells in cotreated explants, may be a key factor in determining lens cell fates in situ. We also cannot rule out that changes in the expression of RTK antagonists, such as Sprouty and Spreds [69,70,82], including those more specific for FGF, such as Sef [83], in these active regions of the lens may be protective of peripheral LECs from any aberrant TGF-β insult of which they have previously been reported to block [69]. ## 5. Conclusions A fine balance between levels of FGF-2 and TGF-β2 can promote differential responses in lens epithelial cells. More specifically, this responsiveness is spatially regulated, with the anterior central lens epithelia more sensitive to EMT induction, whilst peripheral cells primarily undergo fibre differentiation in the presence of high levels of FGF, avoiding apoptotic cell death associated with EMT. This induced heterogeneous population of cells in lens epithelial explants may provide an alternative model better suited to the study of the cellular processes at play in situ, leading to the formation of ASC, and more importantly both fibrotic and pearl forms of PCO. ## References 1. Lovicu F.J., McAvoy J.W.. Growth factor regulation of lens development. Dev. Biol. (2005) 280 1-14. DOI: 10.1016/j.ydbio.2005.01.020 2. Iyengar L., Patkunanathan B., Mcavoy J.W., Lovicu F.J.. Growth factors involved in aqueous humour-induced lens cell proliferation. Growth Factors (2009) 27 50-62. DOI: 10.1080/08977190802610916 3. Iyengar L., Wang Q., Rasko J.E.J., McAvoy J.W., Lovicu F.J.. Duration of ERK1/2 phosphorylation induced by FGF or ocular media determines lens cell fate. Differentiation (2007) 75 662-668. DOI: 10.1111/j.1432-0436.2007.00167.x 4. Lovicu F.J., McAvoy J.W.. FGF-induced lens cell proliferation and differentiation is dependent on MAPK (ERK1/2) signalling. Development (2001) 128 5075-5084. DOI: 10.1242/dev.128.24.5075 5. West-Mays J.A., Pino G., Lovicu F.J.. Development and use of the lens epithelial explant system to study lens differentiation and cataractogenesis. Prog. Retin. Eye Res. (2009) 29 135-143. DOI: 10.1016/j.preteyeres.2009.12.001 6. De Iongh R.U., Lovicu F.J., Hanneken A., Baird A., McAvoy J.W.. FGF receptor-1 (flg) expression is correlated with fibre differentiation during rat lens morphogenesis and growth. Dev. Dyn. (1996) 206 412-426. DOI: 10.1002/(SICI)1097-0177(199608)206:4<412::AID-AJA7>3.0.CO;2-L 7. Chamberlain C.G., McAvoy J.W.. Fibre differentiation and polarity in the mammalian lens: A key role for FGF. Prog. Retin. Eye Res. (1997) 16 443-478. DOI: 10.1016/S1350-9462(96)00034-1 8. Zhao H., Yang T., Madakashira B.P., Thiels C.A., Bechtle C.A., Garcia C.M., Zhang H., Yu K., Ornitz D.M., Beebe D.C.. Fibroblast growth factor receptor signaling is essential for lens fiber cell differentiation. Dev. Biol. (2008) 318 276-288. DOI: 10.1016/j.ydbio.2008.03.028 9. Chamberlain C.G., McAvoy J.W.. Induction of Lens Fibre Differentiation by Acidic and Basic Fibroblast Growth Factor (FGF). Growth Factors (1989) 1 125-134. DOI: 10.3109/08977198909029122 10. Wang D., Wang E., Liu K., Xia C.-H., Li S., Gong X.. Roles of TGFβ and FGF signals during growth and differentiation of mouse lens epithelial cell in vitro. Sci. Rep. (2017) 7 7274. DOI: 10.1038/s41598-017-07619-5 11. Wormstone I.M., Anderson I.K., Eldred J.A., Dawes L.J., Duncan G.. Short-term exposure to transforming growth factor β induces long-term fibrotic responses. Exp. Eye Res. (2006) 83 1238-1245. DOI: 10.1016/j.exer.2006.06.013 12. Wormstone I.M., Tamiya S., Anderson I., Duncan G.. TGF-beta2-induced matrix modification and cell transdif-ferentiation in the human lens capsular bag. Investig. Ophth. Vis. Sci. (2002) 43 2301-2308 13. Boswell B.A., Korol A., West-Mays J.A., Musil L.S.. Dual function of TGFβ in lens epithelial cell fate: Implications for secondary cataract. Mol. Biol. Cell (2017) 28 907-921. DOI: 10.1091/mbc.e16-12-0865 14. Kubo E., Shibata S., Shibata T., Kiyokawa E., Sasaki H., Singh D.P.. FGF2 antagonizes aberrant TGFβ regulation of tropomyosin: Role for posterior capsule opacity. J. Cell. Mol. Med. (2017) 21 916-928. DOI: 10.1111/jcmm.13030 15. Mansfield K.J., Cerra A., Chamberlain C.G.. FGF-2 counteracts loss of TGFbeta affected cells from rat lens explants: Implications for PCO (after cataract). Mol. Vis. (2004) 10 521-532. PMID: 15303087 16. Cerra A., Mansfield K.J., Chamberlain C.G.. Exacerbation of TGF-beta-induced cataract by FGF-2 in cultured rat lenses. Mol. Vis. (2003) 9 689-700. PMID: 14685144 17. de Iongh R.U., Wederell E., Lovicu F.J., McAvoy J.W.. Transforming Growth Factor-β-Induced Epithelial-Mesenchymal Transition in the Lens: A Model for Cataract Formation. Cells Tissues Organs (2005) 179 43-55. DOI: 10.1159/000084508 18. Lovicu F.J., Flokis M., Wojciechowski M., Mao B., Shu D.. Understanding Fibrotic Cataract: Regulation of TGFβ-Mediated Pathways Leading to Lens Epithelial to Mesenchymal Transition (EMT). J. Jpn. Soc. Cataract. Res. (2020) 32 23-32 19. Xu J., Lamouille S., Derynck R.. TGF-beta-induced epithelial to mesenchymal transition. Cell Res. (2009) 19 156-172. DOI: 10.1038/cr.2009.5 20. Nieto M.A., Huang R.Y.-J., Jackson R.A., Thiery J.P.. EMT: 2016. Cell (2016) 166 21-45. DOI: 10.1016/j.cell.2016.06.028 21. Kalluri R., Neilson E.G.. Epithelial-mesenchymal transition and its implications for fibrosis. J. Clin. Investig. (2003) 112 1776-1784. DOI: 10.1172/JCI200320530 22. Martinez G., de Iongh R.U.. The lens epithelium in ocular health and disease. Int. J. Biochem. Cell Biol. (2010) 42 1945-1963. DOI: 10.1016/j.biocel.2010.09.012 23. Marcantonio J.M., Syam P.P., Liu C.S.C., Duncan G.. Epithelial transdifferentiation and cataract in the human lens. Exp. Eye Res. (2003) 77 339-346. DOI: 10.1016/S0014-4835(03)00125-8 24. Lovicu F.J., Steven P., Saika S., McAvoy J.W.. Aberrant Lens Fiber Differentiation in Anterior Subcapsular Cataract Formation: A Process Dependent on Reduced Levels of Pax6. Investig. Ophthalmol. Vis. Sci. (2004) 45 1946-1953. DOI: 10.1167/iovs.03-1206 25. Jiang J., Shihan M.H., Wang Y., Duncan M.K.. Lens Epithelial Cells Initiate an Inflammatory Response Following Cataract Surgery. Investig. Ophthalmol. Vis. Sci. (2018) 59 4986-4997. DOI: 10.1167/iovs.18-25067 26. Liu Y.C., Wilkins M., Kim T., Malyugin B., Mehta J.S.. Cataracts. Lancet (2017) 390 600-612. DOI: 10.1016/S0140-6736(17)30544-5 27. Abdulhussein D., Hussein M.A.. WHO Vision 2020: Have We Done It?. Ophthalmic Epidemiol. (2022) 1-9. DOI: 10.1080/09286586.2022.2127784 28. Trends in prevalence of blindness and distance and near vision impairment over 30 years: An analysis for the Global Burden of Disease Study. Lancet Glob. Health (2021) 9 e130-e143. DOI: 10.1016/S2214-109X(20)30425-3 29. Vasavada A.R., Praveen M.R.. Posterior Capsule Opacification After Phacoemulsification: Annual Review. Asia-Pacific J. Ophthalmol. (2014) 3 235-240. DOI: 10.1097/APO.0000000000000080 30. Karahan E., Er D., Kaynak S.. An Overview of Nd:YAG Laser Capsulotomy. Med. Hypothesis Discov. Innov. Ophthalmol. (2014) 3 45-50. PMID: 25738159 31. Pandey S.K., Apple D.J., Werner L., Maloof A.J., Milverton E.J.. Posterior capsule opacification: A review of the aetiopathogenesis, experimental and clinical studies and factors for prevention. Indian J. Ophthalmol. (2004) 52 99-112. PMID: 15283214 32. Lovicu F.J., Shin E.H., McAvoy J.W.. Fibrosis in the lens. Sprouty regulation of TGFβ-signaling prevents lens EMT leading to cataract. Exp. Eye Res. (2015) 142 92-101. DOI: 10.1016/j.exer.2015.02.004 33. D’Antin J.C., Tresserra F., Barraquer R.I., Michael R.. Soemmerring’s Rings Developed around IOLs, in Human Donor Eyes, Can Present Internal Transparent Areas. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms232113294 34. Wormstone I.M., Wormstone Y.M., Smith A.J.O., Eldred J.A.. Posterior capsule opacification: What’s in the bag?. Prog. Retin. Eye Res. (2020) 82 100905. DOI: 10.1016/j.preteyeres.2020.100905 35. Wormstone I.M., Eldred J.A.. Experimental models for posterior capsule opacification research. Exp. Eye Res. (2016) 142 2-12. DOI: 10.1016/j.exer.2015.04.021 36. Lu C., Yu S., Song H., Zhao Y., Xie S., Tang X., Yuan X.. Posterior capsular opacification comparison between morphology and objective visual function. BMC Ophthalmol. (2019) 19. DOI: 10.1186/s12886-019-1051-z 37. Aslam T.M., Aspinall P., Dhillon B.. Posterior capsule morphology determinants of visual function. Graefe’s Arch. Clin. Exp. Ophthalmol. (2003) 241 208-212. DOI: 10.1007/s00417-003-0626-8 38. Boswell B.A., VanSlyke J.K., Musil L.S.. Regulation of Lens Gap Junctions by Transforming Growth Factor Beta. Mol. Biol. Cell (2010) 21 1686-1697. DOI: 10.1091/mbc.e10-01-0055 39. Shu D.Y., Wojciechowski M., Lovicu F.J.. ERK1/2-mediated EGFR-signaling is required for TGFβ-induced lens epithelial-mesenchymal transition. Exp. Eye Res. (2019) 178 108-121. DOI: 10.1016/j.exer.2018.09.021 40. Musil L.S.. Primary Cultures of Embryonic Chick Lens Cells as a Model System to Study Lens Gap Junctions and Fiber Cell Differentiation. J. Membr. Biol. (2012) 245 357-368. DOI: 10.1007/s00232-012-9458-y 41. Liu J., Hales A.M., Chamberlain C.G., McAvoy J.W.. Induction of cataract-like changes in rat lens epithelial explants by transforming growth factor beta. Investig. Ophthalmol. Vis. Sci. (1994) 35 388-401. PMID: 8112986 42. Shu D.Y., Wojciechowski M.C., Lovicu F.J.. Bone Morphogenetic Protein-7 Suppresses TGFβ2-Induced Epithelial-Mesenchymal Transition in the Lens: Implications for Cataract Prevention. Investig. Ophthalmol. Vis. Sci. (2017) 58 781-796. DOI: 10.1167/iovs.16-20611 43. Lovicu F.J., Schulz M.W., Hales A.M., Vincent L.N., Overbeek P.A., Chamberlain C.G., McAvoy J.W.. TGFβ induces morphological and molecular changes similar to human anterior subcapsular cataract. Br. J. Ophthalmol. (2002) 86 220. DOI: 10.1136/bjo.86.2.220 44. Srinivasan Y., Lovicu F.J., A Overbeek P.. Lens-specific expression of transforming growth factor beta1 in transgenic mice causes anterior subcapsular cataracts. J. Clin. Investig. (1998) 101 625-634. DOI: 10.1172/JCI1360 45. Koike Y., Yozaki M., Utani A., Murota H.. Fibroblast growth factor 2 accelerates the epithelial–mesenchymal transition in keratinocytes during wound healing process. Sci. Rep. (2020) 10 1-13. DOI: 10.1038/s41598-020-75584-7 46. Maruno K.A., Lovicu F.J., Chamberlain C.G., WMcAvoy J.. Apoptosis is a feature of TGFß-induced cataract. Clin. Exp. Optom. (2002) 85 76-82. DOI: 10.1111/j.1444-0938.2002.tb03012.x 47. Strand D.W., Liang Y.-Y., Yang F., Barron D.A., Ressler S.J., Schauer I.G., Feng X.-H., Rowley D.R.. TGF-β in-duction of FGF-2 expression in stromal cells requires integrated smad3 and MAPK pathways. Am. J. Clin. Exp. Urol. (2014) 2 239-248. PMID: 25374926 48. Wojciechowski M.C., Mahmutovic L., Shu D.Y., Lovicu F.J.. ERK1/2 signaling is required for the initiation but not progression of TGFβ-induced lens epithelial to mesenchymal transition (EMT). Exp. Eye Res. (2017) 159 98-113. DOI: 10.1016/j.exer.2017.03.012 49. Shirakihara T., Horiguchi K., Miyazawa K., Ehata S., Shibata T., Morita I., Miyazono K., Saitoh M.. TGF-β regulates isoform switching of FGF receptors and epithelial-mesenchymal transition. EMBO J. (2011) 30 783-795. DOI: 10.1038/emboj.2010.351 50. Wang Q., McAvoy J.W., Lovicu F.J.. Growth Factor Signaling in Vitreous Humor–Induced Lens Fiber Differentiation. Investig. Ophthalmol. Vis. Sci. (2010) 51 3599-3610. DOI: 10.1167/iovs.09-4797 51. de Iongh R.U., Lovicu F.J., Overbeek P.A., Schneider M.D., Joya J., Hardeman E.D., McAvoy J.W.. Requirement for TGFbeta receptor signaling during terminal lens fiber differentiation. Development (2001) 128 3995-4010. DOI: 10.1242/dev.128.20.3995 52. Shu D.Y., Lovicu F.J.. Enhanced EGF receptor-signaling potentiates TGFβ-induced lens epithelial-mesenchymal transition. Exp. Eye Res. (2019) 185 107693. DOI: 10.1016/j.exer.2019.107693 53. Kokawa N., Sotozono C., Nishida K., Kinoshita S.. High total TGF-β2 levels in normal human tears. Curr. Eye Res. (1996) 15 341-343. DOI: 10.3109/02713689609007630 54. Yamamoto N., Itonaga K., Marunouchi T., Majima K.. Concentration of Transforming Growth Factor β2 in Aqueous Humor. Ophthalmic Res. (2005) 37 29-33. DOI: 10.1159/000083019 55. Ochiai Y., Ochiai H.. Higher Concentration of Transforming Growth Factor-β in Aqueous Humor of Glaucomatous Eyes and Diabetic Eyes. Jpn. J. Ophthalmol. (2002) 46 249-253. DOI: 10.1016/S0021-5155(01)00523-8 56. Agarwal P., Daher A.M., Agarwal R.. Aqueous humor TGF-β2 levels in patients with open-angle glaucoma: A meta-analysis. Mol. Vis. (2015) 21 612-620. PMID: 26019480 57. Cousins S.W., McCabe M.M., Danielpour D., Streilein J.W.. Identification of transforming growth factor-beta as an immunosuppressive factor in aqueous humor. Investig. Ophthalmol. Vis. Sci. (1991) 32 2201-2211. PMID: 2071334 58. de Iongh R.U., Lovicu F.J., Chamberlain C.G., McAvoy J.W.. Differential expression of fibroblast growth factor receptors during rat lens morphogenesis and growth. Investig. Ophthalmol. Vis. Sci. (1997) 38 1688-1699. PMID: 9286257 59. Dawes L.J., Sugiyama Y., Lovicu F.J., Harris C.G., Shelley E.J., McAvoy J.W.. Interactions between lens epithelial and fiber cells reveal an intrinsic self-assembly mechanism. Dev. Biol. (2014) 385 291-303. DOI: 10.1016/j.ydbio.2013.10.030 60. Kurosaka D., Kato K., Nagamoto T., Negishi K.. Growth factors influence contractility and alpha-smooth muscle actin expression in bovine lens epithelial cells. Investig. Ophthalmol. Vis. Sci. (1995) 36 1701-1708. PMID: 7601650 61. Cushing M.C., Mariner P.D., Liao J., Sims E.A., Anseth K.S.. Fibroblast growth factor represses Smad-mediated myofibroblast activation in aortic valvular interstitial cells. FASEB J. (2008) 22 1769-1777. DOI: 10.1096/fj.07-087627 62. Ramos C., Becerril C., Montaño M., García-De-Alba C., Ramírez R., Checa M., Pardo A., Selman M.. FGF-1 reverts epithelial-mesenchymal transition induced by TGF-β1 through MAPK/ERK kinase pathway. Am. J. Physiol. Cell. Mol. Physiol. (2010) 299 L222-L231. DOI: 10.1152/ajplung.00070.2010 63. Schevzov G., Whittaker S.P., Fath T., Lin J.J.-C., Gunning P.W.. Tropomyosin isoforms and reagents. Bioarchitecture (2011) 1 135-164. DOI: 10.4161/bioa.1.4.17897 64. Cheng C., Nowak R.B., Amadeo M.B., Biswas S.K., Lo W.-K., Fowler V.M.. Tropomyosin 3.5 protects the F-actin networks required for tissue biomechanical properties. J. Cell Sci. (2018) 131 jcs222042. DOI: 10.1242/jcs.222042 65. Parreno J., Amadeo M.B., Kwon E.H., Fowler V.M.. Tropomyosin 3.1 association with actin stress fibers is required for lens epithelial to mesenchymal transition. Investig. Ophthalmol. Vis. Sci. (2020) 61 2. DOI: 10.1167/iovs.61.6.2 66. Gateva G., Kremneva E., Reindl T., Kotila T., Kogan K., Gressin L., Gunning P.W., Manstein D.J., Michelot A., Lappalainen P.. Tropomyosin Isoforms Specify Functionally Distinct Actin Filament Populations In Vitro. Curr. Biol. (2017) 27 705-713. DOI: 10.1016/j.cub.2017.01.018 67. Wang Q., Stump R., McAvoy J.W., Lovicu F.J.. MAPK/ERK1/2 and PI3-kinase signalling pathways are required for vitreous-induced lens fibre cell differentiation. Exp. Eye Res. (2008) 88 293-306. DOI: 10.1016/j.exer.2008.08.023 68. Le A.-C.N., Musil L.S.. FGF Signaling in Chick Lens Development. Dev. Biol. (2001) 233 394-411. DOI: 10.1006/dbio.2001.0194 69. Shin E.H., Zhao G., Wang Q., Lovicu F.J.. Sprouty gain of function disrupts lens cellular processes and growth by restricting RTK signaling. Dev. Biol. (2015) 406 129-146. DOI: 10.1016/j.ydbio.2015.09.005 70. Susanto A., Zhao G., Wazin F., Feng Y., Rasko J.E., Bailey C.G., Lovicu F.J.. Spred negatively regulates lens growth by modulating epithelial cell proliferation and fiber differentiation. Exp. Eye Res. (2019) 178 160-175. DOI: 10.1016/j.exer.2018.09.019 71. Saika S., Kono-Saika S., Ohnishi Y., Sato M., Muragaki Y., Ooshima A., Flanders K.C., Yoo J., Anzano M., Liu C.-Y.. Smad3 Signaling Is Required for Epithelial-Mesenchymal Transition of Lens Epithelium after Injury. Am. J. Pathol. (2004) 164 651-663. DOI: 10.1016/S0002-9440(10)63153-7 72. Saika S., Miyamoto T., Ishida I., Shirai K., Ohnishi Y., Ooshima A., McAvoy J.W.. TGFbeta-Smad signalling in postoperative human lens epithelial cells. Br. J. Ophthalmol. (2002) 86 1428-1433. DOI: 10.1136/bjo.86.12.1428 73. Stolen C.M., Jackson M.W., Griep A.E.. Overexpression of FGF-2 modulates fiber cell differentiation and survival in the mouse lens. Development (1997) 124 4009-4017. DOI: 10.1242/dev.124.20.4009 74. Kubo E., Shibata T., Singh D.P., Sasaki H.. Roles of TGF β and FGF Signals in the Lens: Tropomyosin Regulation for Posterior Capsule Opacity. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19103093 75. Akatsu Y., Takahashi N., Yoshimatsu Y., Kimuro S., Muramatsu T., Katsura A., Maishi N., Suzuki H., Inazawa J., Hida K.. Fibroblast growth factor signals regulate transforming growth factor-β-induced endothelial-to-myofibroblast transition of tumor endothelial cells via Elk1. Mol. Oncol. (2019) 13 1706-1724. DOI: 10.1002/1878-0261.12504 76. Strutz F., Zeisberg M., Renziehausen A., Raschke B., Becker V., Kooten C.V., Müller G.A.. TGF-β1 induces proliferation in human renal fibroblasts via induction of basic fibroblast growth factor (FGF-2). Kidney Int. (2001) 59 579-592. DOI: 10.1046/j.1523-1755.2001.059002579.x 77. Bordignon P., Bottoni G., Xu X., Popescu A.S., Truan Z., Guenova E., Kofler L., Jafari P., Ostano P., Röcken M.. Dualism of FGF and TGF-β Signaling in Heterogeneous Cancer-Associated Fibroblast Activation with ETV1 as a Critical Determinant. Cell Rep. (2019) 28 2358-2372.e6. DOI: 10.1016/j.celrep.2019.07.092 78. Symonds J.G., Lovicu F.J., Chamberlain C.G.. Posterior capsule opacification-like changes in rat lens explants cultured with TGFβ and FGF: Effects of cell coverage and regional differences. Exp. Eye Res. (2006) 82 693-699. DOI: 10.1016/j.exer.2005.09.008 79. Lovicu F.J., Chamberlain C.G., McAvoy J.W.. Differential effects of aqueous and vitreous on fiber differentiation and extracellular matrix accumulation in lens epithelial explants. Investig. Ophthalmol. Vis. Sci. (1995) 36 1459-1469. PMID: 7775124 80. Kondo T., Ishiga-Hashimoto N., Nagai H., Takeshita A., Mino M., Morioka H., Kusakabe K.T., Okada T.. Expression of transforming growth factor β and fibroblast growth factor 2 in the lens epithelium of Morioka cataract mice. Congenit. Anom. (2014) 54 104-109. DOI: 10.1111/cga.12042 81. Lovicu F.J., McAvoy J.W.. The age of rats affects the response of lens epithelial explants to fibroblast growth factor. An ultrastructural analysis. Investig. Ophthalmol. Vis. Sci. (1992) 33 2269-2278. PMID: 1607238 82. Zhao G., Bailey C.G., Feng Y., Rasko J., Lovicu F.J.. Negative regulation of lens fiber cell differentiation by RTK antagonists Spry and Spred. Exp. Eye Res. (2018) 170 148-159. DOI: 10.1016/j.exer.2018.02.025 83. Newitt P., Boros J., Madakashira B.P., Robinson M.L., Reneker L.W., McAvoy J.W., Lovicu F.J.. Sef is a negative regulator of fiber cell differentiation in the ocular lens. Differentiation (2010) 80 53-67. DOI: 10.1016/j.diff.2010.05.005
--- title: Investigating the Impact of Food Rewards on Children’s Motivation to Participate in Sport authors: - Alanna Shwed - Brenda Bruner - Barbi Law - Mark W. Bruner journal: Children year: 2023 pmcid: PMC10047004 doi: 10.3390/children10030432 license: CC BY 4.0 --- # Investigating the Impact of Food Rewards on Children’s Motivation to Participate in Sport ## Abstract Children who are physically active and involved in organized sport report having the unhealthiest diets. Research suggests excessive calories may be attributed to the prevalence of fast food and candy which are often provided as rewards in sport. This study explored the use of food as a reward in youth sport and the perceived impact it has on children’s motivation to participate in recreational soccer and ice hockey. A multiple instrumental case study approach was utilized. Children aged 4–12 ($$n = 64$$), parents ($$n = 30$$), and coaches ($$n = 18$$) were recruited within central and northeastern Ontario, Canada to participate in focus groups and individual interviews. Transcribed audio recordings underwent inductive thematic analysis. Key themes included: Fun and fast: The culture of food in youth soccer and hockey; (Un)importance of food rewards: The how and why of motivating children in sport; and Youth sport is expensive: Gratitude for sponsorship in youth sport. Themes explain the role of food and food rewards as an element of the youth sport culture as well as the importance of sponsors, regardless of food affiliation, in youth sport. Overall, children’s participation and effort would continue without food rewards; however, they continue to be offered food to motivate and celebrate performance in youth sport. Findings highlight the need to increase knowledge and awareness among parents and coaches on what truly motivates children to help foster healthier strategies for celebrating success and supporting lifelong physical activity. ## 1. Introduction Extensive research exists to support the physical, psychological, and social health benefits for children engaged in regular physical activity [1,2]. Despite these considerable benefits, children meeting physical activity guidelines also report having the unhealthiest diets [3]. Given the high prevalence of children who are involved in sport to obtain their physical activity (i.e., $60\%$ of children and youth aged 6 to 18 years old) [4], the diets of youth engaged in sport merit important consideration. Understandably, children who are more physically active and involved in organized sports eat larger amounts of food than those who are not [5]. Existing research examining unhealthy diets of children indicates that excess calories are coming from fast food (i.e., food sold from restaurants or snack bars, that are prepared quickly and served in packages ready for take away) [6] and sugar-sweetened beverages [3]. Burgeoning evidence suggests these excessive calories may partly be attributed to candy and sports drinks which are often sold at youth sport centers across North America [7]. In addition to what is available at sport centers, research demonstrates that the choice of snacks is influenced by the highly prevalent fast food sponsorship and celebrity endorsement of unhealthy food (i.e., any food high in saturated fats, trans-fatty acids, free sugars, or salt [8], and do not necessarily come from a restaurant or store) in youth sport [9,10]. Sport celebrity endorsement (e.g., Sidney Crosby, a National Hockey League (NHL) star for Tim Hortons in Canada) and fast food restaurants are major sponsors within youth sport and can influence parents’ and children’s eating habits and preferences [11]. Athletes’ endorsement of these unhealthy products (e.g., Alex Morgan, US Women’s National Soccer Team, and Coca Cola) portrays a false message of health and suggests that unhealthy food is part of the successful sport experience [10]. Further, beyond selling unhealthy food at youth sport centers, there is also heavy marketing within the facilities [12]. Food marketing has led some parents to believe that in order for their child to be successful in sport, they must consume the foods advertised [11]. However, the majority of foods marketed in sport are unhealthy [10] and using food in this manner may play a role in establishing children’s preferences for unhealthy foods [7]. Parents state that despite understanding the negative consequences of feeding their child unhealthy food, they still choose these foods as post-game or practice snacks as they enhance the overall experience and act as a reward for participation [9]. Parents’ rationale for using food rewards to influence children’s behavior might be because it is effective [13]. However, they may also simultaneously offer mixed messages to children about their behavior and what role food should play in their lives [13]. The majority of existing literature exploring how food rewards play a role in parenting has focused on parents using preferred foods (i.e., unhealthy) to reward children for eating healthy foods [14]. Research looking beyond utilizing unhealthy foods to encourage children to eat healthy food is limited and brings into question the short-term and long-term implications of incorporating food rewards as motivation for other behaviors. The use of food rewards for physical activity, and in particular sport participation, is an aspect that is often overlooked; specifically, research investigating food in youth sport and fast food sponsorship as they relate to rewards and motivation for sport continuation is lacking [15]. Investigating rewards and motivation in youth sport is important because although most children are active through sport, the participation rate has dropped by $14\%$ over the last decade [5]. Understanding children’s motivation to participate in sport is critical for supporting long-term engagement and lifelong physical activity [16]. Therefore, the purpose of this research was to explore the use of food as a reward in youth sport and the perceived impact it has on the children’s motivation to participate in recreational soccer and ice hockey (referred to from here on in as hockey). Soccer and hockey are the top two sports for participation among Canadian children [17]. Further, they are the only two sports, out of the top five for participation, that are partnered with unhealthy food companies for sponsorship (i.e., Tim Hortons, PepsiCo, Powerade) [18,19]. The aims of this study were to:Explore the use of food as a reward from the child, parent, and coach perspectives;Explore the use of unhealthy food as a mechanism of motivation for participation;Explore the influence of fast food sponsorship on motivational methods and participation in youth sport. ## 2. Materials and Methods A constructivist paradigm, with a relativist ontology, and subjective and transactional epistemology [20] guided this multiple instrumental case study [21]. A multiple instrumental case study allowed for the exploration of youth recreation soccer and hockey within the larger idea of the impact of food as a reward in youth sport [21]. Within the context of this study, the larger idea was food as a reward in youth sport and the cases explored was the impact of using food as a reward in youth recreational soccer and hockey. This multiple instrumental case study explored the role between food rewards and unhealthy (i.e., energy-dense, nutrient-poor) food sponsorship on children’s motivation to participate in recreational soccer and hockey. We investigated the influence that food sponsorship in sport has on children’s desire to participate in soccer and hockey, and parents and coaches’ motivational strategies for children’s participation. Following approval by the institutional research ethics board (Nipissing University Research Ethics Board #: 102217), criterion-based and snowball sampling methods were used to recruit children, parents, and coaches through two youth recreational soccer leagues in northeastern Ontario, Canada (July 2019–August 2019), and two minor hockey associations in central and northeastern Ontario (September 2019–February 2020). Recruitment ended because continued recruitment efforts did not yield more participants. A total of 11 soccer and 21 hockey associations were contacted through email. Ten associations (four soccer, six hockey) responded with interest and shared the study information with coaches and parents. Two associations from each sport had interested participants who emailed the lead author. After a pilot study to ensure the relevance and appropriateness of the research questions, consent and/or assent were obtained from all participants and focus groups, and interviews were conducted. Focus groups with children aged 4–7 and 8–12 were conducted before or after an already scheduled soccer practice or game (at a booth on the side of the field), or hockey practice (in the hockey arena board room). The age group of children was chosen to understand motivational factors for sport participation across various ages and to capture a time when children first enter sport through to adolescence when participation often drops off [22]. Following a modified graphic elicitation procedure outlined by Cammisa and colleagues [23], children aged 4–7 took part in drawing during their focus groups. Participants were asked to draw themselves playing soccer or hockey, their favorite thing about playing, and what they eat before and after soccer or hockey (drawings can be found in supplementary files). Combining drawing with asking children to verbally answer questions helps reduce their reliance on their own language skills and can give more detailed and organized responses [23]. All child focus groups were conducted separately from parents and coaches to limit their influence [24]. Parent focus groups took place in the same location as the child focus groups; however, they occurred during their child’s practice or game. Coach interviews happened in-person or over the phone at a time most convenient for the individual. Focus groups and interviews followed semi-structured interview guides (see Table 1 for questions) and ended after all questions were asked and participants had nothing to add. Interview questions were developed from the literature [22,25], in consensus with the authorship team, and then piloted with four participants. As a greater understanding of context was obtained, additional questions were added to the interview guides after initial interviews and focus groups. Each participant was also asked to complete a short survey to better understand their sport background and contextualize the role sport plays in their lives. Surveys were completed either in-person (parents, children, coaches) or emailed to the participant (coaches) to complete, prior to their interview or focus group. Example questions include: How important is sport? and, How many hours a week are you involved in sport? Descriptive statistics (means, standard deviations) were used to describe the sample population in terms of coach experience, parental time commitment in youth sport, and children’s sport involvement. The focus group and interview audio recordings were transcribed verbatim, and the data were managed using NVivo 12 [26]. Inductive thematic analysis [27] began after the first focus group and ensured we maintained ontological and epistemological authenticity [20]. First, the lead author AS became familiar with the data through conducting the interviews, transcribing the recordings, and reading through the transcripts. AS wrote down initial thoughts but did not start generating codes until the second read through of all transcripts. Codes were generated by grouping meaningful (i.e., relevant to the topic of interest) repeated ideas, and phrases. After all transcripts were coded, broader themes were created by grouping codes and considering how each could be combined to form an overarching theme (codes and themes can be found in supplementary files). BB acted as a critical friend by going through all the codes and challenging the groups of themes until each code group made sense and each theme properly encompassed all codes. AS then presented all themes to the entire authorship team to further refine themes to ensure they aided in telling a useful story in relation to the research question. To ensure rigor and methodological coherence, criteria explained by Tracy [28] for trustworthiness and Smith and McGannon’s [29] updated recommendations were implemented. To enhance rigor, trustworthiness, sincerity, and credibility [28,30], this study included: [1] multiple participant groups (i.e., children, parents, coaches), methods of data collection (i.e., interviews, focus groups, graphic elicitation), and researcher perspectives (i.e., moderators, critical friend, principal and co-investigators); [2] self-reflexivity practices (e.g., field notes, critical conversations between authors) throughout the entire study; [3] an audit trail of researcher and participants’ insights with a flexible interview guide (i.e., modifications and additions to questions were made as more information was learned throughout data collection); [4] member reflections (i.e., asking for participants to reflect on what the researchers heard); [5] the use of a critical friend (AS, BB); and [6] meetings with the entire research team to determine how to present codes and themes in a way that best reflected the knowledge. ## 3. Results A total of 112 participants took part in the study (Table 2). Participants comprised 64 children (13 girls, 51 boys; 13 children aged 4–7, 51 children aged 8–12), 30 parents (17 women, 13 men), and 20 coaches (4 women, 14 men). Focus groups ($$n = 20$$) ranged in size from two to 16 people ($$n = 64$$ children; $$n = 30$$ parents). Individual interviews ($$n = 18$$) were conducted with coaches. Survey information helped to conceptualize the amount of time participants dedicated to youth sport and how important sport was to them. The experience among the coach participants ($$n = 18$$) ranged from 1 to 30 years; however, all coaches that took part in the study viewed sport as extremely important. Parents ($$n = 30$$) involved in the study spent two or more hours a week dedicated to their child’s sport participation and all of them viewed sport as either somewhat or extremely important. Lastly, $70\%$ of children involved in our study said that soccer/hockey is their favorite sport and $92\%$ indicated that sport is either somewhat or extremely important. Three major themes were constructed from the thematic analysis: Fun and fast: The culture of food in youth soccer and hockey;(Un)importance of food rewards: the how and why of motivating children in sport; andYouth sport is expensive: Gratitude for sponsorship in youth sport. ## 3.1. Fun and Fast: The Culture of Food in Youth Soccer and Hockey This theme explains why fast food and unhealthy food are both elements of youth soccer and hockey and how the expectation of these foods can play a role in developing children’s motivation. Participants discussed the expectation of food at sport venues and fun team snacks as well as the convenience of fast food. Many participants discussed what children eat after soccer or hockey at the sport centers. Coaches discussed the types of food that can be found at sport venues: “We play one night…down at [name of field] and they’ve set up the big ice cream truck right at the field…there’s a lot of postgame visits there” (SC6); however, they did not approve: “The greatest thing that ever happened is the [hockey arena’s] canteen is closed for the season so there’s no french fries, pogos [corndogs], you know the fatty fried food” (HC1). Hockey parents explained that depending on the specific hockey arena, there is the potential for healthier food to be offered: “Other rinks [hockey arenas] have great ones…like chicken noodle soup and chili, more healthy stuff. The ones…out of town have gotten better and better every year I’ve noticed, good food” (HFG2 P4). However, children have come to expect unhealthy food at every hockey arena: “They know that there’s going to be treats in here [hockey arena] for sure” (HFG4 P2). The expectation that there is a food stand at every soccer field was not mentioned in this study; however, participants did explain receiving food after soccer is common. Soccer participants explained team snacks are part of the experience. One parent coach explained that post-game or practice snacks are common but vary by team and family: Similarly, parents supported what coaches said in that: “The end of the game it’s varied from parents handing out like almost a bag of chips, a chocolate bar, to a freezie [popsicle] or something” (SFG2 P5). Children echoed parents and coaches by saying “Every time a different parent brings a different treat” (SFG2 A3). Although team snacks appeared to be a prominent element of the youth soccer experience, they were not mentioned by hockey participants. However, hockey participants did explain that unhealthy food before or after hockey is part of the experience. Fast food restaurants were given as examples of where families go to eat before or after hockey; however, the choice of those restaurants was rationalized by the need to eat, and not necessarily their preference for that food. Parents explained they oftentimes need coffee for themselves and use the opportunity to fuel their child before or after hockey: Children also explained that hockey is often around a mealtime and it made sense to eat on the way to the hockey arena and grab something for after: “I’ll wake up in the morning…for an early game you get something from Tim’s [Tim Hortons], usually like a bagel and then I get Timbits [donut holes] or something for after the game” (HFG5 A1). Many discussions from parents and coaches also explained that stopping at a fast food restaurant is often necessary: Hockey participation is important to families and often dictated dietary habits; however, the same reliance on fast food because of time was not mentioned by soccer participants in this study. ## 3.2. (Un)importance of Food Rewards: The How and Why of Motivating Children in Sport This theme discusses rewards and their influence in youth recreational soccer and hockey. Participants explained the prominence of food rewards, why they exist, and why they are not the main motivating factor behind children’s sport participation. Participants indicated that food is often given as a reward. Hockey coaches explained that their most valuable player award often consists of unhealthy food: “I had given them a little certificate to say that they won the player of the game on such date and my, one of my coaches, my assistant coaches, owns a gas bar in town and he’s donated chocolate bars” (HC8). Soccer coaches also said giving out a gift card award to a restaurant chain is required: “In house league [recreational] soccer, I’m expected to hand out a player of the game card…which is sponsored [by] East Side Mario’s [chain restaurant]” (SC6). Children indicated their parents will reward them with food for playing well: “My mom tells me if I get more than five goals then I get a treat” (SFG4 A1). Some coaches reported using rewards to acknowledge their athletes for displays of effort or success: “It would be the most dedicated, most sportsmanlike, top scorer, those types of things” (HC12). Parents said coaches also reward effort in practice: “Last year [coach] gave a Gatorade to the best listener every practice” (HFG4 P4). Parents also reward children, and coaches indicated overhearing parents explaining to their children how they can earn a treat: “We have a couple of kids on the team the parents will go “okay if you do this well you get a Gatorade, if you don’t play well, you don’t get one” kind of thing” (HC13). Hockey parents did not explicitly admit they give their children rewards for participation; however, they did explain that when hockey is early in the morning or if practice is more skills-focused (e.g., power-skating) and not a game, it is more difficult to motivate their child to get ready to go to hockey. One parent said Tim Hortons helps his son get up for early hockey: “Early mornings I occasionally have to throw in a Tim Hortons visit or something” (HFG1 P4). In contrast to hockey parents, soccer parents did indicate that they reward their children for participation: “So, if you’re trying your hardest, you’re running hard, and you’re listening to your coach and you listen to the advice I give you before the game, you’re going to get rewarded” (SFG2 P4). They rationalized the reward because of the hard work displayed by the child: “It’s like here’s your Gatorade because you just ran for two hours and sweat your ass off right” (SFG2 P2). Children supported coaches and parents by explaining how they earn rewards: “Good games, if I played well…skated well, made some good plays, scored some goals” (HFG4 A1) and “Just try our try our best” (SFG4 A1). Despite almost all coach and parent participants indicating that they reward children, they also said that those rewards ultimately do not change effort or participation: “I would say they are excited by them [treats] and enjoy them, but it’s certainly not why they’re there…The treats are a perk, but I don’t think that’s what gets her to the field” (SC6). However, coaches did explain there are positives that come from rewarding children: When asked about how important they think the snacks and rewards are to their children and the reason they play, every parent said they were not important as: “It’s like a bonus, it’s not a motivator…They want to play hockey, they want to score goals, you don’t have to motivate them, they want to go out there and score a goal” (HFG1 P1). Children supported what coaches and parents said by indicating they do not rely on the rewards to participate. When asked if they would still play if they were not given rewards every child said: “I would still play” (HFG7 A13). However, children also appreciated the rewards. Children in soccer explained the post-game rewards help ease the pain of a tough loss: “But on those…days when I did something that I think I did bad[ly] like scored in [my] own goal…then those rewards are really needed to boost my confidence because… I don’t even know if I’d be here [playing soccer] today” (SFG2 A1). When asking parents about why rewards and snacks an element of youth sport despite them not being critical for ensuring continued participation, hockey participants suggested they are a part of the tradition: “I believe the treats are more about tradition” (HC9). One parent explained the only reason these rewards are now an expectation is because they were introduced: Parents and coaches suggested the food at games and practices are something children look forward to as part of the tradition; however, they also enjoy the sport for just the sport itself. Exploring why the absence of rewards would not influence participation rates led to parents explaining they did not need to motivate their child to play. A soccer parent said: “Yeah if I were to tell [name of child] that I wasn’t going to take him to soccer he would throw a fit” (SFG1 P1) and hockey parents similarly said: “I don’t need to motivate at all, my kids want to be here all the time” (HFG4 P1). Children supported what their parents said by suggesting they value the sport more than the rewards they receive from parents and coaches: “I don’t care about the candy; I care about the game” (HFG2 A3) and “I don’t need anything to encourage me to go” (SFG4 A4). It was obvious the children involved in this study truly enjoy playing soccer or hockey without the food rewards. ## 3.3. Youth Sport Is Expensive: Gratitude for Sponsorship in Youth Sport This theme discusses role of sponsorship on participation in youth sport. Participants did not indicate that sponsors influenced their motivation; however, they valued the financial support that help enable participation. When asked about their opinion, most participants (parents, coaches, and children) initially responded by saying they did not have an opinion. When probed further or after considering the question for longer, many participants expressed their gratitude for the support youth soccer and hockey receive from sponsorship. One hockey coach explained, “You know I think…if I had an opinion of them, I’d say thank you very much. If it wasn’t for sponsorship in sport, where would we get the money to run associations” (HC6). Parents in both soccer and hockey talked about how sponsorship helps reduce the cost of their child playing: “If it cuts down the cost for us great…It [sport] gets to be pretty expensive ‘cause once you start putting more than one kid through it” (SFG1 P2). Despite participants’ appreciation for the financial support, parents and coaches explained that there are standards for what companies are allowed to sponsor youth sports: “Bars/pubs…strip joints…E-cigarette stores are not appropriate. You want local businesses to chip in so that people see that they are investing back in the community” (HC12). Participants did not discuss sponsors related to the food children eat; however, coaches said financial support is more important than ensuring all sponsors are healthy food companies: Coaches emphasized the fact that, without the help of sponsors, regardless of what food the company represents, youth sport would not be able to run: “We need money so desperately that so long as it’s coming from an ethical source…we’re not turning it away because it’s McDonald’s” (SC3). ## 4. Discussion This study aimed to examine the use of food as a reward in youth recreational soccer and hockey from child, parent, and coach perspectives. Unhealthy food rewards were prevalent, supporting previous literature examining the food environment and food rewards in youth sport [15,31]. Research has found that parents often use unhealthy food such as candy or chips for post-game treats [9,15], and they are common in youth recreational leagues [32]. However, participants in this study suggested that unhealthy food in soccer and hockey may not be solely used for providing rewards. Hockey parents explained that feeding children with fast food is necessary because of time constraints before or between games, which are normally scheduled around mealtimes. The theme of limited time dictating food choices supports previous literature that child participation in recreational sport leads parents to make changes to their family schedule and structure [31]. Further, although fast food is purchased because of time constraints, this reliance on convenience foods is arguably contributing to the association children have with hockey and receiving an unhealthy food reward. Interestingly, the theme of limited time dictating food choices was not discussed among soccer participants, which might be explained by the difference in timing of the sports. The recreational soccer season in Canada predominantly runs in the summer months (May–August) when children are not in school and work hours of parents might be more flexible. On the other hand, hockey runs during the fall and winter months (September–March) during school, with normal work schedules for many parents. Lastly, although socioeconomic status information was not collected, hockey is an expensive sport, even at the recreational level, and Tim Hortons and McDonalds are inexpensive; for some families there may be economic reasons behind their choice of food to feed their child. Future research is warranted to investigate the economic influence of youth sport on family eating habits. While soccer parents, coaches, and children did not suggest that fast food is an element of youth soccer, all participants in both soccer and hockey did indicate that food rewards are common and an expected part of the sport. Other research has found similar themes from children explaining they expect certain types of food (e.g., French fries) as rewards after hockey because they are part of the routine [33]. However, our study also found that the choice of food rewards is heavily dependent on family habits and what each one deems appropriate. For example, post-game food rewards ranged from slushies from the concession stand to home-prepared sliced watermelon. The contrast in beliefs of what was considered a food reward reflects similar findings to Rafferty and colleagues [9] who found perceptions of team snacks varied between age groups. Parents of younger children expressed concern for ensuring snacks are healthy, whereas parents of older children were more accepting of the typically offered snacks in sport (i.e., unhealthy foods) [9]. No noticeable differences in post-game food rewards between the different ages of children emerged in our study; however, there was a very clear difference in family habits among and within individual teams. Regardless of the contrast in what was considered a treat, in this study parents and coaches explained that food rewards are not common outside of sport for their children and are something normally just associated with soccer or hockey. Children supported their parents by saying they do not always receive treats outside of soccer or hockey but indicated that food rewards from sport happen often (i.e., potentially twice a week for a game and practice). The second aim of this study was to understand the use of unhealthy food as a mechanism of motivation for sport participation among children and youth. Although there is substantial literature on motivation in youth sport, studies exploring the use of food as a reward and its potential influence on motivation is scarce. Most studies have focused on adolescents, and none have included children, particularly the youngest entering sport (i.e., age 4–5). In addition, few studies have explored the role of food in the sport beyond health-related outcomes (e.g., obesity) and dietary habits. Therefore, the findings from this study add to understanding the role of food rewards on children’s motivation to participate in sport. Parents and coaches reported that rewards such as food sometimes help promote effort from children. This supports the previous literature around parents’ use of rewards to evoke a desired behavior from their children [14], and the reasons children are given food rewards after sports [15,31]. Despite parents and coaches indicating that food rewards can help promote effort, motivation, and fun, they are not what children value most or what influences their involvement in soccer or hockey. This study reveals an interesting paradox where parents, coaches, and children all indicated that food rewards do not influence motivation; however, they continue to be utilized. We found that parents and coaches provide food rewards to make the experience positive, celebrate success, continue tradition, and encourage effort, but not to motivate participation. Although children would play soccer or hockey without the treats, the food rewards give them something to look forward to, and they viewed playing soccer or hockey and getting rewarded as a “win, win”. While rewards were not found to be important for participation, they do add to the enjoyment of sport, which ultimately leads to sport continuation [33]. Similarly, research by Elliott and colleagues [15] also found that receiving food rewards were seen as a positive element to being involved in sport by children. Despite the use of rewards in soccer and hockey, parents, coaches, and children emphasized that participation would not look different if these rewards were not a part of soccer and hockey. The finding of parents and coaches using rewards to promote and celebrate effort and success, but ultimately not aiding in children’s motivation to participate or try their best, may be explained by the type of children participating in this study. Children who like soccer or hockey may not be reliant on food rewards to fuel their motivation to participate or put their best effort forward. However, children in this study did appreciate the food rewards, which might be because they value the meaning behind the reward (i.e., acknowledgement, praise for their hard work). Further, it is clear from these results that what motivates children to play most are the successes found in the game (e.g., scoring) and playing with teammates. Children wanting to play soccer and hockey because of the natural outcomes that come with participation and not needing to be motivated by food rewards suggests greater intrinsic motivation [22]. Future research should continue to explore and better understand why parents and coaches continue to reward children for behavior they maintain will already take place or continue without the reinforcement. Further, exploring parents and coaches’ motives behind the use of food rewards might aid in developing education for parents and coaches that resonates with them to help implement effective and healthy strategies for motivating children. Lastly, this study explored the influence of fast food sponsorship on motivation and sport participation among children. All participants could list many different sponsors for youth soccer and hockey; however, there was no conscious influence on choice of treat or motivational methods found in this study. Most participants did not have an opinion on the type of sponsorship except for their gratitude for financial support, which is similar to the findings of Kelly and colleagues [34,35,36]. Unlike previous literature though, the current study did not find that participants had a problem with fast food sponsorship. Most children did not care who sponsored them as long as they got to play, and parents and coaches were only concerned about the appropriateness of the sponsor in the youth context. While parents, coaches, and children in either sport did not frequently consider the potential influence of sponsors, most hockey participants did recognize Tim Hortons as a well-known sponsor in hockey. As Tim Hortons is a Canadian brand which uses high profile NHL players such as Sidney Crosby, Wayne Gretzky, and Tim Horton himself, it is unknown if this is specific to the Canadian culture, or if the association of a fast food restaurant with a particular sport exists elsewhere in a different context. It might be that the influence of fast food and celebrity endorsement (i.e., Sidney Crosby for Tim Hortons) does have an implicit effect on behavior; however, it is at an unconscious level and so is not recognized or acknowledged upon reflection. Research that does not rely on methods of reflection is warranted to capture the influence of fast food sponsorship and food availability on sport participation behavior in the youth sport context. There are several strengths of this research that are important to highlight. This study supports previous literature from the American [9,31] and Australian youth sport setting [15,34,35,36], and has begun to explore food rewards in youth sport in a Canadian context. Research can build on these cases to look beyond youth recreational soccer and hockey in central and northeastern Ontario and children who are already intrinsically motivated. Second, this study examined the perspectives of parents, children, and coaches. Exploring individual realities among people in the same group highlighted the influence of various levels and factors on behavior. Third, this study included children who are just starting sport. Most research examining children’s motivation in sport has looked at children over the age of 11, even though children enter sport much younger [37]. Lastly, this study was a qualitative exploration, allowing for direct contact with participants and discussion where the researcher was not the expert [38]. Participants’ opinions and perspectives were kept intact by allowing them to share their own thoughts instead of categorizing and rating their experiences through predetermined quantitative measures [38]. While this study adds to the literature exploring physical activity and motivation in youth sport, there are some acknowledged limitations. First, there was the potential for social desirability to emerge during focus groups [39]. To help limit social desirability tendencies (e.g., vague answers, inconsistent responses), researchers provided reassurance to participants that all opinions were valued, probes for more information were used, and requests for examples were made to accompany responses. As well, notes were taken throughout the focus group to consider if answers given by participants truly reflected their honest opinions or if social desirability was perceived. Finally, it is possible those who chose not to participate are the ones who rely on food rewards as motivation. All children in this study liked soccer and hockey and genuinely looked forward to playing. It is possible that children who find sport to be important are motivated differently than children who do not enjoy the sport. Therefore, there is a need to understand what methods of motivation work for children who are not intrinsically motivated to play various sports to encourage sport continuation [40] and foster lifelong physical activity habits [16]. ## 5. Conclusions The findings of this research indicate that food rewards are common occurrences in youth soccer and hockey for effort, success, and participation in both games and practices. However, despite food rewards being considered a normal aspect of the sport experience, they do not appear to influence motivation to participate. While children enjoy receiving treats, they are not what children value most. Knowing that using food as a reward is not required to motivate children’s participation in our sample of youth soccer and hockey players suggests a need to promote other methods of motivation. Limiting food rewards and implementing healthier foods or alternative non-food rewards are recommended to continue to help foster enjoyment and sport continuation but also to help minimize the association that has emerged between youth sport and unhealthy food. Further, providing coaches and parents with information about healthy methods for motivating children and strategies that are not unhealthy food rewards, might help facilitate and foster the more intrinsic reasons children enjoy participating. ## References 1. Eime R.M., Young J.A., Harvey J.T., Charity M.J., Payne W.R.. A systematic review of the psychological and social benefits of participation in sport for children and adolescents: Informing development of a conceptual model of health through sport. Int. J. Behav. Nutr. Phys. Act. (2013) 10 98. DOI: 10.1186/1479-5868-10-98 2. Janssen I., Leblanc A.G.. Systematic review of the health benefits of and fitness in school-aged children and youth. Int. J. Behav. Nutr. Phys. Act. (2010) 7 1. DOI: 10.1186/1479-5868-7-40 3. Thivel D., Tremblay M.S., Katzmarzyk P.T., Fogelholm M., Hu G., Maher C., Maia J., Olds T., Sarmiento O.L., Standage M.. Associations between meeting combinations of 24-hour movement recommendations and dietary patterns of children: A 12-country study. Prev. Med. (2019) 118 159-165. DOI: 10.1016/j.ypmed.2018.10.025 4. The Rally Report. (2020) 5. Croll J.K., Neumark-Sztainer D., Story M., Wall M., Perry C., Harnack L.. Adolescents involved in weight-related and power team sports have better eating patterns and nutrient intakes than non−sport involved adolescents. J. Am. Diet. Asso. (2006) 106 709-717. DOI: 10.1016/j.jada.2006.02.010 6. De Vogli R.D., Kouvonen A., Gimeno D.. The influence of market deregulation on fast food consumption and body mass index: A cross-national time series analysis. Bull. World Health Organ. (2014) 92 99-107A. DOI: 10.2471/BLT.13.120287 7. Nelson T.F., Stovitz S.D., Thomas M., LaVoi N.M., Bauer K.W., Neumark-Sztainer D.. Do youth sports prevent pediatric obesity? A systematic review and commentary. Curr. Sport. Med. Rep. (2011) 10 360-370. DOI: 10.1249/JSR.0b013e318237bf74 8. Report of the Commission on Ending Childhood Obesity. (2016) 9. Rafferty A., Gray V.B., Nguyen J., Nguyen-Rodriguez S., Barrack M., Lin S.. Parents report competing priorities influence snack choice in youth sports. J. Nutr. Educ. Behav. (2018) 50 1032-1039. DOI: 10.1016/j.jneb.2018.04.275 10. Bragg M.A., Roberto C.A., Harris J.L., Brownell K.D., Elbel B.. Marketing food and beverages to youth through sports. J. Adolesc. Health (2018) 62 5-13. DOI: 10.1016/j.jadohealth.2017.06.016 11. Smith M., Signal L., Edwards R., Hoek J.. Children’s and parents’ opinions on the sport-related food environment: A systematic review. Obes. Rev. (2017) 18 1018-1039. DOI: 10.1111/obr.12558 12. Prowse R.J., Naylor P.J., Olstad D.L., Carson V., Storey K., Mâsse L.C., Kirk S.F., Raine K.D.. Food marketing in recreational sport settings in Canada: A Cross-sectional audit in different policy environments using the food and Beverage marketing assessment tool for Settings (FoodMATS). Int. J. Behav. Nutr. Phys. Act. (2018) 15 39. DOI: 10.1186/s12966-018-0673-5 13. Puhl R.M., Schwartz M.B.. If you are good you can have a cookie: How memories of childhood food rules link to adult eating behaviors. Eat. Behav. (2003) 4 283-293. DOI: 10.1016/S1471-0153(03)00024-2 14. Roberts L., Marx J.M., Musher-Eizenman D.R.. Using food as a reward: An examination of parental reward practices. Appetite. (2018) 120 318-326. DOI: 10.1016/j.appet.2017.09.024 15. Elliott S., Velardo S., Drummond M., Drummond C.. Sport and CHILDREN’S nutrition: What can we learn from the junior Australian football Setting?. Asia-Pac. J. Health Sport Phys. Educ. (2016) 7 91-104. DOI: 10.1080/18377122.2016.1145427 16. Palomäki S., Hirvensalo M., Smith K., Raitakari O., Männistö S., Hutri-Kähönen N., Tammelin T.. Does organized sport participation during youth predict healthy habits in adulthood? A 28-year longitudinal study. Scand. J. Med. Sci. Sport. (2018) 28 1908-1915. DOI: 10.1111/sms.13205 17. Bulletin 1: Participation in Sport among Children and Youth. (2013) 18. Canada Soccer Sponsors & Partnerships. (2020) 19. Hockey Canada Partnerships. (2019) 20. Tamminen K.A., Poucher Z.A., Hackfort D., Schinke R.. Research philosophies. The Routledge International Encyclopedia of Sport and Exercise Psychology (2020) 21. Sparkes A.C., Smith B.. Qualitative Research Methods in Sport, Exercise and Health: From Process to Product (2014) 22. Ryan R.M., Deci E.L.. Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. Am. Psychol. (2000) 55 68-78. DOI: 10.1037/0003-066X.55.1.68 23. Cammisa M., Montrone R., Caroli M.. Development and results of a new methodology to perform focus group with preschool children on their beliefs and attitudes on physical activity. Int. J. Ped. Obes. (2011) 6 22-27. DOI: 10.3109/17477166.2011.613658 24. Mauthner M.. Methodological aspects of collecting data from children: Lessons from three research projects. Child. Soc. (1997) 11 16-28. DOI: 10.1111/j.1099-0860.1997.tb00003.x 25. Glanz K., Rimer B.K., Viswanath K.. Health Behavior: Theory, Research, and Practice (2015) 26. 26. QRS International NVivo Qualitative Analysis SoftwareQRS International Pty Ltd.Ruggell, Liechtenstein2019. NVivo Qualitative Analysis Software (2019) 27. Braun V., Clarke V.. To saturate or not to saturate? Questioning data saturation as a useful concept for thematic analysis and sample-size rationales. Qual. Res. Sport Exerc. Health. (2019) 13 201-216. DOI: 10.1080/2159676X.2019.1704846 28. Tracy S.J.. Qualitative quality: Eight “big-tent” criteria for excellent qualitative research. Qual. Inq. (2010) 16 837-851. DOI: 10.1177/1077800410383121 29. Smith B., McGannon K.R.. Developing rigor in qualitative research: Problems and opportunities within sport and exercise psychology. Int. Rev. Sport Exerc. Psychol. (2017) 11 101-121. DOI: 10.1080/1750984X.2017.1317357 30. Shenton A.K.. Strategies for ensuring trustworthiness in qualitative research projects. Educ. Inf. (2004) 22 63-75. DOI: 10.3233/EFI-2004-22201 31. Chircop A., Shearer C., Pitter R., Sim M., Rehman L., Flannery M., Kirk S.. Privileging physical activity over healthy eating: Time to choose?. Health Prom. Int. (2013) 30 418-426. DOI: 10.1093/heapro/dat056 32. Caswell M.S., Hanning R.M.. Adolescent perspectives of the recreational ice hockey food environment and influences on eating behaviour revealed through photovoice. Pub. Health Nutr. (2018) 21 1255-1265. DOI: 10.1017/S1368980018000289 33. Gardner L.A., Magee C.A., Vella S.A.. Enjoyment and behavioral intention predict organized youth sport participation and dropout. J. Phys. Act. Health (2017) 14 861-865. DOI: 10.1123/jpah.2016-0572 34. Kelly B., Baur A.L., Bauman A.E., King L., Chapman K., Smith B.J.. Promoting Health and Nutrition through Sport: Attitudes of the Junior Sporting Community.. (2011) 35. Kelly B., Baur L.A., Bauman A.E., King L., Chapman K., Smith B.J.. Views of children and parents on limiting unhealthy food, drink and alcohol sponsorship of elite and children’s sports. Pub. Health Nutr. (2013) 16 130-135. DOI: 10.1017/S1368980012001188 36. Kelly B., Baur L.A., Bauman A.E., King L., Chapman K., Smith B.J.. “Food company sponsors are kind, generous and cool”: (Mis)conceptions of junior sports players. Int. J. Behav. Nutr. Phys. Act. (2011) 8 95. DOI: 10.1186/1479-5868-8-95 37. Purcell L.. Sport readiness in children and youth. Paediatr. Child. Health (2005) 10 343-344. DOI: 10.1093/pch/10.6.343 38. Willig C., Rogers W.S.. The Sage Handbook of Qualitative Research in Psychology (2017) 39. King M.F., Bruner G.C.. Social desirability bias: A neglected aspect of validity testing. Psychol. Mark. (2000) 17 79-103. DOI: 10.1002/(SICI)1520-6793(200002)17:2<79::AID-MAR2>3.0.CO;2-0 40. Rottensteiner C., Tolvanen A., Laakso L., Konttinen N.. Youth athletes’ motivation, perceived competence, and persistence in organized team sports. J. Sport Behav. (2015) 38 1-18
--- title: Knocking Down CDKN2A in 3D hiPSC-Derived Brown Adipose Progenitors Potentiates Differentiation, Oxidative Metabolism and Browning Process authors: - Yasmina Kahoul - Xi Yao - Frédérik Oger - Maeva Moreno - Souhila Amanzougarene - Mehdi Derhourhi - Emmanuelle Durand - Raphael Boutry - Amélie Bonnefond - Philippe Froguel - Christian Dani - Jean-Sébastien Annicotte - Christophe Breton journal: Cells year: 2023 pmcid: PMC10047013 doi: 10.3390/cells12060870 license: CC BY 4.0 --- # Knocking Down CDKN2A in 3D hiPSC-Derived Brown Adipose Progenitors Potentiates Differentiation, Oxidative Metabolism and Browning Process ## Abstract Human induced pluripotent stem cells (hiPSCs) have the potential to be differentiated into any cell type, making them a relevant tool for therapeutic purposes such as cell-based therapies. In particular, they show great promise for obesity treatment as they represent an unlimited source of brown/beige adipose progenitors (hiPSC-BAPs). However, the low brown/beige adipocyte differentiation potential in 2D cultures represents a strong limitation for clinical use. In adipose tissue, besides its cell cycle regulator functions, the cyclin-dependent kinase inhibitor 2A (CDKN2A) locus modulates the commitment of stem cells to the brown-like type fate, mature adipocyte energy metabolism and the browning of adipose tissue. Here, using a new method of hiPSC-BAPs 3D culture, via the formation of an organoid-like structure, we silenced CDKN2A expression during hiPSC-BAP adipogenic differentiation and observed that knocking down CDKN2A potentiates adipogenesis, oxidative metabolism and the browning process, resulting in brown-like adipocytes by promoting UCP1 expression and beiging markers. Our results suggest that modulating CDKN2A levels could be relevant for hiPSC-BAPs cell-based therapies. ## 1. Introduction Obesity is considered the main risk factor for type 2 diabetes (T2D), mainly due to the excessive accumulation of adipose tissue (AT) [1]. The expansion of AT in obese individuals is a direct cause of the comorbidities, due to the excessive accumulation of triglycerides (TG) within adipocytes, leading to inflammation and insulin resistance. In mammals, there are two major types of AT that are anatomically and functionally distinct: white (WAT) and brown (BAT). White adipocytes store excess energy as TG and release free fatty acids as energy substrate during periods of negative energy balance. BAT differs from WAT by its cellular origin, and is specialized in energy expenditure and the production of heat, mainly through active fat oxidation [2]. Elevated energy expenditure in BAT is correlated with high expression levels of a specific mitochondrial protein named uncoupling protein 1 (UCP1). More recently, the presence of a subtype of thermogenic adipocytes within WAT that also expresses UCP1 has been reported. These inducible adipocytes, named beige, are distinct from white and brown adipocytes. They mainly arise from noradrenergic stimulation or cold exposure. The conversion of white adipocytes into brown-like adipocytes is called browning [3]. Obese individuals are characterized by increased WAT mass and decreased brown and beige AT mass and activity [4]. Increasing energy expenditure by BAT activation or by promoting the browning of WAT may represent a new therapeutic avenue to prevent insulin resistance in obesity and T2D [5]. In humans, cold exposure enhances metabolic activity and thermogenesis in BAT. This increase is accompanied by increased insulin sensitivity in diabetic patients [6]. The transplantation of BAT or brown adipocytes isolated from human adipose progenitors (APs) into the visceral cavity of mice reverses metabolic syndrome and T2D, constituting a potential translatable therapeutic tool to improve metabolic health [7]. Since beige adipocytes can arise through de novo differentiation from undifferentiated APs or via the conversion of mature white adipocytes into UCP1-positive cells, referred to as transdifferentiation [8,9,10], the identification of selective molecular pathways and underlying mechanisms involved in beige adipocyte biogenesis may represent a first step towards innovative therapeutic options. Genome-wide association studies have established that several single nucleotide polymorphisms, including loss-of function mutations in the cyclin-dependent kinase inhibitor 2A (CDKN2A) locus, affect glycemia, insulin values and T2D risk, implying a role in glucose and insulin sensitivity regulation [11,12]. The human CDKN2A locus encodes two proteins, the Cyclin Dependent Kinase inhibitory (CDKI) p16INK4a protein and the p53 regulatory protein p14ARF (p19ARF in mice). The p16INK4a protein is a potent CDKI preventing the binding of CDK$\frac{4}{6}$ to Cyclin D, controlling the CDK4-pRB-E2F1 pathway; whereas p14ARF mainly exerts its activity via the inhibition of MDM2, a ubiquitin-ligase that promotes the degradation of p53 [11,12]. In AT, besides its cell cycle regulator functions (i.e., anti-proliferative and tumor suppressor), the CDKN2A locus also controls the commitment of stem cells to the brown-like type fate and mature adipocyte energy metabolism [13,14,15]. We have shown that mice with a germline disruption of the Cdkn2a gene (Cdkn2a−/−) fed a high-fat diet are protected against diet induced obesity (DIO) by increasing thermogenesis via inguinal (ing) WAT beiging, resulting in improved insulin sensitivity associated with the activation of the PKA pathway [16]. In this study, we have also observed that CDKN2A expression is increased in adipocytes from obese, compared to lean, subjects [16]. Consistent with these findings, a recent study reported that silencing Cdkn2a expression in cold-inducible beige APs results in a rejuvenation of beige adipocyte formation, restoring cold-induced thermogenesis in old mice [13]. The authors also showed that silencing Cdkn2a expression in UCP1+ cells within ingWAT that display progenitor-like characteristics stimulates new beige fat formation through cell proliferation via a cell-autonomous role [17]. Overall, these data indicate the existence of an inverse correlation between the expression level of CDKN2A and beige adipocyte activity, further supporting the notion that cell-cycle genes may be involved in controlling a white-to-beige/brown fat transition that involves APs, beige adipocyte expansion and their activity in a cell-autonomous manner. Human induced pluripotent stem cells (hiPSCs) have the potential to be differentiated into any cell type, making these cells an unlimited source for studying cell-based therapy. In particular, several studies have established the therapeutic potential of hiPSCs differentiated into brown adipocytes progenitors (hiPSC-BAP) against obesity and associated metabolic disorders [18]. However, the limitation of the use of hiPSC-BAPs in 2D cultures is due to their low adipocyte capacity and their low expression levels of UCP1 [18]. Here, to overcome this limit, we report a new method of 3D culture, via the formation of an organoid-like structure, which enhances the capacity for differentiation and the browning process of hiPSC-BAPs [19]. We have previously reported that silencing CDKN2A expression during hiPSC-BAP adipogenic differentiation in 2D cultures promotes UCP1 expression [16]. In this study, we investigated the effects of CDKN2A silencing in hiPSC-BAP in improved 3D cultures. Understanding how Cdkn2a can relay to initiate a thermogenic program in hiPSC-BAPs is a first step to envisage activating beiging as a new putative therapy to alleviate the effects of obesity and to prevent insulin resistance and T2D. RNA-sequencing (RNA-seq) analysis and kinase activity profiling of hiPSC-BAP further demonstrate that CDKN2A silencing enhances pathways involved in adipogenesis, oxidative metabolism and the browning process, resulting in the reprogramming of brown-like adipocytes by promoting UCP1 expression and beiging markers. ## 2.1. Cell Culture and siRNA Experiments hiPSC-BAPs, derived from the hiPSC line NOK6 were grown in standard tissue culture conditions at 37 °C with $5\%$ CO2 as previously described [20]. The growth medium was DMEM low glucose supplemented with L-glutamine (2 mM), penicillin–streptomycin 5000 IU/mL–5000 g/mL (Pen/Strep), $10\%$ FBS and 2.5 ng/mL FGF2. ## 2.1.1. Generation of hiPSC-Derived Brown-like Adipospheres The generation of spheroids and their adipogenic differentiation was performed as we recently described in detail [21]. Briefly, 1 × 106 hiPSC-BAPs were seeded per well of 24 well Ultra-Low Attachment (ULA) plate (Corning 3473, Fischer scientific, Illkirch-Graffenstaden, France) for three days for spheroid formation. Then, to differentiate spheroids, the growth medium was changed to a differentiation medium composed of EBM-2 (Lonza, Colmar, France) supplemented with $0.1\%$ FCS, IBMX (0.5 mM), dexamethasone (0.25 μM), T3 (0.2 nM), insulin (170 nM), rosiglitazone (1 μM), SB431542 (5 μM), and a EGM-2 cocktail (Lonza, CC-3121) including ascorbic acid, hy-drocortisone and EGF. IBMX and dexamethasone were maintained only for the first 3 days of differentiation. SB431542 and EGF were removed after the first 9 days of differentiation. Spheroids were maintained in the differentiation medium for up to 20 days, with the medium changed once a week. ## 2.1.2. siRNA Transfection siRNAs (Human CDKN2A siRNASMART pool, GEHealth Bio-Sciences, Rosersberg, Sweden) were transfected at the time when hiPSC-BAPs were in suspension for spheroid formation. One hundred nM siRNAs were transfected in a medium containing $60\%$ DMEM low glucose, $40\%$ MCDB-201, 1× ITS, dexamethasone (10−9 M), and ascorbic sodium acid (100 mM) using HiPerFect (Qiagen, Courtaboeuf, France) transfection reagent as described by the supplier. Cells were then maintained in conditions to form spheroids and were induced to differentiate as described above. ## 2.2. RNA Extraction and RNA-Sequencing Total RNA was extracted from the hiPSC-BAP 3D at D0 and D10 of differentiation using TRIzolTM Reagent (Sigma-Aldrich, Saint-Quentin-Fallavier, France). The quality of the RNAs was verified with RNA 6000 nanochips on the agilent 2100 bioanalyser. Purified RNA (200 ng) was used for the library preparation. Briefly, RNA libraries were prepared using the TruSeq Stranded mRNA Library Preparation Kit (Illumina, San Diego, CA, USA) following the manufacturer’s instructions. The libraries were sequenced on the NextSeq system (Illumina) using a paired-end 2x75 bp protocol. Three biological replicates per condition were sequenced. The GEO accession number for the sequencing data was GSE223241. ## 2.3. Proteins Extraction and PamGene Kinase Assay Proteins from spheroids and adipospheres were extracted for PamGene kinase assay as previously described [16]. Tyrosine (PTK) and serine/threonine kinase (STK) activity was investigated with PTK and STK microarrays purchased from PamGene (PamGene International BV, ’s-Hertogenbosch, The Netherlands). The experiments were performed as described in the manufacturer’s instructions. ## 2.4. Bioinformatic Analysis For RNA sequencing, the demultiplexing of the sequence data was performed using bcl2fastq Conversion Software (Illumina; bcl2fastq v2.19.1). The trimming of adapter sequences was performed using cutadapt software (version 1.7.1). Reads quality control was assessed using FastQC (v0.11.5). Subsequently, sequence reads from FASTQ files were aligned to the human genome GRCh38, downloaded from Ensembl 108. Alignment was performed using STAR aligner (version 2.5.2b). Over 19 millions of 75 base pairs PE-reads reads were generated per sample. The normalized counts of the different genes and isoforms were performed using RSEM (v1.2.31) using a GTF from Ensembl 108. Finally differential expression was performed using R version 3.6.3 and DESeq2 package v1.24.0. An adjusted p-value < 0.05, Log2FC > 1 and LogFC < −1 were set as thresholds. We then performed pathway analysis using the core analysis function of Ingenuity Pathway analysis (IPA) (Qiagen) and the Gene Set Enrichment Analysis (GSEA) was done using GSEA software version 4.3.2 (GSEA; http://software.broadinstitute.org/gsea/ (accessed on 3 October 2022)). All GSEA data showed had a p-value < 0.05. For Pamgene analysis, image acquisition and data analysis were performed according to the manufacturer’s instructions as previously described [16]. Data and upstream kinase analysis were conducted using the Bionavigator software v.6.3.67.0 developed by PamGene. Peptides and kinases with an adjusted p-value < 0.05, LogFC > 1 and LogFC < −1 were set as thresholds. ## 2.5. Statistical Analysis Data are presented as mean ±SEM. Statistical analyses were performed using unpaired two-tailed Student’s t-test, using GraphPad Prism software. Differences were considered statistically significant at $p \leq 0.05$ (* $p \leq 0.05$; $p \leq 0.01$ and * $p \leq 0.001$). ## 3.1. Characterization of the Differentiation Process of hiPSC-BAPs into Adipocytes in 3D Culture Adipogenic differentiation of hiPSC-BAPs was performed as summarized in Figure 1. Briefly, hiPSC-BAPs were plated, transfected with siRNA, and differentiation was triggered 3 days later, for 10 days. RNA-seq and Pamgene experiments were performed before (at D0) and after (at D10) differentiation. ## 3.1.1. Transcriptome Analysis of the Adipogenic Differentiation of hiPSC-BAPs in 3D Culture Expression profile differences in the transcriptome of spheroids before differentiation (D0) vs. adipospheres after differentiation (D10) were determined by RNA-seq analysis. The 3D culture markedly affects the hiPSC-BAP mRNA expression levels. Transcriptomic analysis revealed 3484 significantly differentially expressed genes (1644 up-regulated and 1840 down-regulated) between D0 and D10 groups (Figure 2A). Overall, RNA-seq analysis revealed that adipogenesis (i.e., upregulation of PPARγ and CEBPα; downregulation of DIO2), markers of mature adipocytes (i.e., upregulation of ADIPOQ and PLIN1), oxidative metabolism pathways and browning adipocyte capacity (i.e., upregulation of FABP4, CIDEA, PGC1α and UCP1) are markedly activated in 10 days-differentiated 3D adipospheres (Figure S1). We also found that mRNA expression levels of DIO2, an enzyme that catalyzes T4 to T3 conversion [22], were markedly downregulated. Given that T3 is present in the differentiation medium of hiPSC-BAPs, the reduction of DIO2 may reflect active adipocyte differentiation which already adopts a brown-like phenotype [23]. After IPA and GSEA analysis, we found that several pro-adipogenic pathways were markedly modified in the D10 vs. D0 groups. We observed the activation of the cholesterol biosynthesis, LXR/RXR and PPAR signaling pathways (Figure 2C and Figure S3); and the repression of the sirtuin, matrix metalloprotease, acute phase response, osteoarthritis, hepatic fibrosis and TGFβ signaling pathways (Figure 2B and Figure S3). Rosiglitazone, CEBPs, IL4 and Vascular Endothelial Growth Factor (VEGF) are major upstream regulators of up-regulated pathways (Figure S2). Other over-expressed pathways were those involved in cellular oxidative metabolism such as oxidative phosphorylation, fatty acid oxidation, ketogenesis, as well as amino acid and noradrenaline degradation pathways (Figure 2C and Figure S3) which are essential for adipogenesis and browning adipocyte capacity [24,25]. ## 3.1.2. Kinome Profiling of the Adipogenic Differentiation of hiPSC-BAPs in 3D Culture Then, we used Pamgene arrays containing serine/threonine (STK) and phosphotyrosine (PTK) peptides that were incubated with protein lysates of hiPSC-BAP adipospheres before (D0) and after differentiation (D10), as previously described [16]. A global decrease in phosphorylation of both phosphorylation sites was observed at D10 vs. D0 (Figure 3A,B). Significant differences in phosphorylation for 5 out of 144 peptides (STK, Figure 3A and Table S2) and for 14 out of 196 peptides (PTK, Figure 3B and Table S2) were evidenced. Using a combined Bionavigator and IPA analysis to identify potential upstream kinases, we found several kinases that displayed differential STK (Figure 3C and Table S1) and PTK (Figure 3D and Table S1) phosphorylation at D10 vs. D0. In line with RNA-seq analysis, global pro-adipogenic pathways whose phosphorylation levels were modified were identified, such as cell cycle regulation, adipogenesis, VEGF, fibroblast growth factor (FGF), as well as PTEN and JAK2/STAT3 signaling pathways (STK, Figure 4A and PTK, Figure 4B), which are implicated in proliferation/differentiation during the early stages of adipogenesis [24,25]. Among them, the intracellular mitogen-activated protein kinase (MAPK) and the three pathways: extracellular signal-regulated kinases (ERK1, 2, 5 and 7), Jun N-terminal kinases (JNKs) and p38 (Figure 3C and Table S1), as well as the FGF receptor family (FGFR 1, 2, 3 and 4) (Figure 3D and Table S1), involved in proliferative activity during adipogenesis [24], displayed lower phosphorylation. ## 3.2. Knock-Down of Cdkn2a Potentiates the Capacity of Adipogenic Differentiation of Spheroids at D0 Given that Cdkn2a might be required in the APs-specific browning process, we decided to assess selective molecular pathways and the underlying mechanisms involved in this process in CDKN2A-deficient hiPSC-BAPs. Expression profile differences in spheroid transcriptome at spheroid stages of differentiation (D0, progenitor stage) between CDKN2A-deficient and control spheroids were determined by RNA-seq. Transcriptomic analysis revealed 245 significantly differentially expressed genes (121 up-regulated and 124 down-regulated) between both groups (Figure 5A). The reduction in CDKN2A mRNA expression levels was validated in spheroids at D0 (Figure S4). Overall, RNA-seq analysis showed that CDKN2A-deficient spheroids exhibit greater adipogenic potential (i.e., downregulation of DIO2) with an anti-inflammatory profile (Figure S4). Computational analysis indicated that pro-adipogenic pathways such as LXR/RXR, PPAR and CXCR4 (chemokine receptor) are activated (Figure 5C and Figure S5); whereas pathways involved in inflammatory response and TGFβ signaling pathways were repressed (Figure 5B and Figure S5) in CDKN2A-deficient vs. control spheroids [24]. Pro-adipogenic factors such as PTGER2 (prostaglandin receptor), VEGF, AREG (retinoic acid signaling) andFOXM1 (Forkhead Box M1) are major upstream regulators of up-regulated pathways (Figure S5), and pro-inflammatory signaling pathways (IL6, IL1α, IL17α, NFκB, IL1β, TNF) are major upstream regulators of down-regulated pathways (Figures S5 and S6) [24]. Up-regulation of the molecular pathways involved in oxidative activity was also observed (Figure S6); whereas no significant change was evidenced in UCP1 RNA expression levels at the D0 progenitor stage (Figure S4). The CDKN2A products p16INK4a and p19ARF are key regulators of the activity of kinases involved in cell proliferation and senescence [16]. We postulated that modified kinase activity may be involved in the browning process and we performed a global kinome analysis in CDKN2A-deficient hiPSC-BAPs. Significant differences in phosphorylation levels for 1 out of 144 peptides (STK, Figure 6A and Table S4) and for 10 out of 196 peptides (PTK, Figure 6B and Table S4) were evidenced between CDKN2A-deficient and control spheroids. Using a combined Bionavigator and IPA analysis, we identified several potential modulated signaling pathways and upstream kinases (STK, Figure 6C; PTK, Figure 6D and Table S3). Among them, glucocorticoid (GC) receptor (GR) signaling pathways (STK, Figure 7A), immune pathways (CD28 signaling in T-helper cells, IL15 production, CTA4 signaling in cytotoxic T lymphocyte), as well as FGF, NGF, Focal Adhesion Kinase (FAK) and insulin receptor signaling (PTK, Figure 7B and Table S3), which are implicated in proliferation/differentiation during the early stages of adipogenesis [24], were evidenced. ## 3.3. CDKN2A Invalidation Potentiates Cellular Oxidative Metabolism and Browning Process of Adipospheres at D10 The silencing of CDKN2A significantly affects the adiposphere mRNA expression levels at D10 (Figure S7). Transcriptomic analysis revealed 610 significantly differentially expressed genes (406 up-regulated and 204 down-regulated) between CDKN2A-deficient and control adipospheres (Figure 8A). The reduction of CDKN2A mRNA expression levels was validated in spheroids at D10 (Figure S7). Overall, RNA-seq analysis revealed that adipogenesis (i.e., upregulation of PPARγ and CEBPα), markers of mature adipocyte (i.e., upregulation of ADIPOQ and PLIN1), oxidative metabolism and browning process pathways (i.e., upregulation of FABP4, SREBF1, CIDEA, PGC1α, UCP1) are markedly activated in 10 days-differentiated CDKN2A-deficient 3D adipospheres (Figure S7). Computational analysis revealed that, in addition to pro-adipogenic pathways already highlighted at D0, global cellular oxidative metabolism (glycolysis, oxidative phosphorylation, TCA cycle, fatty acid β oxidation, ketogenesis, leucine and valine degradation) and the browning process (white adipose tissue browning pathway, AMPK signaling) are markedly activated [25] (Figure 8C and Figure S9). PPARγ and Sterol Regulatory Element Binding Transcription Factor (SREBF 1 and 2) are major upstream regulators of up-regulated pathways (Figure S8). Several kinase pathways such as p70S6K and PI3K/AKT signaling and sirtuin signaling pathway, which affect the proliferation and differentiation of pre-adipocytes, are repressed [24] (Figure 8B). We then analyzed the effect of the knock-down of CDKN2A on the kinome of adipospheres. Significant differences in phosphorylation for 6 out of 144 peptides (STK, Figure 9A and Table S6) and for 8 out of 196 peptides (PTK, Figure 9B and Table S6) were evidenced. Using the Bionavigator analysis to identify potential upstream kinases, we identified several signaling pathways that displayed modified STK (Figure 9C and Table S5) and PTK (Figure 9D and Table S5) phosphorylation. Following combined computional analysis, we observed that, in addition to pro-adipogenic and kinase pathways already highlighted by RNA-seq, AMPK and p38 MAPK which are key players of the browning process [25], are markedly modulated. Differences in phosphorylation linked to the modulation of Gαq- and Gαs-coupled G protein-coupled receptors (GPCR) (STK, Figure 10A) and pro-inflammatory signaling pathways (IL15, IL7) (PTK, Figure 10B) are also modulated [24]. ## 4. Discussion Several studies have pointed out the therapeutic potential of hiPSC-BAP as a promising novel therapy to alleviate the effects of obesity and T2D [18,19]. However, their low capacity for differentiation in brown-like adipocytes in 2D cultures hampers their use for further therapeutic approaches [19]. In recent years, 3D cell culture techniques have received much attention, as these might provide more accurate models of tissues. Indeed, 3D cultures generate changes in lipid accumulation and gene expression, which may lead to a better and closer in vivo differentiation [26]. In order to improve the differentiation capacity of 2D cultures, we developed a novel and more efficient method of using 3D cultures of hiPSC-BAPs, via the formation of an organoid-like structure [19]. Our data confirm previous experiments showing that differentiation into adipospheres improves adipogenesis and browning process capacities compared to conventional monolayer BAP differentiation [16]. Fate decisions of multipotent progenitor cells to differentiate into adipocytes are driven by specific signaling pathways. In particular, the adipogenic process occurs in two major phases: commitment to APs and terminal differentiation, which are determined by modified transcriptional, epigenomic and metabolic activities [24]. Here, we showed that the differentiation of hiPSC-BAP from spheroids to adipospheres in a 3D culture results in marked transcriptomic and phosphorylation changes. Comparative transcriptome and kinome analyses of spheroids before differentiation (D0) vs. adipospheres after differentiation (D10) revealed that adipogenesis, oxidative metabolism pathways and browning adipocyte capacity are markedly activated in 10 days-differentiated 3D adipospheres. RNA-seq analysis revealed the dynamic expression changes that occur during the commitment of APs toward adipocyte differentiation (i.e., repression of osteoarthritis and hepatic fibrosis pathways). TGFβ and sirtuin signaling pathways, which have emerged as critical anti-adipogenic players, were downregulated. TGFβ 1 and 2 and SIRT1 inhibit PPARγ and CEBPα expression [27,28]. TGF-β 1 inhibition suppresses the proliferation and induces the differentiation of hiPSC [19]. By contrast, transcription factor signaling pathways (LXR/RXR, CEBPs and PPARγ), which are the master regulators of adipogenesis [29], were activated. These events are required to promote the growth arrest and differentiation of pre-adipocytes and the progressive expression of a lipogenic transcriptional program (activation of glycolysis, oxidative phosphorylation, fatty acid oxidation and ketogenesis pathways). In line with these findings, the decrease in levels of phosphorylation of MAPK and ERK, JNK, p38 signal-regulated kinases as well as the FGF pathway, which are key regulators of early adipogenic events [24], was evidenced by Pamgene. This might reflect the terminal differentiation of 3D adipospheres into mature adipocytes. In basal 3D culture conditions, the increase in UCP1 [25], IL4 [30] and VEGF [31] expression levels, which is key to the brown adipocyte lineage, suggests that adipocytes already adopt a brown-like phenotype. Indeed, IL-4 enhances the differentiation of APs into committed beige adipogenic precursors [32], and VEGF is synthesized and promotes the angiogenesis in BAT [33]. The activation of cholesterol biosynthesis [34] and angiogenesis (i.e., upregulation of VEGF) [33], and the repression of matrix metalloprotease [35] and hypoxia/inflammation (i.e., dowregulation of HIF1α [36] and acute phase response (APR)) enriched pathways, might reflect an active adipocyte-like remodeling and expansion of the adiposphere. The APR is an early response to inflammation which hampers lipid and glucose utilization in adipocytes [37]. In line with its canonical role in cell-cycle progression and differentiation, the CDKN2A locus is well known to promote adipogenesis [15]. It might also be a key determinant of brown adipocyte fate, although underlying mechanisms and cellular pathways remain elusive [13,15]. Thus, we next assessed molecular pathways involved in the browning process in CDKN2A-deficient hiPSC-BAPs. CDKN2A-deficient spheroids at D0 exhibit greater adipogenic potential with an anti-inflammatory profile (Figure 11). However, no increase in the browning process was evidenced at this stage. In addition to repressed TGFβ and activated adipogenic pathways already highlighted in basal conditions, the most striking observations were the identification of additional modulated signaling pathways, namely the activation of the CXCR4 and the repression of multiple pro-inflammatory signaling pathways. GR and insulin signaling pathways also displayed significant differences in phosphorylation levels. GCs, present in most adipogenic differentiation cocktails, are potent inducers of adipogenesis in vitro. Pre-adipocytes from humans express GR through which GCs stimulate the expression of PPARγ and C/EBPα during adipogenesis [38,39]. Activation of GR decreases pro-inflammatory cytokine expression [40] which is known to inhibit adipogenesis through various pathways, thus constraining the hyperplastic expandability of AT [41]. Insulin is also a powerful inducer of stem cell commitment to adipogenesis via the activation of the PI3K/Akt and MAPK signaling pathways to promote pro-adipogenic transcription [42]. CXCR4 promotes proliferation of APs and is required for the acquisition of brown adipocyte features. It also prevents inflammation [43,44]. Strikingly, a marked enrichment of oxidative metabolism and browning process pathways was observed in 10 days-differentiated CDKN2A-deficient 3D adipospheres (Figure 11). Although a global increase in adipogenesis and cellular oxidative pathways was already evidenced in basal conditions, the silencing of CDKN2A potentiates these pathways along with WAT browning pathways at adiposphere stages of differentiation. The findings of SREBF1 and PPARγ/RXR as major upstream regulators of up-regulated pathways might reflect their dual role in regulating adipogenic and lipogenic pathways [45]. The marked transcriptional activation of AMPK and p38 MAPK and the modulation of the levels of phosphorylation of AMPK and Gα signaling pathways reinforce the idea that adipospheres are fully committed to differentiate into mature brown-like adipocytes [25]. In line with these findings, most of the molecular pathways (i.e., sirtuin, ERK/MAPK, FAK, PI3K, p70S6K) that control pre-adipocyte proliferation [24] were down-regulated, suggesting the terminal differentiation of mature adipocytes (Figure 11). IL15, whose production pathway displays differential phosphorylation, is also known to lower the proliferation rate of pre-adipocytes [46]. On the one hand, AMPK inhibits adipogenesis via blocking the early mitotic clonal expansion. AMPK has a dual role in adipogenesis. AMPK blocks the early mitotic clonal expansion, and later activates the differentiation of pre-adipocytes into mature brown adipocytes [47]. Several studies have reported that AMPK signaling is instrumental in the browning as well as in the energy expenditure of beige adipocytes [48]. Activating intracellular AMPK increases intracellular cAMP and phosphorylates PKA resulting in induced intracellular lipolysis in BAT [48]. p38 MAPK signaling is also a key player in browning [49]. p38 MAPK is a downstream effector kinase of cAMP/PKA signaling in brown adipocytes [50]. During the early phase of adipogenesis, both cAMP and GC signalling pathways promote transcriptional activation, resulting in the commitment of APs to a pre-adipocyte fate and the differentiation of pre-adipocytes [51]. Gαs signaling via GPCRs that activate cAMP/PKA signalling and UCP1-dependent thermogenesis also regulates brown/beige adipocytes [52]. One limitation of our study is the lack of functional tests to further investigate the brown fat properties of CDKN2A-deficient adipospheres at the cellular level. Additional experiments to compare phenotypic differences in control and CDKN2A-deficient adipospheres are also needed to better appreciate whether CDKN2A contributes to increased BAT functions and/or morphology. Moreover, at this stage, we cannot rule out that knocking-down CDKN2A in hiPSC-BAPs could stimulate cell proliferation. However, no difference in size or cell phenotype was observed microscopically between control and CDKN2A-deficient adipospheres throughout the differentiation process and up to 21 days of culture (data not shown). In addition, the RNA-seq data of control and CDKN2A-deficient spheroids and adipospheres did not reveal marked modifications in the expression levels of genes involved in signaling pathways that control proliferation. Thus, it suggests that modulating CDKN2A expression in hiPSC-BAPs does not lead to uncontrolled proliferation of adipospheres. Given that silencing CDKN2A expression in hiPSC-BAPs has limited effect on cell proliferation, it is tempting to speculate that this locus indeed drives alternative pathways from those used for regulating the cell cycle to potentiate the browning process in a 3D system. Here, we demonstrated that CDKN2A plays an important role in brown-like adipogenic recruitment and maturation in a cell-autonomous manner. Our data emphasize the potential effects of this locus in progenitor cells on the browning process, using alternative pathways from those used for regulating the cell cycle. In particular, we showed that AMPK, p38 MAPK and Gαs/cAMP/PKA signaling pathways are key targets of CDKN2A silencing (Figure 11). Thus, additional studies are needed to further delineate the contributions of these kinases and to identify both direct and indirect activators underlying the induction of the browning process in CDKN2A-deficient stem cells. Thus, targeting alternative CDKN2A signaling pathways that may not be involved in tumor suppressive and anti-proliferative effects, but which are driving the browning process in APs, may represent a new strategy to reprogram the cellular response and develop therapeutic approaches against obesity and T2D. ## 5. Conclusions In conclusion, our results suggest that the CDKN2A locus is an important regulator of adipogenesis, oxidative metabolism and the browning process in a cell-autonomous manner. ## References 1. Qasim A., Turcotte M., de Souza R.J., Samaan M.C., Champredon D., Dushoff J., Speakman J.R., Meyre D.. On the Origin of Obesity: Identifying the Biological, Environmental and Cultural Drivers of Genetic Risk among Human Populations. Obes. Rev. (2018) 19 121-149. DOI: 10.1111/obr.12625 2. Goodpaster B.H., Sparks L.M.. Metabolic Flexibility in Health and Disease. Cell Metab. (2017) 25 1027-1036. DOI: 10.1016/j.cmet.2017.04.015 3. Herz C.T., Kiefer F.W.. Adipose Tissue Browning in Mice and Humans. J. Endocrinol. (2019) 241 R97-R109. DOI: 10.1530/JOE-18-0598 4. Kajimura S., Spiegelman B.M., Seale P.. Brown and Beige Fat: Physiological Roles beyond Heat Generation. Cell Metab. (2015) 22 546-559. DOI: 10.1016/j.cmet.2015.09.007 5. Wu J., Boström P., Sparks L.M., Ye L., Choi J.H., Giang A.-H., Khandekar M., Virtanen K.A., Nuutila P., Schaart G.. Beige Adipocytes Are a Distinct Type of Thermogenic Fat Cell in Mouse and Human. Cell (2012) 150 366-376. DOI: 10.1016/j.cell.2012.05.016 6. Hanssen M.J.W., Hoeks J., Brans B., van der Lans A.A.J.J., Schaart G., van den Driessche J.J., Jörgensen J.A., Boekschoten M.V., Hesselink M.K.C., Havekes B.. Short-Term Cold Acclimation Improves Insulin Sensitivity in Patients with Type 2 Diabetes Mellitus. Nat. Med. (2015) 21 863-865. DOI: 10.1038/nm.3891 7. White J.D., Dewal R.S., Stanford K.I.. The Beneficial Effects of Brown Adipose Tissue Transplantation. Mol. Aspects Med. (2019) 68 74-81. DOI: 10.1016/j.mam.2019.06.004 8. Hepler C., Vishvanath L., Gupta R.K.. Sorting out Adipocyte Precursors and Their Role in Physiology and Disease. Genes Dev. (2017) 31 127-140. DOI: 10.1101/gad.293704.116 9. Berry R., Rodeheffer M.S.. Characterization of the Adipocyte Cellular Lineage in Vivo. Nat. Cell Biol. (2013) 15 302-308. DOI: 10.1038/ncb2696 10. Gao H., Volat F., Sandhow L., Galitzky J., Nguyen T., Esteve D., Åström G., Mejhert N., Ledoux S., Thalamas C.. CD36 Is a Marker of Human Adipocyte Progenitors with Pronounced Adipogenic and Triglyceride Accumulation Potential. Stem Cells (2017) 35 1799-1814. DOI: 10.1002/stem.2635 11. Hannou S.A., Wouters K., Paumelle R., Staels B.. Functional Genomics of the CDKN2A/B Locus in Cardiovascular and Metabolic Disease: What Have We Learned from GWASs?. Trends Endocrinol. Metab. (2015) 26 176-184. DOI: 10.1016/j.tem.2015.01.008 12. Morris A.P., Voight B.F., Teslovich T.M., Ferreira T., Segrè A.V., Steinthorsdottir V., Strawbridge R.J., Khan H., Grallert H., Mahajan A.. Large-Scale Association Analysis Provides Insights into the Genetic Architecture and Pathophysiology of Type 2 Diabetes. Nat. Genet. (2012) 44 981-990. DOI: 10.1038/ng.2383 13. Berry D.C., Jiang Y., Arpke R.W., Close E.L., Uchida A., Reading D., Berglund E.D., Kyba M., Graff J.M.. Cellular Aging Contributes to Failure of Cold-Induced Beige Adipocyte Formation in Old Mice and Humans. Cell Metab. (2017) 25 166-181. DOI: 10.1016/j.cmet.2016.10.023 14. Svensson P.-A., Wahlstrand B., Olsson M., Froguel P., Falchi M., Bergman R.N., McTernan P.G., Hedner T., Carlsson L.M.S., Jacobson P.. CDKN2B Expression and Subcutaneous Adipose Tissue Expandability: Possible Influence of the 9p21 Atherosclerosis Locus. Biochem. Biophys. Res. Commun. (2014) 446 1126-1131. DOI: 10.1016/j.bbrc.2014.03.075 15. Kahoul Y., Oger F., Montaigne J., Froguel P., Breton C., Annicotte J.-S.. Emerging Roles for the INK4a/ARF (CDKN2A) Locus in Adipose Tissue: Implications for Obesity and Type 2 Diabetes. Biomolecules (2020) 10. DOI: 10.3390/biom10091350 16. Rabhi N., Hannou S.A., Gromada X., Salas E., Yao X., Oger F., Carney C., Lopez-Mejia I.C., Durand E., Rabearivelo I.. Cdkn2a Deficiency Promotes Adipose Tissue Browning. Mol. Metab. (2018) 8 65-76. DOI: 10.1016/j.molmet.2017.11.012 17. Park J., Shin S., Liu L., Jahan I., Ong S.-G., Xu P., Berry D.C., Jiang Y.. Progenitor-like Characteristics in a Subgroup of UCP1+ Cells within White Adipose Tissue. Dev. Cell (2021) 56 985-999.e4. DOI: 10.1016/j.devcel.2021.02.018 18. Hafner A.-L., Contet J., Ravaud C., Yao X., Villageois P., Suknuntha K., Annab K., Peraldi P., Binetruy B., Slukvin I.I.. Brown-like Adipose Progenitors Derived from Human Induced Pluripotent Stem Cells: Identification of Critical Pathways Governing Their Adipogenic Capacity. Sci. Rep. (2016) 6 32490. DOI: 10.1038/srep32490 19. Yao X., Dani V., Dani C.. Human Pluripotent Stem Cells: A Relevant Model to Identify Pathways Governing Thermogenic Adipocyte Generation. Front. Endocrinol. (2020) 10 932. DOI: 10.3389/fendo.2019.00932 20. Mohsen-Kanson T., Hafner A.-L., Wdziekonski B., Takashima Y., Villageois P., Carrière A., Svensson M., Bagnis C., Chignon-Sicard B., Svensson P.-A.. Differentiation of Human Induced Pluripotent Stem Cells into Brown and White Adipocytes: Role of Pax3. Stem Cells (2014) 32 1459-1467. DOI: 10.1002/stem.1607 21. Yao X., Dani C.. A Simple Method for Generating, Clearing, and Imaging Pre-Vascularized 3D Adipospheres Derived from Human IPS Cells. Methods Mol. Biol. (2022) 2454 495-507. DOI: 10.1007/7651_2021_360 22. Bianco A.C., Salvatore D., Gereben B., Berry M.J., Larsen P.R.. Biochemistry, Cellular and Molecular Biology, and Physiological Roles of the Iodothyronine Selenodeiodinases. Endocr. Rev. (2002) 23 38-89. DOI: 10.1210/edrv.23.1.0455 23. Wagner M.S., Wajner S.M., Dora J.M., Maia A.L.. Regulation of Dio2 Gene Expression by Thyroid Hormones in Normal and Type 1 Deiodinase-Deficient C3H Mice. J. Endocrinol. (2007) 193 435-444. DOI: 10.1677/JOE-07-0099 24. Ambele M.A., Dhanraj P., Giles R., Pepper M.S.. Adipogenesis: A Complex Interplay of Multiple Molecular Determinants and Pathways. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21124283 25. Machado S.A., Pasquarelli-do-Nascimento G., da Silva D.S., Farias G.R., de Oliveira Santos I., Baptista L.B., Magalhães K.G.. Browning of the White Adipose Tissue Regulation: New Insights into Nutritional and Metabolic Relevance in Health and Diseases. Nutr. Metab. (2022) 19 61. DOI: 10.1186/s12986-022-00694-0 26. Shen J.X., Couchet M., Dufau J., de Castro Barbosa T., Ulbrich M.H., Helmstädter M., Kemas A.M., Zandi Shafagh R., Marques M.-A., Hansen J.B.. 3D Adipose Tissue Culture Links the Organotypic Microenvironment to Improved Adipogenesis. Adv. Sci. (2021) 8 2100106. DOI: 10.1002/advs.202100106 27. Ahdjoudj S., Kaabeche K., Holy X., Fromigué O., Modrowski D., Zérath E., Marie P.J.. Transforming Growth Factor-Beta Inhibits CCAAT/Enhancer-Binding Protein Expression and PPARgamma Activity in Unloaded Bone Marrow Stromal Cells. Exp. Cell Res. (2005) 303 138-147. DOI: 10.1016/j.yexcr.2004.09.013 28. Majeed Y., Halabi N., Madani A.Y., Engelke R., Bhagwat A.M., Abdesselem H., Agha M.V., Vakayil M., Courjaret R., Goswami N.. SIRT1 Promotes Lipid Metabolism and Mitochondrial Biogenesis in Adipocytes and Coordinates Adipogenesis by Targeting Key Enzymatic Pathways. Sci. Rep. (2021) 11 8177. DOI: 10.1038/s41598-021-87759-x 29. Lefterova M.I., Haakonsson A.K., Lazar M.A., Mandrup S.. PPARγ and the Global Map of Adipogenesis and Beyond. Trends Endocrinol. Metab. (2014) 25 293-302. DOI: 10.1016/j.tem.2014.04.001 30. Su S., Guntur A.R., Nguyen D.C., Fakory S.S., Doucette C.C., Leech C., Lotana H., Kelley M., Kohli J., Martino J.. A Renewable Source of Human Beige Adipocytes for Development of Therapies to Treat Metabolic Syndrome. Cell Rep. (2018) 25 3215-3228.e9. DOI: 10.1016/j.celrep.2018.11.037 31. Nishio M., Yoneshiro T., Nakahara M., Suzuki S., Saeki K., Hasegawa M., Kawai Y., Akutsu H., Umezawa A., Yasuda K.. Production of Functional Classical Brown Adipocytes from Human Pluripotent Stem Cells Using Specific Hemopoietin Cocktail without Gene Transfer. Cell Metab. (2012) 16 394-406. DOI: 10.1016/j.cmet.2012.08.001 32. Lizcano F., Vargas D., Gómez Á., Torrado A.. Human ADMC-Derived Adipocyte Thermogenic Capacity Is Regulated by IL-4 Receptor. Stem Cells Int. (2017) 2017 2767916. DOI: 10.1155/2017/2767916 33. Jin H., Li D., Wang X., Jia J., Chen Y., Yao Y., Zhao C., Lu X., Zhang S., Togo J.. VEGF and VEGFB Play Balancing Roles in Adipose Differentiation, Gene Expression, and Function. Endocrinology (2018) 159 2036-2049. DOI: 10.1210/en.2017-03246 34. Chung S., Parks J.S.. Dietary Cholesterol Effects on Adipose Tissue Inflammation. Curr. Opin. Lipidol. (2016) 27 19-25. DOI: 10.1097/MOL.0000000000000260 35. Chavey C., Mari B., Monthouel M.-N., Bonnafous S., Anglard P., Van Obberghen E., Tartare-Deckert S.. Matrix Metalloproteinases Are Differentially Expressed in Adipose Tissue during Obesity and Modulate Adipocyte Differentiation. J. Biol. Chem. (2003) 278 11888-11896. DOI: 10.1074/jbc.M209196200 36. Wagegg M., Gaber T., Lohanatha F.L., Hahne M., Strehl C., Fangradt M., Tran C.L., Schönbeck K., Hoff P., Ode A.. Hypoxia Promotes Osteogenesis but Suppresses Adipogenesis of Human Mesenchymal Stromal Cells in a Hypoxia-Inducible Factor-1 Dependent Manner. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0046483 37. Venteclef N., Jakobsson T., Steffensen K.R., Treuter E.. Metabolic Nuclear Receptor Signaling and the Inflammatory Acute Phase Response. Trends Endocrinol. Metab. (2011) 22 333-343. DOI: 10.1016/j.tem.2011.04.004 38. Wu Z., Bucher N.L., Farmer S.R.. Induction of Peroxisome Proliferator-Activated Receptor Gamma during the Conversion of 3T3 Fibroblasts into Adipocytes Is Mediated by C/EBPbeta, C/EBPdelta, and Glucocorticoids. Mol. Cell Biol. (1996) 16 4128-4136. DOI: 10.1128/MCB.16.8.4128 39. Smas C.M., Chen L., Zhao L., Latasa M.J., Sul H.S.. Transcriptional Repression of Pref-1 by Glucocorticoids Promotes 3T3-L1 Adipocyte Differentiation. J. Biol. Chem. (1999) 274 12632-12641. DOI: 10.1074/jbc.274.18.12632 40. Desarzens S., Faresse N.. Adipocyte Glucocorticoid Receptor Has a Minor Contribution in Adipose Tissue Growth. J. Endocrinol. (2016) 230 1-11. DOI: 10.1530/JOE-16-0121 41. Jiang N., Li Y., Shu T., Wang J.. Cytokines and Inflammation in Adipogenesis: An Updated Review. Front. Med (2019) 13 314-329. DOI: 10.1007/s11684-018-0625-0 42. Cignarelli A., Genchi V.A., Perrini S., Natalicchio A., Laviola L., Giorgino F.. Insulin and Insulin Receptors in Adipose Tissue Development. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20030759 43. van den Berg S.M., van Dam A.D., Rensen P.C.N., de Winther M.P.J., Lutgens E.. Immune Modulation of Brown(Ing) Adipose Tissue in Obesity. Endocr. Rev. (2017) 38 46-68. DOI: 10.1210/er.2016-1066 44. Kurita K., Ishikawa K., Takeda K., Fujimoto M., Ono H., Kumagai J., Inoue H., Yokoh H., Yokote K.. CXCL12-CXCR4 Pathway Activates Brown Adipocytes and Induces Insulin Resistance in CXCR4-Deficient Mice under High-Fat Diet. Sci. Rep. (2019) 9 6165. DOI: 10.1038/s41598-019-42127-8 45. Ahmadian M., Suh J.M., Hah N., Liddle C., Atkins A.R., Downes M., Evans R.M.. PPARγ Signaling and Metabolism: The Good, the Bad and the Future. Nat. Med. (2013) 19 557-566. DOI: 10.1038/nm.3159 46. Fuster G., Almendro V., Fontes-Oliveira C.C., Toledo M., Costelli P., Busquets S., López-Soriano F.J., Argilés J.M.. Interleukin-15 Affects Differentiation and Apoptosis in Adipocytes: Implications in Obesity. Lipids (2011) 46 1033-1042. DOI: 10.1007/s11745-011-3594-5 47. Ahmad B., Serpell C.J., Fong I.L., Wong E.H.. Molecular Mechanisms of Adipogenesis: The Anti-Adipogenic Role of AMP-Activated Protein Kinase. Front. Mol. Biosci. (2020) 7 76. DOI: 10.3389/fmolb.2020.00076 48. van der Vaart J.I., Boon M.R., Houtkooper R.H.. The Role of AMPK Signaling in Brown Adipose Tissue Activation. Cells (2021) 10. DOI: 10.3390/cells10051122 49. Leiva M., Matesanz N., Pulgarín-Alfaro M., Nikolic I., Sabio G.. Uncovering the Role of P38 Family Members in Adipose Tissue Physiology. Front. Endocrinol. (2020) 11 572089. DOI: 10.3389/fendo.2020.572089 50. Bordicchia M., Liu D., Amri E.-Z., Ailhaud G., Dessì-Fulgheri P., Zhang C., Takahashi N., Sarzani R., Collins S.. Cardiac Natriuretic Peptides Act via P38 MAPK to Induce the Brown Fat Thermogenic Program in Mouse and Human Adipocytes. J. Clin. Investig. (2012) 122 1022-1036. DOI: 10.1172/JCI59701 51. Petersen R.K., Madsen L., Pedersen L.M., Hallenborg P., Hagland H., Viste K., Døskeland S.O., Kristiansen K.. Cyclic AMP (CAMP)-Mediated Stimulation of Adipocyte Differentiation Requires the Synergistic Action of Epac- and CAMP-Dependent Protein Kinase-Dependent Processes. Mol. Cell. Biol. (2008) 28 3804-3816. DOI: 10.1128/MCB.00709-07 52. Klepac K., Kilić A., Gnad T., Brown L.M., Herrmann B., Wilderman A., Balkow A., Glöde A., Simon K., Lidell M.E.. The Gq Signalling Pathway Inhibits Brown and Beige Adipose Tissue. Nat. Commun. (2016) 7 10895. DOI: 10.1038/ncomms10895
--- title: Overshoot of the Respiratory Exchange Ratio during Recovery from Maximal Exercise Testing in Young Patients with Congenital Heart Disease authors: - Marco Vecchiato - Andrea Ermolao - Emanuele Zanardo - Francesca Battista - Giacomo Ruvoletto - Stefano Palermi - Giulia Quinto - Gino Degano - Andrea Gasperetti - Massimo A. Padalino - Giovanni Di Salvo - Daniel Neunhaeuserer journal: Children year: 2023 pmcid: PMC10047014 doi: 10.3390/children10030521 license: CC BY 4.0 --- # Overshoot of the Respiratory Exchange Ratio during Recovery from Maximal Exercise Testing in Young Patients with Congenital Heart Disease ## Abstract Introduction: The overshoot of the respiratory exchange ratio (RER) after exercise is reduced in patients with heart failure. Aim: The present study aimed to investigate the presence of this phenomenon in young patients with congenital heart disease (CHD), who generally present reduced cardiorespiratory fitness. Methods: In this retrospective study, patients with CHD underwent a maximal cardiopulmonary exercise testing (CPET) assessing the RER recovery parameters: the RER at peak exercise, the maximum RER value reached during recovery, the magnitude of the RER overshoot and the linear slope of the RER increase after the end of the exercise. Results: In total, 117 patients were included in this study. Of these, there were 24 healthy age-matched control subjects and 93 young patients with CHD (transposition of great arteries, Fontan procedure, aortic coarctation and tetralogy of Fallot). All patients presented a RER overshoot during recovery. Patients with CHD showed reduced aerobic capacity and cardiorespiratory efficiency during exercise, as well as a lower RER overshoot when compared to controls. RER magnitude was higher in the controls and patients with aortic coarctation when compared to those with transposition of great arteries, previous Fontan procedure, and tetralogy of Fallot. The RER magnitude was found to be correlated with the most relevant cardiorespiratory fitness and efficiency indices. Conclusions: The present study proposes new recovery indices for functional evaluation in patients with CHD. Thus, the RER recovery overshoots analysis should be part of routine CPET evaluation to further improve prognostic risk stratifications in patients with CHD. ## 1. Introduction Congenital heart disease (CHD) accounts for nearly one-third of all major congenital anomalies and its birth prevalence worldwide is suggested to vary [1]. Recent data extracted from the European Surveillance of Congenital Anomalies estimated the average total prevalence of CHD in Europe is around 8.0 per 1000 births [2]. In this patient population, long-term survival is decreased [3], with lesion severity and repair status as major risk factors for excess mortality [4]. However, due to improvements in medical, surgical, and intensive care interventions, the life expectancy of patients born with CHD has been rising over time [3]. Cardiorespiratory fitness was found to be highly heterogeneous both within and between individuals with CHD diagnoses [5]. In this context, cardiopulmonary exercise testing (CPET) has emerged as an important tool for risk stratification and may guide clinicians in assessing prognosis and planning interventions in CHD patients [6,7]. CPET may also be useful for diagnostic purposes as well as for decision-making making. Most of the studies about CPET have focused on the cardiopulmonary response during exercise; however, there is less evidence about the respiratory gas indices during recovery with most studies focusing only on oxygen uptake (VO2) kinetics [8,9]. Most studies on the recovery phase of patients with heart disease are on adult patients with heart failure (HF) [10,11,12]. These studies demonstrated that patients with HF exhibit an increased VO2 delay during the recovery phase after maximal CPET compared to controls and this delay is associated with the severity of diseases [13]. Recently, the transient increase, defined as overshoot, of some CPET parameters during the recovery phase, such as the respiratory exchange ratio (RER), has aroused scientific and clinical interest [12]. The magnitudes of these parameters have been compared between HF patients and healthy subjects, demonstrating that overshoots tended to be more pronounced in subjects with better cardiopulmonary function during exercise [12]. Cardiovascular recovery after exercise appears to be faster in children than in adults [14,15], but data about the recovery after CPET in young patients with CHD is limited. VO2 recovery kinetics and heart rate (HR) recovery are prolonged in patients with different types of CHD [16,17] but little is known about the prognostic and diagnostic value of this data and how it can guide clinical decision-making [18,19]. Therefore, the present study aimed to evaluate the behavior of the respiratory gas exchange indices during recovery in young patients with CHD, assessing the impact of different conditions compared with an age-matched healthy control group. Furthermore, it will be discussed how these recovery parameters might be used for prognostic risk stratification in clinical routine. ## 2.1. Study Design and Population Characteristics This was a retrospective observational cross-sectional study that included all young patients with CHD (aged between 7 and 20 years) who were evaluated at our Sports and Exercise Medicine Division between 2018 and 2021 for cardiovascular screening/follow-up, sports eligibility assessment, and/or exercise prescription [20]. Patients with different CHD were compared to highlight any functional and prognostic differences between their CPET exercise and recovery parameters. The selected CHD were the 4 most represented within our population: transposition of great arteries (TGA), patients with univentricular CHD who underwent Fontan procedure (Fon), aortic coarctation (CoA), and tetralogy of Fallot (ToF). Other or complex CHD, as well as all patients with beta-blocker therapy and/or pacemaker, were excluded from this study. The other exclusion criteria were related to absolute contraindications to CPET evaluation, as well as musculoskeletal disease that would impede maximal exercise testing. Moreover, tests with gas monitoring of fewer than four minutes during the recovery phase were excluded. A control group of apparently healthy subjects was added, consisting of children who were referred for pre-participation screening or for minor complaints during exercise, such as chest pain, palpitations or breathing difficulties, but were declared negative after the diagnostic process. The control group was selected to match the patient study population with regard to gender, age, and body mass index (BMI). In accordance with legal regulations, the Code of Medical Ethics and the Declaration of Helsinki, subjects were duly informed of the risks, benefits, and stress deriving from the study protocol and signed a written informed consent form. This study was approved by the local Ethics Committee for Clinical Research (protocol code 302n/AO/22). ## 2.2. Cardiopulmonary Exercise Testing For each patient, personal history was collected and a physical examination was conducted. Each subject underwent a standardized, incremental, maximal 12-leads ECG-monitored CPET (Masterscreen CPX system Jaeger, Carefusion, Hoechberg, Germany) using a treadmill (T170 DE-med, h/p/cosmos, Nussdorf-Traunstein, Germany) until a rating of perceived exertion (RPE) ≥ $\frac{18}{20}$ of Borg Scale was reached and metabolic, cardiovascular, or ventilatory signs of exhaustion appeared. Blood pressure was measured both at rest and during CPET and its recovery phase. The age-predicted heart rate (HR) was calculated with the following formula: (220-age) bpm. The respiratory gas exchange (VO2, VCO2) and ventilation (VE) were monitored through the breath-by-breath mode and at least until the fourth minute of recovery. The first ventilatory threshold (VT) was identified through the V-slope method. When the VT was not clearly identifiable, it was determined by the consensus of two physicians within the following group (A.E., G.D., A.G., or D.N.). The respiratory compensation point (RCP) was determined with the same principle considering the ventilatory equivalents and the partial pressure of end-tidal carbon dioxide (PETCO2). The VE/VCO2 slope was calculated as the coefficient of linear regression from the beginning of the exercise (removing possible initial hyperventilation) to the RCP. The oxygen uptake efficiency slope (OUES) was determined by the slope of the regression line between VO2 and the logarithm of VE [21,22]. ## 2.3. Overshoot Analyses The RER overshoot was analyzed by assessing five parameters (Figure 1): ## 2.4. Ventricular Function Assessment Furthermore, all subjects were assessed to obtain data regarding ventricular systolic function. Data regarding ventricular systolic function were obtained by echocardiographic evaluations, which have been performed in the context of the routine follow-up of these patients at the Department of Women’s and Children’s Health of University of Padova. The parameter chosen to quantify the systolic function of the left ventricle is the ejection fraction (LVEF), which indicates the ratio (expressed as a percentage) between the volume of blood expelled during systole from the left ventricle and the end-diastolic volume. For the systolic function of the right ventricle, on the other hand, tricuspid annular plane systolic excursion (TAPSE), which is the displacement of the tricuspid valve plane towards the cardiac apex during ventricular systole, as well as the fractional area change (FAC), as shortening percentage of the right ventricle between systole and diastole, were evaluated. ## 2.5. Statistical Analyses Data are expressed as a mean ± the standard deviation. The normality was assessed using the Shapiro–Wilk test. T-tests for the normally distributed variables and Mann–Whitney U tests for the non-normally distributed variables were used. The various classes of CHD were compared with each other and with controls by an ANOVA test for normally distributed variables and with a non-parametric test for non-normally distributed variables. Patients with CHD were further classified to investigate the RER recovery parameters in subgroups with potential prognostic differences. Patients were grouped according to the VE/VCO2 slope into ventilatory classes: I (VE/VCO2 slope < 30), II (VE/VCO2 slope between 30 and 35.9), and III (VE/VCO2 slope between 36 and 44.9); no patients belonged to ventilatory class IV (VE/VCO2 slope > 45). The correlations were evaluated with Pearson’s index for normally distributed variables and with Spearman’s index for non-normally distributed variables. The statistical analyses were executed with IBM SPSS Statics software version 26. A statistical significance level of p ≤ 0.05 was applied. ## 3.1. Patients Selection In total, 131 young subjects were initially recruited for the aim of the study, including 103 patients with CHD and 28 control subjects. In the CHD group, seven patients were excluded from the study because it was not possible to clearly identify an RER max during the time recorded; one patient was excluded due to a sampling error, and two patients were excluded because they did not reach the criteria for metabolic exhaustion. Therefore, the CHD study group was made up of 93 subjects: 23 TGA, 22 Fon, 24 CoA, and 24 ToF. All patients with TGA correction underwent the arterial switch procedure. Moreover, 18 patients of the Fontan group presented a left dominant ventricle, whereas 7 had a right dominant ventricle and 1 patient underwent a staged biventricular conversion. The degree of pulmonary valve regurgitation and right ventricle outflow tract stenosis was rather heterogeneous in the ToF group. In the control group, three patients were excluded because it was not possible to clearly identify a RER max during the recovery phase, and one patient because he was unable to reach the needed metabolic criteria for exhaustion. The control group, therefore, was composed of 24 subjects. ## 3.2. Baseline Characteristics The general anthropometric and clinical characteristics of the study participants are represented in Table 1. Resting systolic and diastolic blood pressures were higher in CHD patients than in healthy controls ($$p \leq 0.022$$ and $p \leq 0.001$, respectively). Moreover, statistically significant differences were found between the 4 subgroups of CHD in systolic blood pressure ($$p \leq 0.001$$), with patients with CoA having the highest mean resting SBP (119.92 ± 14.65 mmHg). None of the subjects included in the control group had a low blood oxygen saturation at rest, while five patients with CHD (all belonging to the Fon group) desaturated at rest before the exercise phase. ## 3.3. Cardiopulmonary Exercise Testing and Echocardiographic Assessments All patients performed their respective maximal CPET with the same protocol (Bruce Ramp) reaching a RPE ≥ $\frac{18}{20}$ on the Borg scale, with no reported symptoms. The results of the CPET and the main indices of cardiac contractility analyzed with echocardiography are shown in Table 2. HR peak was lower in the CHD group than in the control group ($$p \leq 0.013$$). The comparison of HR peak between the subgroups was also statistically significant ($$p \leq 0.001$$), with patients with Fontan having the lowest median (176 bpm). Even the HR recovery after one minute was lower in patients with CHD compared to controls ($$p \leq 0.007$$), with Fon patients showing the slowest recovery. As for the oxygen pulse (O2 pulse), an anomalous behavior was recorded in $30\%$ of patients with CHD, whereas all healthy controls had a normal O2 pulse behavior during the test. Patients with CHD showed lower values of the O2 pulse in terms of percentage of predicted, when compared to healthy controls ($$p \leq 0.005$$) with still Fon patients presenting lower values compared to the other three CHD subgroups. The analysis of peripheral saturation at peak exercise (SpO2 peak) showed that 19 patients with CHD desaturated at peak exercise (12 from the Fon group), while none of the healthy controls had a lower-than-normal peripheral saturation. Pairwise comparisons between the various parameters are shown in Supplementary Table S1. Most significant differences have been found between the Fontan group, which had the greatest functional impairment, and the CoA and control groups. CoA patients showed higher HR peak and HR peak (%) compared to ToF and Fon patients but similar to TGA patients. Furthermore, aerobic capacity was significantly lower in patients with CHD compared to healthy subjects; statistically significant differences were also displayed between the four subgroups (in both cases $p \leq 0.001$) with Fon patients recording the lowest VO2 peak (32.05 ± 5.90 mL/min/kg; 76.64 ± $14.40\%$ of predicted) and CoA patients with the highest aerobic capacity (40.98 ± 8.40 mL/min/kg; 99.00 ± $17.10\%$ of predicted). ## 3.4. Overshoot Analysis Table 3 shows the comparison between the parameters concerning the phenomenon of RER overshoot during the recovery phase between study groups. All included patients showed an increase, defined as an overshoot of the RER after exercise. Although during exercise patients with CHD and controls showed a similar RER peak, the behavior of the RER during recovery was significantly different. Moreover, RER max, RER mag, and RER slope revealed a lower RER overshoot for patients with CHD when compared to the healthy controls (Figure 2). However, no statistically significant difference was found for the RER peak, RER max, RER slope and Time to RER max parameters when CHD subgroups were compared. Only the RER mag was higher in the controls and patients with CoA when compared with the Fon, ToF, and TGA groups. The correlations between RER overshoot parameters during the recovery phase and some of the main cardiorespiratory fitness and efficiency indices were assessed (Table 4). HR peak showed significant correlations with both RER max (ρ = 0.323; $p \leq 0.001$) and RER mag ($r = 0.366$; $p \leq 0.001$). A significant negative correlation between RER mag and HR/VO2 slope was displayed, as well as a positive correlation between RER max and HRRec after one minute. Although the time to RER max and the RER slope showed slight correlations with the principal CPET parameters, RER max and RER mag were significantly correlated with important cardiorespiratory fitness and efficiency indices, such as VO2 peak and OUES. Moreover, when grouped by ventilatory classes, the RER recovery parameters, except for time to RER max, were significantly higher in patients of ventilatory class I compared with patients belonging to ventilatory classes II and III (Figure 3). No statistically significant correlations between the RER recovery parameters and resting echocardiographic data were found, except between RER max and TAPSE. ## 4. Discussion To the best of the authors’ knowledge, this is the first study evaluating the overshoot parameters of the respiratory gas exchange and specifically the behavior of the RER during recovery from maximal CPET in young patients with CHD. The main results of the present study are the following:All patients with CHD presented an overshoot of the RER during recovery after maximal CPET.Patients with CHD showed reduced RER recovery overshoot compared to healthy subjects. Although there are significant differences regarding the cardiopulmonary response during exercise between the subgroups of CHD, no differences in the RER recovery parameters were evident. RER recovery parameters significantly correlated with the most important cardiorespiratory fitness and efficiency indices, independently from the RER peak reached during exercise. ## 4.1. Why Is the CPET Recovery Phase Relevant in Patients with CHD? Currently, the cardiopulmonary response during exercise has been widely studied in different populations but there is still little evidence of the CPET parameters’ behavior during the recovery phase [8]. Some authors described delayed kinetics of VO2 recovery in patients with HF after maximal and submaximal incremental exercise testing compared to healthy subjects [13,23,24], showing that these findings were associated with a worse prognosis in these patients [24,25]. Slow recovery of energy stores in skeletal muscles was deemed to be responsible for the delayed VO2 recovery [26]. In addition, in patients with HF, a delayed recovery of VE and VCO2 was also found. This phenomenon has been attributed to the retention of CO2 in the muscles after exercise, justifying the consequent increase in ventilation to maintain a state of eucapnia [24]. In this regard, it is noteworthy to underline that parameters describing the recovery phase seemed to have a more significant correlation than peak values with muscle strength in both healthy controls and patients with HF [27]. In the latter, the phenomenon of VO2 overshoot during the first part of the recovery phase has also been described: it is defined by a further VO2 growth compared to the peak values [28]. This overshoot has been found in some cardiac patients and it seems to be associated with a worse prognosis [29]. Other authors also found a paradoxical increase in cardiac output in the recovery phase after CPET [11], which may explain VO2 overshoot. This increase in cardiac output would be attributable to a reduction of peripheral vascular resistances at the end of the exercise but also to the contribution of skeletal muscles to repay the oxygen debt or to a relatively slower decline in the blood concentration of catecholamines during recovery [11,30]. More recently, the overshoot phenomenon of gas exchange indices, such as RER and VE/VO2 during the recovery phase after maximal CPET, has been described [12]. Takayanagi et al. identified an attenuation of this phenomenon in patients with HF when compared to healthy subjects [12]. The overshoot of gas exchange indices seems to be a direct consequence of VE and VCO2 returning to normal more slowly than VO2, due to the carbon dioxide deposits produced by the anaerobic metabolism during exercise [12]. Some authors reported a delay in gas exchange recovery in patients with different types of CHD, but they mainly focused on the VO2 recovery kinetics [16,17]. Since most of the literature concerning the evaluation of CHD by CPET focuses on the exercise phase, this study aimed to analyze the behavior of the main cardiopulmonary indices during recovery in a population of young patients with CHD and to compare it with an age-matched healthy population. The RER was chosen as the most suitable parameter to be evaluated as it reflects simultaneously both the VO2 and VCO2 trend, with possibly slowed kinetics during the recovery phase [16,17,31]. ## 4.2. Exercise Phase Patients with CHD are subjects who, despite the improvement in medical and surgical therapies that occurred over the last decades, are still forced to live their whole life with the pathophysiological alterations due to their disease and to the sequelae of surgical interventions. These alterations mainly involve the cardiovascular system with consequent functional limitations, but, in complex CHD the whole oxygen transport system might be affected [32,33]. Therefore, a comprehensive functional evaluation with maximal CPET is strongly recommended in current guidelines [34]. Patients with CHD have reduced cardiorespiratory fitness and efficiency during exercise when compared to healthy controls. This is widely demonstrated in the literature and confirms the possible presence of cardiogenic limitations to exercise in patients with CHD [35]. Indeed, children and adults with CHD, in particular after surgical repair, have lower maximal aerobic and functional capacity compared to controls [36,37]. A recent systematic review and meta-analysis investigating children and adolescents with CHD reported a lower exercise capacity and cardiorespiratory efficiency compared with healthy controls as shown by worse VO2 peak, maximal power, VE/VCO2 slope, O2 pulse, and HR max [38]. Moreover, exercise capacity differs significantly across the spectrum of CHD [35]. Simple CHD present usually better exercise parameters compared to complex CHD. In our study, CoA patients had a higher HR peak compared to ToF and Fon patients. Moreover, Fon patients presented mild chronotropic incompetence compared to other CHD subgroups, probably related to an abnormal cardiac filling rather than sinoatrial node dysfunction. In accordance with previous studies, patients with complex CHD, in particular patients with univentricular circulation who underwent a Fontan procedure, presented the lowest VO2 peak and highest VE/VCO2 slope values [39]. On the other hand, patients with CoA and patients with TGA, in particular those who received an arterial switch operation, were found to have the highest VO2 peak and lowest VE/VCO2 slope values [40]. Our work confirms this previous evidence with Fon patients presenting the lowest maximal and submaximal cardiorespiratory fitness during exercise. Indeed, a proportion of them showed desaturation at rest and during exercise, with the worst values for aerobic capacity and ventilatory efficiency among the CHD groups investigated. Moreover, despite patients with TGA presenting a lesion with high complexity, restoring correct anatomy and physiology during early life seem to ensure an almost normal exercise capacity and cardiorespiratory fitness [35]. ## 4.3. Recovery Phase The focus of this study was related to the RER recovery parameters after maximal CPET. Patients with CHD presented significantly reduced RER max, RER mag, and RER slope compared to healthy controls, despite the RER peak value during exercise being comparable between the two groups (1.22 ± 0.11 vs. 1.23 ± 0.12). These findings seem to confirm that, in subjects with cardiogenic limitations to physical exercises, such as patients with CHD, the RER overshoot phenomenon appears to be reduced [12]. Although the included chronic conditions do not share the same pathophysiological mechanism, it is possible that the underlying cardiac impairment leading to the reduced RER overshoot of patients with CHD may be similar to what has been described in patients with HF [12]. A delay in VO2 recovery kinetics and HR recovery has already been demonstrated in young patients with different CHD [16]. An impaired right-sided hemodynamic and central autonomic nervous activity may lead to a delay in recovery indices [18] with possible implications also for clinical decision-making [19]. Most studies on patients with CHD have focused on adult populations [40]. In one of the few works investigating the recovery phase of a young population with CHD, patients with ToF presented diminished exercise capacity and slower recovery of VO2 and VCO2 compared to healthy subjects. Those patients with the worst exercise capacity also showed the slowest recovery indices [31]. This delay seemed to correlate to ventricular contractility indices, suggesting the crucial role of ventricular function during the recovery after physical exercise [31]. Interestingly, patients with different classes of CHD did not show significant variances in RER recovery indices. This confirms previous studies on adults showing that gas exchange recovery after exercise testing is prolonged in patients with CHD, independently of the congenital heart lesion [17]. However, comparing our results with prior studies, RER mag of young patients with CHD (44.4 ± $14.8\%$) was higher compared to older patients with HF, kidney transplant recipients but also healthy older subjects (21.4 ± $12.4\%$, 28.4 ± $12.7\%$ and 29.3 ± $10\%$, respectively) [12,41]. These findings suggest that age appears to be a crucial factor in determining this phenomenon in the recovery phase, as subjects with the established cardiac disease show higher values than healthy older subjects, even in CHD with the lowest cardiorespiratory fitness (RER mag in Fon group: 42.31 ± $13.10\%$). ## 4.4. RER Overshoot and Cardiorespiratory Fitness/Efficiency The RER recovery indices showed interesting correlations with cardiorespiratory efficiency even in CHD, corroborating that it is possible to implement the CPET evaluation in this clinical population. HR/VO2 slope describes the subject’s ability to adequately raise HR to meet the increased metabolic demands during exercise, and it is a cardiocirculatory efficiency index that has been poorly studied in the literature so far, particularly in patients with CHD [42]. A significant negative correlation was found between RER mag and HR/VO2 slope (r = −0.232, $$p \leq 0.004$$). It could be hypothesized that patients with a hyperkinetic response during exercise and thus lower cardiocirculatory efficiency have also a reduced RER overshoot in the recovery phase, probably due to cardiac limitations. Furthermore, patients with better exercise tolerance and thus higher HR peak have shown a more significant RER overshoot. It needs to be investigated whether this observation is due to cardiac limitations regarding the chronotropic response or simply due to lower exercise tolerance. The RER mag showed significant correlations with relevant indices of cardiorespiratory fitness and efficiency such as VO2 peak and OUES, which were comparable with those previously described for patients with HF [12]. This shows how the overshoot phenomenon is closely related to maximal and submaximal aerobic capacity. Different from previous works, no correlation between RER mag and VE/VCO2 slope was found [12,41]. This could be explained by the fact that the population in this study was young and might not have yet developed a relevant degree of ventilatory-perfusion mismatch. Alternatively, since RER peak and RER max seem to negatively correlate with VE/VCO2 slope, these data may suggest that ventilatory-perfusion mismatch has a huger impact on exercise tolerance and affects less the recovery phase [43]. To evaluate the potential clinical application of the RER overshoot, patients with CHD were grouped according to their ventilatory classes, which reflect the cardiorespiratory efficiency and a possible ventilation-perfusion mismatch during exercise. Patients with better ventilatory classes showed higher RER recovery overshoots compared with those belonging to worse ventilatory classes, similar to what was reported about kidney transplant recipients [41]. Furthermore, a vigorous RER overshoot seems to be an index of better cardiorespiratory performance and a better prognosis in patients with CHD. This supports the proposal that the analysis of CPET metrics during recovery may provide valid additional information for the test interpretation. Finally, correlations between RER overshoot and resting biventricular function were investigated. No significant correlations between RER mag and echocardiographic parameters were found, as previously reported between RER mag and LVEF in patients with HF, suggesting the absence of a direct relationship between RER overshoot and ventricular function at rest [12]. However, there is still the need to study how limitations in the response of cardiac output during exercise may influence the CPET overshoot recovery parameters. In this regard, data from invasive and/or non-invasive measurements of cardiac output during exercise could be useful to better understand the direct impact of cardiac limitations on these recovery metrics. ## 4.5. Limitations and Perspectives This was a retrospective study assessing the recovery phase after maximal CPET based on routinely performed clinical assessments. The sample size was limited because a long-lasting evaluation of the recovery phase with gas exchange data was not routinely performed in our laboratory before January 2018, when a dedicated recovery protocol was created. A larger sample and specific trials are needed to investigate the impact of ventricular function (during rest and exercise) on the RER recovery overshoot, as echocardiographic data have been assessed for clinical purposes and were thus not available for all patients. Moreover, those patients with a peak RER < 1.1 were excluded from the study, for consistency with previous literature and to avoid possible confounding in the assessment of the recovery CPET parameters, especially RER. The recovery phase after exercise has been poorly explored in pathological populations and often with heterogeneous methodologies. The present study could help to highlight the possibility of incorporating and standardizing variables of the recovery phase in CPET interpretation, aiming to improve the diagnostic and prognostic stratification of these patients. Future trials should analyze the behavior of gas exchange indices after maximal exercise testing in populations with different functional limitations, to improve the understanding of the pathophysiological mechanisms that determine its behavior and the clinical interpretation of this phenomenon. Finally, further studies are needed, aiming to prospectively investigate the prognostic value of the RER overshoot parameters on hard clinical endpoints. ## 5. Conclusions The present study highlights the role of functional assessment in patients with CHD. An overshoot of RER during recovery after maximal CPET is commonly observed in young patients with CHD but this phenomenon appears to be lower compared to healthy controls, suggesting a possible connection with cardiogenic limitations during exercise. Indeed, RER mag was different in the study’s subgroups where healthy subjects and patients with aortic coarctation showed a significantly higher RER overshoot compared to patients with TGA, previous Fontan procedure, and ToF. RER recovery overshoots correlated with prognostically relevant CPET indices of cardiorespiratory fitness and efficiency, showing lower values in patients with significant ventilatory-perfusion mismatch. The evaluation of CPET recovery parameters should be further investigated and implemented in clinical settings to increase scientific evidence and provide additional information to improve risk stratification in patients with CHD. ## References 1. van der Linde D., Konings E.E.M., Slager M.A., Witsenburg M., Helbing W.A., Takkenberg J.J.M., Roos-Hesselink J.W.. Birth Prevalence of Congenital Heart Disease Worldwide: A Systematic Review and Meta-Analysis. J. Am. Coll Cardiol. (2011) 58 2241-2247. DOI: 10.1016/j.jacc.2011.08.025 2. Dolk H., Loane M., Garne E.. Congenital Heart Defects in Europe: Prevalence and Perinatal Mortality, 2000 to 2005. Circulation (2011) 123 841-849. DOI: 10.1161/CIRCULATIONAHA.110.958405 3. Best K.E., Rankin J.. Long-Term Survival of Individuals Born With Congenital Heart Disease: A Systematic Review and Meta-Analysis. J. Am. Heart Assoc. (2016) 5. DOI: 10.1161/JAHA.115.002846 4. Oliver J.M., Gallego P., Gonzalez A.E., Garcia-Hamilton D., Avila P., Yotti R., Ferreira I., Fernandez-Aviles F.. Risk Factors for Excess Mortality in Adults with Congenital Heart Diseases. Eur. Heart J. (2017) 38 1233-1241. DOI: 10.1093/eurheartj/ehw590 5. Amedro P., Gavotto A., Guillaumont S., Bertet H., Vincenti M., de La Villeon G., Bredy C., Acar P., Ovaert C., Picot M.C.. Cardiopulmonary Fitness in Children with Congenital Heart Diseases versus Healthy Children. Heart (2018) 104 1026-1036. DOI: 10.1136/heartjnl-2017-312339 6. Warnes C.A., Williams R.G., Bashore T.M., Child J.S., Connolly H.M., Dearani J.A., del Nido P., Fasules J.W., Graham T.P., Hijazi Z.M.. ACC/AHA 2008 Guidelines for the Management of Adults with Congenital Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Develop Guidelines on the Management of Adults with Congenital Heart Disease). Circulation (2008) 118 e714-e833. DOI: 10.1161/CIRCULATIONAHA.108.190690 7. Wadey C.A., Weston M.E., Dorobantu D.M., Pieles G.E., Stuart G., Barker A.R., Taylor R.S., Williams C.A.. The Role of Cardiopulmonary Exercise Testing in Predicting Mortality and Morbidity in People with Congenital Heart Disease: A Systematic Review and Meta-Analysis. Eur. J. Prev. Cardiol. (2022) 29 513-533. DOI: 10.1093/eurjpc/zwab125 8. Guazzi M.. “Recovering” the Recognition for VO2 Kinetics During Exercise Recovery in Heart Failure: A Good Practice in Need of More Exercise. J. ACC Heart Fail. (2018) 6 340-342. DOI: 10.1016/j.jchf.2018.01.020 9. Guazzi M., Bandera F., Ozemek C., Systrom D., Arena R.. Cardiopulmonary Exercise Testing: What Is Its Value?. J. Am. Coll Cardiol. (2017) 70 1618-1636. DOI: 10.1016/j.jacc.2017.08.012 10. Arena R., Myers J., Guazzi M.. Cardiopulmonary Exercise Testing Is a Core Assessment for Patients With Heart Failure. Congest. Heart Fail. (2011) 17 115-119. DOI: 10.1111/j.1751-7133.2011.00216.x 11. Tanabe Y., Takahashi M., Hosaka Y., Ito M., Ito E., Suzuki K.. Prolonged Recovery of Cardiac Output after Maximal Exercise in Patients with Chronic Heart Failure. J. Am. Coll Cardiol. (2000) 35 1228-1236. DOI: 10.1016/S0735-1097(00)00517-9 12. Takayanagi Y., Koike A., Nagayama O., Nagamine A., Qin R., Kato J., Nishi I., Himi T., Kato Y., Sato A.. Clinical Significance of the Overshoot Phenomena of Respiratory Gas Indices during Recovery from Maximal Exercise Testing. J. Cardiol. (2017) 70 598-606. DOI: 10.1016/j.jjcc.2017.03.012 13. Bailey C.S., Wooster L.T., Buswell M., Patel S., Pappagianopoulos P.P., Bakken K., White C., Tanguay M., Blodgett J.B., Baggish A.L.. Post-Exercise Oxygen Uptake Recovery Delay: A Novel Index of Impaired Cardiac Reserve Capacity in Heart Failure. JACC Heart Fail. (2018) 6 329-339. DOI: 10.1016/j.jchf.2018.01.007 14. Falk B., Dotan R.. Child-Adult Differences in the Recovery from High-Intensity Exercise. Exerc. Sport Sci. Rev. (2006) 34 107-112. DOI: 10.1249/00003677-200607000-00004 15. Zanconato S., Cooper D.M., Armon Y.. Oxygen Cost and Oxygen Uptake Dynamics and Recovery with 1 Min of Exercise in Children and Adults. J. Appl. Physiol. (1991) 71 993-998. DOI: 10.1152/jappl.1991.71.3.993 16. Singh T.P., Alexander M.E., Gauvreau K., Curran T., Rhodes Y., Rhodes J.. Recovery of Oxygen Consumption after Maximal Exercise in Children. Med. Sci. Sports Exerc. (2011) 43 555-559. DOI: 10.1249/MSS.0b013e3181f683b8 17. Greutmann M., Rozenberg D., Le T.L., Silversides C.K., Granton J.T.. Recovery of Respiratory Gas Exchange after Exercise in Adults with Congenital Heart Disease. Int. J. Cardiol. (2014) 176 333-339. DOI: 10.1016/j.ijcard.2014.06.033 18. Lurz P., Riede F.T., Taylor A.M., Wagner R., Nordmeyer J., Khambadkone S., Kinzel P., Derrick G., Schuler G., Bonhoeffer P.. Impact of Percutaneous Pulmonary Valve Implantation for Right Ventricular Outflow Tract Dysfunction on Exercise Recovery Kinetics. Int. J. Cardiol. (2014) 177 276-280. DOI: 10.1016/j.ijcard.2014.09.014 19. Giardini A., Specchia S., Coutsoumbas G., Donti A., Formigari R., Fattori R., Oppido G., Gargiulo G., Picchio F.M.. Impact of Pulmonary Regurgitation and Right Ventricular Dysfunction on Oxygen Uptake Recovery Kinetics in Repaired Tetralogy of Fallot. Eur. J. Heart Fail. (2006) 8 736-743. DOI: 10.1016/j.ejheart.2006.01.012 20. Neunhaeuserer D., Battista F., Mazzucato B., Vecchiato M., Meneguzzo G., Quinto G., Niebauer J., Gasperetti A., Vida V., di Salvo G.. Exercise Capacity and Cardiorespiratory Fitness in Children with Congenital Heart Diseases: A Proposal for an Adapted NYHA Classification. Int. J. Environ. Res. Public Health (2022) 19. DOI: 10.3390/ijerph19105907 21. Hollenberg M., Tager I.B.. Oxygen Uptake Efficiency Slope: An Index of Exercise Performance and Cardiopulmonary Reserve Requiring Only Submaximal Exercise. J. Am. Coll Cardiol. (2000) 36 194-201. DOI: 10.1016/S0735-1097(00)00691-4 22. Vecchiato M., Neunhaeuserer D., Quinto G., Bettini S., Gasperetti A., Battista F., Vianello A., Vettor R., Busetto L., Ermolao A.. Cardiopulmonary Exercise Testing in Patients with Moderate-Severe Obesity: A Clinical Evaluation Tool for OSA?. Sleep Breath. (2021) 26 1115-1123. DOI: 10.1007/s11325-021-02475-0 23. Nanas S., Nanas J., Kassiotis C., Nikolaou C., Tsagalou E., Sakellariou D., Terovitis I., Papazachou O., Drakos S., Papamichalopoulos A.. Early Recovery of Oxygen Kinetics after Submaximal Exercise Test Predicts Functional Capacity in Patients with Chronic Heart Failure. Eur. J. Heart Fail. (2001) 3 685-692. DOI: 10.1016/S1388-9842(01)00187-8 24. Cohen-Solal A., Laperche T., Morvan D., Geneves M., Caviezel B., Gourgon R.. Prolonged Kinetics of Recovery of Oxygen Consumption After Maximal Graded Exercise in Patients With Chronic Heart Failure. Circulation (1995) 91 2924-2932. DOI: 10.1161/01.CIR.91.12.2924 25. Scrutinio D., Passantino A., Lagioia R., Napoli F., Ricci A., Rizzon P.. Percent Achieved of Predicted Peak Exercise Oxygen Uptake and Kinetics of Recovery of Oxygen Uptake after Exercise for Risk Stratification in Chronic Heart Failure. Int. J. Cardiol. (1998) 64 117-124. DOI: 10.1016/S0167-5273(98)00019-9 26. Belardinelli R., Barstow T.J., Nguyen P., Wasserman K.. Skeletal Muscle Oxygenation and Oxygen Uptake Kinetics Following Constant Work Rate Exercise in Chronic Congestive Heart Failure. Am. J. Cardiol. (1997) 80 1319-1324. DOI: 10.1016/S0002-9149(97)00672-3 27. Zavin A., Arena R., Joseph J., Allsup K., Daniels K., Christian Schulze P., Lecker S., Forman D.E.. Dynamic Assessment of Ventilatory Efficiency during Recovery from Peak Exercise to Enhance Cardiopulmonary Exercise Testing. Eur. J. Prev. Cardiol. (2013) 20 779-785. DOI: 10.1177/2047487312446563 28. Suzuki T., Koike A., Nagayama O., Sakurada K., Tsuneoka H., Kato J., Yamashita T., Yamazaki J.. Overshoot Phenomena of Respiratory Gas Variables During Exercise Recovery in Cardiac Patients. Circ. J. (2012) 76 876-883. DOI: 10.1253/circj.CJ-11-1279 29. Fortin M., Turgeon P.Y., Nadreau É., Grégoire P., Maltais L.G., Sénéchal M., Provencher S., Maltais F.. Prognostic Value of Oxygen Kinetics During Recovery From Cardiopulmonary Exercise Testing in Patients With Chronic Heart Failure. Can. J. Cardiol. (2015) 31 1259-1265. DOI: 10.1016/j.cjca.2015.02.015 30. Toste A., Soares R., Feliciano J., Andreozzi V., Silva S., Abreu A., Ramos R., Santos N., Ferreira L., Ferreira R.C.. Prognostic Value of a New Cardiopulmonary Exercise Testing Parameter in Chronic Heart Failure: Oxygen Uptake Efficiency at Peak Exercise—Comparison with Oxygen Uptake Efficiency Slope. Rev. Port. De Cardiol. (2011) 30 781-787. DOI: 10.1016/S0870-2551(11)70026-9 31. Coomans I., de Kinder S., van Belleghem H., de Groote K., Panzer J., de Wilde H., Mosquera L.M., François K., Bové T., Martens T.. Analysis of the Recovery Phase after Maximal Exercise in Children with Repaired Tetralogy of Fallot and the Relationship with Ventricular Function. PLoS ONE (2020) 15. DOI: 10.1371/journal.pone.0244312 32. Rychik J., Goldberg D.J.. Late Consequences of the Fontan Operation. Circulation (2014) 130 1525-1528. DOI: 10.1161/CIRCULATIONAHA.114.005341 33. Feinstein J.A., Benson D.W., Dubin A.M., Cohen M.S., Maxey D.M., Mahle W.T., Pahl E., Villafae J., Bhatt A.B., Peng L.F.. Hypoplastic Left Heart Syndrome: Current Considerations and Expectations. J. Am. Coll Cardiol. (2012) 59 S1-S42. DOI: 10.1016/j.jacc.2011.09.022 34. Stout K.K., Daniels C.J., Aboulhosn J.A., Bozkurt B., Broberg C.S., Colman J.M., Crumb S.R., Dearani J.A., Fuller S., Gurvitz M.. 2018 AHA/ACC Guideline for the Management of Adults With Congenital Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation (2019) 139 e698-e800. DOI: 10.1161/CIR.0000000000000603 35. Kempny A., Dimopoulos K., Uebing A., Moceri P., Swan L., Gatzoulis M.A., Diller G.P.. Reference Values for Exercise Limitations among Adults with Congenital Heart Disease. Relation to Activities of Daily Life--Single Centre Experience and Review of Published Data. Eur. Heart J. (2012) 33 1386-1396. DOI: 10.1093/eurheartj/ehr461 36. Inuzuka R., Diller G.P., Borgia F., Benson L., Tay E.L.W., Alonso-Gonzalez R., Silva M., Charalambides M., Swan L., Dimopoulos K.. Comprehensive Use of Cardiopulmonary Exercise Testing Identifies Adults with Congenital Heart Disease at Increased Mortality Risk in the Medium Term. Circulation (2012) 125 250-259. DOI: 10.1161/CIRCULATIONAHA.111.058719 37. Kipps A.K., Graham D.A., Harrild D.M., Lewis E., Powell A.J., Rhodes J.. Longitudinal Exercise Capacity of Patients with Repaired Tetralogy of Fallot. Am. J. Cardiol. (2011) 108 99-105. DOI: 10.1016/j.amjcard.2011.02.349 38. Villaseca-Rojas Y., Varela-Melo J., Torres-Castro R., Vasconcello-Castillo L., Mazzucco G., Vilaró J., Blanco I.. Exercise Capacity in Children and Adolescents With Congenital Heart Disease: A Systematic Review and Meta-Analysis. Front. Cardiovasc. Med. (2022) 9 990. DOI: 10.3389/fcvm.2022.874700 39. Fredriksen P.M., Veldtman G., Hechter S., Therrien J., Chen A., Warsi M.A., Freeman M., Liu P., Siu S., Thaulow E.. Aerobic Capacity in Adults with Various Congenital Heart Diseases. Am. J. Cardiol. (2001) 87 310-314. DOI: 10.1016/S0002-9149(00)01364-3 40. Diller G.P., Dimopoulos K., Okonko D., Li W., Babu-Narayan S.V., Broberg C.S., Johansson B., Bouzas B., Mullen M.J., Poole-Wilson P.A.. Exercise Intolerance in Adult Congenital Heart Disease: Comparative Severity, Correlates, and Prognostic Implication. Circulation (2005) 112 828-835. DOI: 10.1161/CIRCULATIONAHA.104.529800 41. Patti A., Neunhaeuserer D., Gasperetti A., Baioccato V., Vecchiato M., Battista F., Marchini F., Bergamin M., Furian L., Ermolao A.. Overshoot of the Respiratory Exchange Ratio during Recovery from Maximal Exercise Testing in Kidney Transplant Recipients. Int. J. Environ. Res. Public Health (2021) 18. DOI: 10.3390/ijerph18179236 42. Neunhaeuserer D., Gasperetti A., Savalla F., Gobbo S., Bullo V., Bergamin M., Foletto M., Vettor R., Zaccaria M., Ermolao A.. Functional Evaluation in Obese Patients Before and After Sleeve Gastrectomy. Obes. Surg. (2017) 27 3230-3239. DOI: 10.1007/s11695-017-2763-x 43. Malhotra R., Bakken K., D’Elia E., Lewis G.D.. Cardiopulmonary Exercise Testing in Heart Failure. JACC Heart Fail. (2016) 4 607-616. DOI: 10.1016/j.jchf.2016.03.022
--- title: High-Throughput Quantitative Screening of Glucose-Stimulated Insulin Secretion and Insulin Content Using Automated MALDI-TOF Mass Spectrometry authors: - Clément Philippe Delannoy - Egon Heuson - Adrien Herledan - Frederik Oger - Bryan Thiroux - Mickaël Chevalier - Xavier Gromada - Laure Rolland - Philippe Froguel - Benoit Deprez - Sébastien Paul - Jean-Sébastien Annicotte journal: Cells year: 2023 pmcid: PMC10047017 doi: 10.3390/cells12060849 license: CC BY 4.0 --- # High-Throughput Quantitative Screening of Glucose-Stimulated Insulin Secretion and Insulin Content Using Automated MALDI-TOF Mass Spectrometry ## Abstract Type 2 diabetes (T2D) is a metabolic disorder characterized by loss of pancreatic β-cell function, decreased insulin secretion and increased insulin resistance, that affects more than 537 million people worldwide. Although several treatments are proposed to patients suffering from T2D, long-term control of glycemia remains a challenge. Therefore, identifying new potential drugs and targets that positively affect β-cell function and insulin secretion remains crucial. Here, we developed an automated approach to allow the identification of new compounds or genes potentially involved in β-cell function in a 384-well plate format, using the murine β-cell model Min6. By using MALDI-TOF mass spectrometry, we implemented a high-throughput screening (HTS) strategy based on the automation of a cellular assay allowing the detection of insulin secretion in response to glucose, i.e., the quantitative detection of insulin, in a miniaturized system. As a proof of concept, we screened siRNA targeting well-know β-cell genes and 1600 chemical compounds and identified several molecules as potential regulators of insulin secretion and/or synthesis, demonstrating that our approach allows HTS of insulin secretion in vitro. ## 1. Introduction Type 2 Diabetes (T2D) is characterized by high blood glucose levels and develops due to inadequate pancreatic β-cell function (i.e., insulin secretion) and peripheral insulin resistance. In Europe, the global prevalence of diabetes is currently estimated at $8\%$ of the population, with T2D representing about $90\%$ of cases [1]. Although lifestyle modification is the first reference treatment for T2D, modifying the patient’s habits is often ineffective to stabilize glycemia and combinations of pharmacological treatments (metformin, sulfonylurea, incretin enhancers, GLP-1 mimetics, etc.) are often required to treat T2D [2,3]. Yet, long-term control of glycemia remains a challenge for most patients, particularly when β-cell capacity to secrete insulin decreases with age [4]. In the context of research aimed at discovering new therapeutic targets for T2D, the identification of new natural or synthetic compounds or target genes that have the capacity to restore or increase insulin secretion and content is essential. Therefore, developing high-throughput screening (HTS) strategies to rapidly measure insulin secretion or synthesis is crucial to address the unmet need of new efficient molecules for T2D patients. Currently, several strategies have been developed based on assays aiming at measuring secreted hormones or specific proteins (e.g., insulin or c-peptide) [5], mainly through Enzyme-Linked Immuno Sorbent Assay (ELISA) approaches. However, this technique, although widely used in academic research and clinical diagnosis, has several limitations, such as the technical detection of the target protein, its precise quantification, the duration of the experiment and the costs of the reagents. Altogether, these limitations preclude the use of ELISA strategies for real HTS approaches to evaluate the action of a large collection of potential drugs that can stimulate insulin secretion. Along with ELISA, complementary techniques for the high-throughput screening of insulin secretion have recently emerged with a high potential for T2D therapy. To overcome the experimental and technical limits of ELISA, several laboratories have developed genetically engineered cellular tools to measure β-cell function. In these modified cells, the use of immunofluorescence microscopy or luciferase-based approaches allows the measurement of a modified C-peptide [5,6,7,8,9]. Although easy to handle, these genetically modified tools partially address the functionality of the β-cell since they do not directly assess the level of production of endogenous insulin content or the secretion of insulin in response to a physiological stimulation such as glucose. Therefore, to tackle these limitations, we developed a new strategy aiming at implementing an automated approach that could allow HTS of drugs that efficiently modulate glucose-stimulated insulin secretion. This automated process, using mass spectrometry to directly measure insulin, is not only faster but also less expensive, enabling for the first time the implementation of high-throughput exploratory strategies to identify new biological targets or bioactive compounds through the screening of chemical or siRNA libraries. Here, we describe a high-throughput screening approach based on matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) to quantify insulin production and secretion in a mouse model of β-cells, the Min6 cell line. We automated, in a 384-well format, siRNA-based loss-of-function of candidate genes to study their effect on insulin secretion. Finally, as a proof of concept, we screened more than 1600 chemical compounds and identified several drugs that modulate insulin secretion or content. ## 2.1. Cell Culture and Treatments MIN6 cells (Addexbio) were maintained in 25 mM glucose, Glutamax DMEM medium (Gibco, 31966-021, Waltham, MA, USA), supplemented with $15\%$ heat-inactivated fetal bovine, 100 μg/mL penicillin-streptomycin and 55 μM beta-mercaptoethanol (Gibco) and cultured in a humidified atmosphere with $5\%$ CO2 at 37 °C. Cells were seeded at 20,000 cells/well using a Multidrop Combi dispenser (Thermo Fisher Scientific, Waltham, MA, USA) into 384-well plate black µclear. Cells were then treated with different chemical compounds and/or siRNA, as described below (see Supplementary Table S1 for the list of siRNA used in this study). ## 2.2. Automated Glucose-Stimulated Insulin Secretion (GSIS) in 384-Well Format For GSIS experiments, Min6 cells were plated in a 384-well plate (2.104 cells/well) and incubated in 80 µL of starvation buffer Krebs Ringer buffer (KRB) [NaCl 115 mM; KCl 4.7 mM; CaCl2 2.6 mM; NaHCO3 20 mM; MgSO4 1.2 mM; KH2PO4 1.2 mM; HEPES 16 mM] supplemented with $0.5\%$ fatty-acid free BSA for 1 h at 37 °C and $5\%$ CO2. The automated GSIS pipeline protocol started on the BioCel platform system (Agilent, Santa Clara, CA, USA) including integrated devices for incubation, washing, distribution, pipetting the liquid handling system, microplate stacking or sealing systems. All the tasks were sequentially operated on each device using the Direct Drive Robot (DDR) arm (Agilent Biotechnologies) positioned at the center of the platform. At the beginning of the run, microplates placed at 37 °C in $5\%$ CO2 atmosphere in the Incubator (Liconic, Mauren, Liechtenstein) were conveyed via a telescopic lift arm on the deck of the platform. The cells were first washed from their culture medium by 4 washes of KRB-BSA with the EL406 washer distributor liquid handling device (Biotek, Agilent, Santa Clara, CA, USA) suitable for 384-well plates. Control of the dispense (with tubes angled to 20°) and aspiration of liquid due to the dual-action manifold system of the device permitted the complete reduction of the loose of cell layers during this washing step. The cells were then incubated for 1 h at 37 °C in KRB-BSA for the starvation stage. To avoid BSA interferences in MALDI-TOF mass spectrometry analysis, Min6 cells were washed 5 times with BSA-free KRB buffer supplemented with 2.8 mM glucose after starvation using a Biotek washing robot. After 1 h at 37 °C, cells were washed 5 times with KRB buffer and supplemented with 2.8 mM glucose by distribution of a solution of KRB-glucose using the disposable peristaltic pump cassette system (EL406 washer distributor), for 1 h at 37 °C and $5\%$ CO2. Then, 40 µL of the supernatant was collected for insulin quantification in 2.8 mM glucose conditions using a Bravo liquid handler and cells were subsequently incubated in 80 µL of KRB buffer containing 20 mM glucose for 1 h at 37 °C and $5\%$ CO2. A 100 µL sample of the supernatant was collected for insulin quantification in 20 mM glucose conditions. The intracellular insulin content was recovered in 40 µL of lysis buffer containing $75\%$ ethanol and $1.5\%$ HCl. Microplates with all collected samples were sealed with a PlateLoc Thermal Microplate Sealer (Agilent Technologies) after the collecting step with Bravo and frozen at −80 °C until analysis. Insulin concentration was measured through ELISA according to the manufacturer’s instructions (Mercodia, Uppsala, Sweden) or mass spectrometry as described below. Samples were frozen at −80 °C before further processing. ## 2.3. Automated siRNA Reverse Transfection in 384-Well Format To validate a potential functional effect on GSIS, we selected two siRNA targeting genes previously known to control insulin secretion (ON-TARGETplus SMARTpool siRNA targeting mouse Pdx1 (L-040402-01-0005) and mouse Kcnj11 (L-042183-00-0005)) and two controls (non-targeting negative controls and siGLO fluorescent control). siRNA transfection was performed by reverse transfection using $0.375\%$ Dharmafect1 transfection reagent (GE Dharmacon) and 50 nM siRNA. siRNAs (200 nL of a 20 µM stock; 0.16 pmol/well) were dispensed into 384-well assay plates (Greiner Bio-One; 78109) using an Echo 550 Series Liquid Handler (Labcyte, San José, CA, USA). On the day of assay, plates that contained siRNAs were thawed and equilibrated to room temperature. Dharmafect1 transfection reagent (0.3 µL/well in DMEM media, Gibco) was added to the assay plates using a Multidrop Combi dispenser (Thermo Fisher Scientific, Waltham, MA, USA), with a standard tube dispensing cassette. After 20 min of incubation at room temperature, cells (65 µL of 300,000 cells/mL; 20,000 cells/well) were added to the assay plates with the Multidrop Combi dispenser and standard tube dispensing cassette. Assay plates were incubated at 37 °C, $5\%$ CO2 in a controlled-humidity incubator and cultured for 48 h before GSIS assays. Samples were frozen at −80 °C before further processing. ## 2.4. Automated Incubation with Chemical Compounds Min6 cells (80 µL of 250,000 cells/mL; 20,000 cells/well) were added to the assay plates with the Multidrop Combi dispenser (ThermoFisher) and standard tube dispensing cassette. Assay plates were incubated at 37 °C, $5\%$ CO2 in a controlled-humidity incubator during 24 h. Repaglinide (Euromedex—R1426, Souffelweyersheim, France) and diazoxyle (CliniSciences—BG0437, Nanterre, France) were used at a concentration of 100 nM and 100 µM. A library containing 1600 lead-like molecules was obtained from Asinex company (Moscow, Russia). The compounds were dissolved at a concentration of 10 mM in DMSO and one batch of this library was distributed in a 384-well LDV microplate (LP-0200, Beckman coulter, Pasadena, CA, USA). For the automated screening assay, 200 nL of 1600 Asinex compounds (5 microplates) or reference products were transferred in intermediate microplates (Greiner Bio-One; 78109, Vilvoorde, Belgium) from a source 384-well LDV microplate using a nanoacoustic transfer device (Echo 550), then stored at −20 °C until use. Eighteen hours before the automated GSIS protocol, using the BioCel platform, intermediate microplates were filled with 60 µL of cell culture medium with an EL406 washer distributor (Cassette for distribution). Then, cell microplates previously placed in the incubator were sent to the EL406 washer distributor for the aspiration step in order to leave a remaining volume of 40 µL per well. On the deck of the Bravo device, compounds diluted in medium in the intermediate microplate were mixed 4 times and a volume of 20 µL was transferred in the cell microplate for a treatment at the final concentration of 10 µM. Cells were treated overnight with different compounds and samples were frozen at −80 °C before further processing. ## 2.5. Automated MALDI-TOF Mass Spectrometry Analysis The MALDI target were prepared in an automated way using a Biomek NXp liquid handler (Beckman Coulter, Fullerton, CA, USA). The detailed robot routine is available in the Supplementary Materials. First, the 384-well microplate (Greiner) containing the previously prepared samples was placed onto the robot deck, along with the other required labware. Then, for cellular content samples only, 40 µL of MS-grade water was added to the 40 µL of samples to dilute it by a factor of 2. The dilution was performed by mixing 10 µL of the resulting solution 3 times. Then, 30 µL of a stock solution of bovine insulin (Sigma-Aldrich, Saint Louis, MO, USA) at 10 μg.mL−1 in 3 mM hydrochloric acid (pH 2.5) (for content samples, 5 μg.mL−1 for extracellular high and low glucose samples) was separated into a second 384-well microplate. In parallel, 15 µL of the MALDI matrix solution containing sinapic acid (Sigma-Aldrich) at 10 mg.mL−1 in $\frac{50}{49}$/1:acetonitrile/MS-grade water/trifluoroacetic acid (Sigma Aldrich) was added in each well of a third separate 384-well microplate. The different samples and solutions were then mixed in the following order: First, 10 µL of the bovine insulin solution was transferred into the microplate containing the samples and the two solutions were mixed by aspirating/refouling 10 µL of the mixture 10 times. Then, 15 µL of the resulting mixture was transferred into the plate containing the matrix and the two solutions were mixed by aspirating/refouling 10 µL of the mixture 10 times. Following this, 2 µL of the resulting sample/bovine insulin/matrix mixture was immediately deposited as a drop on each spot of a MALDI MTP384 Polish steel target (Bruker Daltonics, Bremen, Germany). This was made possible by the creation and 3D printing of an adapter to allow the use of the MALDI target by the robot. Its design is presented in the Supplementary Materials (Figure S1). The drops were then dried at room temperature, and the MALDI target was introduced in the source chamber of an Autoflex Speed (Bruker Daltonics, Bremen, Germany). All MALDI-TOF analyses were performed in linear positive mode using an in-house method for insulin detection (LP_5-20_kDa.par), created from the manufacturer’s automatic method LP_5-20_kDa.par. Equipment parameters were as follows: voltage values of ion sources #1 and #2 set as 19.00 and 16.50 keV, respectively; voltage values of reflectron #1 and #2 set as 21.00 and 9.50 keV, respectively; lens tension 8.00 keV; pulsed extraction 120 ns; laser intensity between 60 and $70\%$; laser global attenuator offset set to $41\%$; attenuator offset set to $32\%$; attenuator range set to $25\%$; detector gain voltage set to 2600 V (+300 V boost); Smartbeam parameter set to ultra and sample rate and digitizer settings set to 4.00 GS/s. The MS signals were acquired by summing 5000 laser shots per spectrum. Prior to each analysis, the spectrometer was calibrated using the monoisotopic values of the manufacturer’s Protein Calibration Standard I calibrant (Bruker Daltonics, Billerica, MA, USA), containing insulin ([M+H]+—m/$z = 5734.51$), ubiquitin I ([M+H]+—m/$z = 8565.76$), cytochrome C ([M+H]+—m/$z = 12$,360.97), myoglobin ([M+H]+—m/$z = 16$,952.30), cytochrome C ([M+2H]2+—m/$z = 6180.99$) and myoglobin ([M+2H]2+—m/$z = 8476.65$). The calibrant was prepared by mixing 5 µL of calibrant diluted mixture prepared according to manufacturer’s specification with 5 µL of a 10 mg.mL−1 HCCA matrix in $\frac{50}{49.9}$/0.1:acetonitrile/MS-grade water/trifluoroacetic acid, and 2 µL were then spotted on a Polished Steel 384 MALDI MTP target. Mass spectra of the sample were first visualized using FlexAnalysis software (version 3.4; Bruker Daltonics, Billerica, MA, USA), after their calibration. To produce a comprehensive document, providing the intensity and area ratio of the murine and bovine insulin, as well quality control, in a way that is simple and easily understood by non-specialist experimenters, a short program was coded. It was written in VBA (code available in the Supplementary Materials), and linked to an Excel sheet (Microsoft, Redmond, WA, USA). After an automated export for the peak list from the mass spectra using FlexAnalysis software into a dedicated Excel sheet, this program can first perform an automated quality control, based on the intensity and the mass of the bovine insulin. Then, it automatically detects the murine insulin and performs the ratio in area or intensity compared to bovine insulin, depending on the user’s preference. Finally, it sorts up all information and presents it in a clear colored view as a plate map, associated with ratio values and corresponding bar charts to quickly see the compounds that produce the highest signal compared to the blank and others. ## 2.6. Measurement of Transfection Efficiency by Flow Cytometry To assess the transfection efficiency, Min6 cells were transfected with a siGLO (Green Transfection Indicator D-001630-01-05, Dharmacon, Lafayette, CO, USA) labelled with fluorescein (CF/FAM/FITC)), using different concentrations of siGLO and transfection reagent. Transfection efficiency was determined 48 h after transfection using flow cytometry. Cells were acquired on a BD LSR Fortessa flow cytometer, and fluorescence of transfected Min6 cells was quantified by flow cytometry. ## 2.7. Immunofluorescence Images from transfected Min6 cells with siGLO and labeled with Hoechst 33342 (40 ng/mL, Thermofisher) were acquired with the microscope IN Cell Analyser 6000 (GE Healthcare Life Sciences, Buc, France) in 10× magnification in non-confocal mode with a DAPI filter set (ex.405/em.455 nm) and an FITC filter set (ex.488/em.525 nm) on the automated high content screening platform (Agilent Technologies, Santa Clara, CA, USA) (Equipex Imaginex Biomed, Institut Pasteur de Lille, France). ## 2.8. RNA Extraction and Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) Total RNA was extracted from Min6 cells using the RNeasy Plus Microkit (Qiagen, Hilden, Germany) following manufacturer’s instructions. Gene expression was measured after reverse transcription by quantitative real-time PCR (qRT-PCR) with FastStart SYBR Green master mix (Roche) using a LightCycler Nano or LC480 instruments (Roche, Basel, Switzerland). qRT-PCR results were normalized to endogenous cyclophilin reference mRNA levels as previously described [10]. The results are expressed as the relative mRNA level of a specific gene expression using the formula 2−ΔCt. The complete list of primers is presented in Supplementary Table S2. ## 2.9. Immunoblotting Experiments Cells were washed with cold PBS and lysed in radioimmunoprotein-assay (RipA) buffer (10 mM Tris/HCl, 150 mM NaCl, $1\%$ (v/v) Triton X-100, $0.5\%$ (w/v) sodium deoxycholate, $0.1\%$ (w/v) sodium dodecyl sulfate and protease inhibitors, pH 7.4), and maintained under constant agitation for 30 min. Cell extracts were then centrifuged at 16,000× g for 20 min at 4 °C). Protein concentration was determined by the BCA protein assay kit according to the manufacturer’s instructions. Equal amounts of protein were resolved on $10\%$ SDS-PAGE under reducing conditions and proteins were transferred to a nitrocellulose membrane. Blots were incubated with primary antibodies directed against PDX1 (1:1000, Abcam, ab47267, Cambridge, UK), alpha-tubulin (1:1000, Invitrogen, 32-2700, Waltham, MA, USA), washed three times with PBS-$0.05\%$ tween and followed by incubation with secondary antibodies, directed against goat anti-mouse or rabbit HRP-conjugated (1:5000, Sigma-Aldrich). The visualization of immunoreactive bands was performed using the enhanced chemiluminescence plus Western blotting detection system (GE Healthcare). Quantification of protein signal intensity was performed by volume densitometry using ImageJ 1.47t software (NIH, Bethesda, MD, USA). ## 2.10. Statistical Analysis Data are expressed as mean ± SEM. Statistical analysis were performed using GraphPad Prism software (version 9.3, Boston, MA, USA) with two-way ANOVA with Bonferroni’s post-test for multiple comparisons as indicated in the figure legends. Differences were considered statistically significant at p value < 0.05 (* < 0.05; < 0.01; * < 0.001; **** < 0.0001). For the high-throughput screens, a score similar to the statistical Z-score for each test compound was calculated using the formula: Z-score = (x − μ)/σ, where x is the murine insulin relative intensity from a compound-treated well, μ is the murine insulin relative intensity from the DMSO-treated wells on the same plate and σ is the standard deviation of the murine insulin relative intensity signals of the DMSO-treated wells across all plates. ## 3.1. Automation of Glucose-Stimulated Insulin Secretion (GSIS) Assay In order to develop an automated protocol to measure insulin secretion in the mouse insulinoma Min6 cell line, we first miniaturized the GSIS protocol in 384-well plates (Supplementary Figure S2) and validated that these cells do respond to glucose stimulation in these culture conditions. Min6 cells were seeded and, 48 h later, the cells were subjected to GSIS. For MALDI-TOF mass spectrometry analysis, removing BSA from the KRB buffer is a key step to limit BSA interferences and salt contaminations. Therefore, after one hour of starvation, the cells were washed five times with BSA-free KRB buffer supplemented with 2.8 mM glucose and incubated for 1 h at 37 °C. Then, half of the supernatant was collected using a Bravo liquid handler, and the cell plates were complemented with KRB buffer containing glucose at 20 mM and incubated for 1 h at 37 °C. The high glucose fractions were then collected and cells were lysed to measure insulin content. To confirm the efficiency of the automated protocol, GSIS samples were subjected to an ELISA assay to measure insulin secretion in response to low and high glucose concentrations. Randomly selected samples from two separate 384-well plates were measured. Our data show that Min6 cells display a secretion rate which increases significantly after glucose stimulation, which represents 3 to $5\%$ of the total insulin content (Figure 1A) or a 6- to 8-fold increase of insulin secretion when Min6 cells were stimulated from 2.8 mM to 20 mM of glucose (Figure 1B). These data demonstrate that our GSIS protocol is functional in 384-well plates and that GSIS of Min6 cells can be fully automated in 384-well plates following our protocol. ## 3.2. Automated Analysis of GSIS in 384-Well Plates through MALDI-TOF Mass Spectrometry Having established the miniaturized GSIS protocol, we next wanted to quantitatively detect insulin from GSIS assays through an automated approach that should be at least as sensitive, reliable and quantitative as ELISA and that can be automated. To this end, we selected MALDI-TOF mass spectrometry. First, we set up a spiked-based strategy that allows a precise quantification and normalization of the insulin protein present in our samples. Since the murine insulin has a molecular weight of 5803 Da, we selected bovine insulin as an internal standard, which has a molecular weight of 5733 Da and is assumed to present an ionizability very close to the one of murine insulin as its amino acid composition is very similar. To quantitatively measure insulin from GSIS supernatants, samples were spiked with known concentrations of bovine insulin. Following MALDI-TOF mass spectrometry, a final mass spectrum was obtained, corresponding to the sum of 5000 laser shots (Figure 2A). The relative intensity emitted by the detected murine insulin was normalized to the relative intensity of our spiked bovine insulin internal standard. Importantly, we observed that mixing our internal standard with our samples did not affect the signal strength of the detected murine insulin (Figure 2B). Then, to demonstrate the reliability of MALDI-TOF mass spectrometry to detect insulin, samples from GSIS experiments performed in 384-well plates were analyzed through an ELISA assay and data were compared to MALDI-TOF mass spectrometry results. When we measured automated GSIS through ELISA or MALDI-TOF mass spectrometry, we could not observe differences in GSIS results between these two approaches. The analysis of two independent plates demonstrated that Min6 cells secreted between six to eight times more insulin at 20 mM glucose compared to 2.8 mM glucose, independent of the method of detection, i.e., mass spectrometry or ELISA (Figure 2C). These results demonstrate a strong reliability between both detection methods to measure insulin protein and suggest that MALDI-TOF mass spectrometry is as sensitive, as reliable and as quantitative as ELISA for GSIS measurements. ## 3.3. Automated siRNA Transfection Combined with GSIS We next wanted to apply the automated GSIS protocol to the discovery of new potential genes involved in insulin secretion. Following our experimental strategy described above, a fully automated siRNA reverse transfection protocol was implemented with the aim to detect variations of insulin secretion upon knocking-down specific genes. We first determined the optimal transfection conditions using a 6-FAM fluorescent-labeled control siRNA. This step is crucial to ensure an efficient transfection rate to potentially obtain a significant knock-down of the expression of target genes. Using siGLO as a fluorescent control to assess transfection efficiency, we could determine the optimal concentration for transfection reagents, siRNA and time of incubation (Supplementary Figure S3). Once the automated miniaturized transfection and GSIS protocols were validated, we tested two positive control siRNA: Pdx1 and Kcnj11, two genes previously shown to negatively modulate insulin secretion. Knock-down of Pdx1 gene in Min6 cells was validated both at the transcriptomic and protein levels (Supplementary Figure S4). Upon knock-down, GSIS samples were then analyzed by MALDI spectrometry and we detected lower relative intensities for murine insulin in cells treated with siRNA targeting Pdx1 and Kcnj11 (Figure 3A,B). Indeed, intensity ratios under 2.8 mM and 20 mM glucose conditions showed a ten-fold increase in insulin secretion for siControl-treated Min6 cells, whereas a six-fold increase in siPdx1-treated cells and a four-fold increase in siKcnj11-treated cells were observed (Figure 3C). These results were further independently confirmed through an ELISA assay where siRNAs targeting Kcnj11 and Pdx1 induced a significant decrease in insulin secretion after a stimulation with 20 mM glucose (Figure 3D). Again, mass spectrometry analysis further confirmed the results obtained through ELISA approaches (Figure 3E). Altogether, these data demonstrate that an automated siRNA reverse transfection protocol is efficient to evaluate potential functional effects of knocking-down genes in Min6 cells and suggest that this approach provides a robust automated platform to evaluate GSIS in HTS approaches. ## 3.4. Automated Chemical Treatment Combined with GSIS As a proof of concept, we next wanted to implement HTS strategies using our Min6 automated GSIS protocol coupled to mass spectrometry analysis. To demonstrate the feasibility of our approach, the use of pharmacological activators (repaglinide) and inhibitors (diazoxide) of insulin secretion has been undertaken to observe insulin secretion variations in Min6 cells. After treatment of the cells with these compounds, MALDI-TOF mass spectrometry analysis was employed to measure insulin secretion in response to these drugs. Our data revealed that these compounds were effective in modulating glucose-stimulated insulin secretion (Figure 4A), as demonstrated using automated MALDI TOF mass spectrometry or ELISA assays. Indeed, upon stimulation with 2.8 mM glucose conditions, Min6 cells treated with 100 nM of repaglinide had a higher basal secretion rate than untreated cells or cells incubated with 100 μM of diazoxide (Figure 4B). Upon 20 mM glucose condition treatment, a two-fold increase and a two-fold decrease in insulin secretion in the presence of repaglinide and diazoxide, respectively, was observed (Figure 4C,D), which were again confirmed through an ELISA assays (Figure 4E). In addition, the intracellular insulin content of Min6 cells exposed to repaglinide was decreased by $30\%$ when compared to other conditions (Supplementary Figure S5A–C), yet the total amount of insulin was similar in Min6 for all conditions of treatments (Supplementary Figure S5D). As expected, cells stimulated with repaglinide showed decreased intracellular insulin content when compared to control, untreated cells. Indeed, treating Min6 cells with repaglinide induced a $70\%$ decrease in the intracellular insulin content, and, conversely, diazoxide increased by $30\%$ insulin content compared to control, untreated cells (Figure 5A). Thanks to our high throughput MALDI-TOF approach, we analyzed 160 vehicle-treated samples, 62 repaglinide-treated samples and 62 diazoxide-treated samples, and observed that their analysis followed a normal distribution (Figure 5B and Figure S6). In addition, we could repeat this experiment with other bioactive molecules known to induce insulin secretion, such as forskolin or IBMX, and confirmed our results obtained with repaglinide (Supplementary Figure S7). After GSIS, the intensity levels corresponding to insulin observed for the cells treated with different secretagogues were markedly lower than the control, untreated cells, suggesting that these compounds may provoke intracellular insulin leakage (Supplementary Figure S7). ## 3.5. Automated Library Screening to Measure Insulin Intracellular Content after GSIS Since our automated cell culture protocol, MALDI-TOF mediated insulin detection and quantification and GSIS pipeline is adapted for high-throughput applications, we performed a pilot screening experiment in Min6 cells using a collection of 1600 compounds from ASINEX library and DMSO, repaglinide and diazoxide as controls. Here, we focused on measuring insulin content only. As observed previously, repaglinide induced a significantly lower intracellular insulin content compared to the negative controls (Figure 6). This screening was carried out in duplicate in order to be able to calculate a z-score for all the compounds tested. The DMSO control cells presented a z-score comprised in a window between z = −1 and $z = 1.$ The cells treated with repaglinide had a z-score between z = −1.5 and z = −3. We assumed that cells treated with compounds with the same order of magnitude of repaglinide z-score may be considered as molecules that stimulate insulin secretion. Conversely, compounds leading to a z-score close to controls or diazoxide may be considered as, at least, non-effective or inhibitors of insulin secretion. Among the 1600 compounds tested, we observed that some molecules decreased insulin content after glucose stimulation (Figure 6), suggesting that they may stimulate insulin secretion. Amongst those molecules, we observed different compounds with a z-score close to those of repaglinide, the positive control. Altogether, our results demonstrate that our approach using MALDI-TOF mass spectrometry can be automated and efficiently used for HTS of libraries of siRNA or chemical compounds. ## 4. Discussion Here, we describe a new automated methodology to accurately measure insulin secretion in vitro and identify potential modulators of insulin secretion in pancreatic beta cells. This new process, based on automated GSIS and MALDI-TOF mass spectrometry, was developed in a 384-well plate format to enable high-throughput analyses and identify new chemical substances or genes that potentially modulate insulin secretion or cell content. We demonstrated the efficiency of our assay by performing a pilot experiment using 1600 chemical compounds and selected siRNAs, in the mouse Min6 beta cell line. Indeed, we have developed and implemented this technological process to test the efficiency of screening different chemical compounds, but also siRNA targeting bona fide β-cell identity genes. Unlike other techniques such as ELISA assay, our protocol is based on direct insulin detection and not on antigen recognition. In addition, we demonstrate that the MALDI-TOF mass spectrometry strategy is as reliable as ELISA detection, while being much less expensive than existing insulin detection techniques. High-throughput screening for insulin secretion is an approach with high potential for valorization. In the case of the insulin-secreting pancreatic β-cell, these approaches are limited by several technological constraints inherent to the methods for detection and quantification of this hormone (i.e., ELISA). To overcome these experimental limitations, several laboratories have reported the use of modified cell-based tools to indirectly measure the level of secretory activity of mouse-derived Min6 cells by fluorescence microscopy or measurement of modified peptide-C luciferase [6,7,8,9]. Although simple to manipulate, these artificial, genetically engineered tools only very partially interrogate β-cell functionality as it assesses neither endogenous insulin production nor its secretion under glucose stimulation conditions. It is with the objective of addressing these two analytical criteria that we have proposed to deploy our complete procedure of automation of the GSIS protocol followed by MALDI-TOF mass spectrometry to envision, in the future, a siRNA library (approx. 20,000 genes) and chemical library (>100,000 compounds) screening strategy in the Min6 pancreatic β-cell model. A similar study has already been undertaken on the INS-1 $\frac{832}{13}$ pancreatic beta cell model, with the screening of 1200 compounds, but unlike our study, the compounds used have already known targets [11,12]. The scope of the technology is likely very broad, as MS detection is extremely versatile. The most direct applications are related to the high-throughput screening for other molecules that modulate the secretion of other peptide hormones such as Glucagon-like peptide 1 (GLP-1). In the context of the development of personalized medicine, our analysis process could be deployed to define the most appropriate treatment to restore the hormonal secretion of a patient suffering from T2D. This concept of searching for alternative therapeutics could naturally be extended to other disciplinary fields such as oncology or the study of neurodegenerative diseases. Nevertheless, the use of our process in a precise context remains subject to the prior knowledge of the protein(s) of interest. Although encouraging, our study has several limitations. The use of the murine β-cell model Min6 is convenient, but translation to human cells remains to be undertaken. It may be relevant to confirm some of the target and/or small molecules identified in Min6 cells in the EndoC-BH1 human model, which has been shown to be useful for the identification of modulators of human beta-cell insulin secretion [13]. In addition, siRNA mediated knock-down may not be efficient enough to identify genes potentially involved in β-cell function. The use of CrispR/Cas9, as performed in human islets [14] or in EndoC-BH1 [15], may help to better identify genes that are directly involved in insulin secretion, strengthening the translation to human. Interestingly, a new generation of human beta cell line, EndoC-HB5 cells, is compatible with HTS experiments with a good reactivity to glucose stimulation. ## References 1. Lin X., Xu Y., Pan X., Xu J., Ding Y., Sun X., Song X., Ren Y., Shan P.-F.. Global, Regional, and National Burden and Trend of Diabetes in 195 Countries and Territories: An Analysis from 1990 to 2025. Sci. Rep. (2020) 10 14790. DOI: 10.1038/s41598-020-71908-9 2. Artasensi A., Pedretti A., Vistoli G., Fumagalli L.. Type 2 Diabetes Mellitus: A Review of Multi-Target Drugs. Molecules (2020) 25. DOI: 10.3390/molecules25081987 3. Nisal K., Kela R., Khunti K., Davies M.J.. Comparison of Efficacy between Incretin-Based Therapies for Type 2 Diabetes Mellitus. BMC Med. (2012) 10. DOI: 10.1186/1741-7015-10-152 4. Aguayo-Mazzucato C.. Functional Changes in Beta Cells during Ageing and Senescence. Diabetologia (2020) 63 2022-2029. DOI: 10.1007/s00125-020-05185-6 5. Heyduk E., Moxley M.M., Salvatori A., Corbett J.A., Heyduk T.. Homogeneous Insulin and C-Peptide Sensors for Rapid Assessment of Insulin and C-Peptide Secretion by the Islets. Diabetes (2010) 59 2360-2365. DOI: 10.2337/db10-0088 6. Kalwat M.A., Wichaidit C., Nava Garcia A.Y., McCoy M.K., McGlynn K., Hwang I.H., MacMillan J.B., Posner B.A., Cobb M.H.. Insulin Promoter-Driven Gaussia Luciferase-Based Insulin Secretion Biosensor Assay for Discovery of β-Cell Glucose-Sensing Pathways. ACS Sens. (2016) 1 1208-1212. DOI: 10.1021/acssensors.6b00433 7. Burns S.M., Vetere A., Walpita D., Dančík V., Khodier C., Perez J., Clemons P.A., Wagner B.K., Altshuler D.. High-Throughput Luminescent Reporter of Insulin Secretion for Discovering Regulators of Pancreatic Beta-Cell Function. Cell Metab. (2015) 21 126-137. DOI: 10.1016/j.cmet.2014.12.010 8. Pappalardo Z., Chopra D.G., Hennings T.G., Richards H., Choe J., Yang K., Baeyens L., Ang K., Chen S., Arkin M.. A Whole-Genome RNA Interference Screen Reveals a Role for Spry2 in Insulin Transcription and the Unfolded Protein Response. Diabetes (2017) 66 1703-1712. DOI: 10.2337/db16-0962 9. Kalwat M.A., Niedergang F., Vitale N., Gasman S.. High-Throughput Screening for Insulin Secretion Modulators. Exocytosis and Endocytosis: Methods and Protocols (2021) 131-138 10. Bourouh C., Courty E., Rolland L., Pasquetti G., Gromada X., Rabhi N., Carney C., Moreno M., Boutry R., Caron E.. The Transcription Factor E2F1 Controls the GLP-1 Receptor Pathway in Pancreatic β Cells. Cell Rep. (2022) 40 111170. DOI: 10.1016/j.celrep.2022.111170 11. Wu W., Shang J., Feng Y., Thompson C.M., Horwitz S., Thompson J.R., Macintyre E.D., Thornberry N.A., Chapman K., Zhou Y.-P.. Identification of Glucose-Dependent Insulin Secretion Targets in Pancreatic β Cells by Combining Defined-Mechanism Compound Library Screening and SiRNA Gene Silencing. J. Biomol. Screen (2008) 13 128-134. DOI: 10.1177/1087057107313763 12. Aamodt K.I., Aramandla R., Brown J.J., Fiaschi-Taesch N., Wang P., Stewart A.F., Brissova M., Powers A.C.. Development of a Reliable Automated Screening System to Identify Small Molecules and Biologics That Promote Human β-Cell Regeneration. Am. J. Physiol. Endocrinol. Metab. (2016) 311 E859-E868. DOI: 10.1152/ajpendo.00515.2015 13. Szczerbinska I., Tessitore A., Hansson L.K., Agrawal A., Lopez A.R., Helenius M., Malinowski A.R., Gilboa B., Ruby M.A., Gupta R.. Large-Scale Functional Genomics Screen to Identify Modulators of Human β-Cell Insulin Secretion. Biomedicines (2022) 10. DOI: 10.3390/biomedicines10010103 14. Bevacqua R.J., Dai X., Lam J.Y., Gu X., Friedlander M.S.H., Tellez K., Miguel-Escalada I., Bonàs-Guarch S., Atla G., Zhao W.. CRISPR-Based Genome Editing in Primary Human Pancreatic Islet Cells. Nat. Commun. (2021) 12 2397. DOI: 10.1038/s41467-021-22651-w 15. Grotz A.K., Abaitua F., Navarro-Guerrero E., Hastoy B., Ebner D., Gloyn A.L.. A CRISPR/Cas9 Genome Editing Pipeline in the EndoC-ΒH1 Cell Line to Study Genes Implicated in Beta Cell Function. Wellcome Open Res. (2020) 4 150. DOI: 10.12688/wellcomeopenres.15447.2
--- title: High Levels of Leptin and Adipsin Are Associated with Clinical Activity in Early Rheumatoid Arthritis Patients with Overweight and Periodontal Infection authors: - Consuelo Romero-Sánchez - Juliette De Avila - Alejandro Ramos-Casallas - Lorena Chila-Moreno - Nathaly Andrea Delgadillo - Philippe Chalem-Choueka - César Pacheco-Tena - Juan Manuel Bello-Gualtero - Wilson Bautista-Molano journal: Diagnostics year: 2023 pmcid: PMC10047025 doi: 10.3390/diagnostics13061126 license: CC BY 4.0 --- # High Levels of Leptin and Adipsin Are Associated with Clinical Activity in Early Rheumatoid Arthritis Patients with Overweight and Periodontal Infection ## Abstract Adipokines are associated with the pathogenesis of rheumatoid arthritis (RA) and are potential biomarkers of disease activity, periodontitis, and obesity. The aim of this was to establish the association between adipokine profile, RA disease activity, body mass index, and periodontal infection. This study evaluated 51 patients with early-RA and 51 controls including serum rheumatological markers, adipokine levels, detection of Porphyromonas gingivalis and serum anti-Porphyromonas gingivalis antibodies, clinical and periodontal measurements. Statistical analyses were run with SPSS® V26, with a logistic regression model to confirm associations. The results show high levels of leptin were more frequent in patients ($$p \leq 0.001$$) who simultaneously showed a higher frequency of Porphyromonas gingivalis ($$p \leq 0.004$$). Patients with concomitant presence of Porphyromonas gingivalis, high clinical activity score, and overweight were correlated with high levels of leptin (OR, 7.20; $95\%$ CI, 2.68–19.33; $$p \leq 0.0001$$) and adipsin (OR, 2.69; $95\%$ CI, 1.00–7.28; $$p \leq 0.005$$). The conclusion is that high levels of leptin and adipsin are associated with greater clinical activity in early-RA patients with overweight and periodontal infection, whereby overweight and Porphyromonas gingivalis may enhance RA activity. This may represent a pathological mechanism between these conditions, where adipokines seem to have a key role. ## 1. Introduction Rheumatoid arthritis (RA) is a chronic, inflammatory joint disease of autoimmune nature which can lead to accumulating joint damage and irreversible disability. The disease is complex and involves environmental factors that trigger it in genetically susceptible individuals [1]. Periodontitis mediated by Porphyoromonas gingivalis (P. gingivalis) infection precedes RA and is a likely factor in the onset and maintenance of the autoimmune response [2]. Some authors recently estimated a $1.49\%$ prevalence in Colombia, through the Community Oriented Program for Control of Rheumatic Diseases strategy [3]. As previously described, patients with early stages of RA (eRA) present a significant incidence of periodontal inflammation (possibly due to periodontal microbiota changes influenced by antirheumatic drugs (DMARDs)) [4] and obesity. The latter is a risk factor related to higher disease activity, functional ability, and health-related quality of life in patients [5]. Based on the evidence regarding the relationship between obesity and periodontitis [6], these two conditions could be linked due to the production of adipokines [7]. Adipokines are a variety of chemical agents (including cytokines and chemokines), which are secreted by adipocytes and non-adipocyte cells (such as fibroblasts, vascular cells, etc.), involved in metabolic and immunity processes [8]. Obesity is a chronic subinflammatory state associated with an altered adipokine profile, with high levels of leptin and reduced adiponectin expression [9]. A similar condition is observed in periodontitis, where this dysregulation correlates with periodontal damage and BMI [10,11]. The proinflammatory profile observed in both conditions could be a related mechanism link. Several studies may support this hypothesis, demonstrating that periodontal ligament cells constitutively secrete adipokines, and external factors can modulate their secretion [12,13]. The association between adipokines and RA development previously described [14] indicates a correlation with radiographic joint damage progression. Leptin seems to be involved in RA, showing positive associations with acute phase markers, disease activity and BMI [15]. In the case of adipsin, serum levels are correlated with BMI, interleukin-6 (IL-6), and erythrocyte sedimentation rate (ESR) in eRA patients [16]. Regarding vaspin, serum levels are associated with the development of clinical manifestations in individuals with serum arthritis biomarkers [17]. The evidence of the potential impact of adipokines as a factor associated with the risk to individuals with developing eRA is limited [15], as most of the studies include patients with established RA [5,14]. Although the mechanisms underlying these associations are not well understood, together, high levels of leptin, the presence of P. gingivalis, and overweight may be relevant conditions associated with the development of RA [18]. However, the possibility exists that the adipokine profile is not related to clinical and periodontal conditions in those patients. Hence, this study aimed to establish the association between adipokine profile, rheumatic activity, BMI, and periodontal infection in eRA patients compared to disease control group. ## 2. Materials and Methods This study simultaneously recruited controls and patients with a recent diagnosis of RA, according to the ACR/EULAR2010 criteria [19] with fewer than 2 years of disease evolution, in the Department of Rheumatology and Immunology at the Hospital Militar Central, and Fundación Instituto de Reumatología Fernando Chalem, in Bogotá-Colombia from June 2015 to February 2017. The early-RA (eRA) group was composed of patients between 18 and 65 years old, under only conventional treatment (nonsteroidal anti-inflammatory drugs—NSAIDs, disease-modifying antirheumatic drugs—DMARDs, and/or steroids). The exclusion criteria were ongoing infectious diseases, diagnosis of another autoimmune/autoinflammatory disease, current malignancies, diabetes mellitus, actual orthodontic appliances, antibiotic treatment in the last 3 months, periodontal therapy in the last 6 months, pregnancy, or breastfeeding. The control group was composed of individuals between 18 and 65 years old, with working or environmental conditions similar to the eRA group. The exclusion criteria for this group were ongoing infectious diseases, diagnosis of another autoimmune/autoinflammatory disease, having familiar history of autoimmune or autoinflammatory diseases, current malignancies, diabetes mellitus, actual orthodontic appliances, antibiotic treatment in the last 3 months, periodontal therapy in the last 6 months, pregnancy, or breastfeeding. Arterial hypertension and overweight were not exclusion variables. This study was conducted with a nonprobability sampling. Based on the difficulties of recruiting eRA patients in the early stages of the disease and the strict inclusion criteria, the selection was made by convenience. All individuals signed the informed consent form approved by the Hospital Militar Central Institutional Ethics Committee (codes HMC 2016-041 and 2016-099). ## 2.1. Clinical Examination Rheumatologists measured the level of RA disease activity using the disease activity score 28 (DAS28), the simplified disease activity index (SDAI), the routine assessment of patient index data 3 (RAPID3), and the multidimensional health assessment questionnaire (MDHAQ). All patients were evaluated by calibrated periodontists, who performed periodontal examinations. A full-mouth examination was performed including selected sites on each permanent tooth, excluding third molars. All patients were evaluated for periodontitis based on the 2017 criteria of the World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions; the stages of periodontitis were determined based on CAL and PD values in interproximal sites [20]. Additionally, periodontal indices, including PD, (inter-examiner intra-class correlation coefficient (IE-ICC), 0.94–0.96) CAL (IE-ICC, 0.92–0.96), bleeding on probing (IE-ICC, 0.88–0.90), plaque index (IE-ICC, 0.94–0.98), and gingival index (IE-ICC, 0.85–0.90), were evaluated in full-mouth examinations including the selected sites. ## 2.2. Adipokine Levels and Serum Markers Measurements Leptin and vaspin quantifications were calculated using an indirect enzyme-linked immunosorbent assay (ELISA) (Diasource, KAP2281, and MyBioSource, MBS267502 kits, respectively). Quantification of adiponectin, resistin, and adipsin were performed using Luminex xMAP® technology (MILLIPLEX® MAP, HADK1MAG61K03, Merck KGaA, Darmstadt, Germany). The quantification of IL-6 and CRP were performed using chemiluminescence technology (Immulite 1000, Siemens® Cat. LKP1 No. 10381411, 10286287, Gwynedd, UK) with detectable values of >3.4 pg/mL for IL-6 and high values of >3 mg/L for CRP. The quantitative measurement of IgG/IgA ACPAs in the serum was performed using a commercially available ELISA kit (Quanta lite® CCP 3.1 IgG/IgA, INNOVA Diagnostics, San Diego, CA, USA) (positive ≥ 20 IU/mL). The measurement of rheumatoid factor (RF) was calculated using a kinetic turbidimetry technique (Spinreact®, 110705, Girona, Spain) (positive ≥ 20 IU/mL), and the ESR using quantitative capillary photometry technology (Alifax Spa®, Padova, Italy) (normal value < 20 mm/h). All the tests were run following the manufacturers’ instructions. ## 2.3. Detection of DNA of P. gingivalis, IgG1 and IgG2 Anti-P. gingivalis Antibodies DNA of P. gingivalis was detected through a quantitative PCR (qPCR) technique. To determine IgG1 and IgG2 antibodies against P. gingivalis, indirect ELISA in-house assays were performed. These tests were run as described in a previous study [21]. ## 2.4. Statistical Analysis To explore the associations between adipokine levels, rheumatologic activity, BMI, and P. gingivalis, an analysis was performed using χ2 or Fisher’s exact tests, comparisons were evaluated by Mann–Whitney U tests, and a logistic regression model was made to confirm associations including those variables which showed associations by bivariate analysis. All analyses were conducted using IBM-SPSS software V26 for Windows with a $95\%$ confidence. The effect of sample size on the results was calculated using the matched study power module from Epidat V 4.2. ## 3. Results In this study, 62 patients with eRA and 62 controls were included. A total of 51 patients and controls matched by age and sex were evaluated. ## 3.1. Sociodemographic Characteristics The predominant gender was female ($80.39\%$), where BMI score > 25 was higher in patients ($50.98\%$), and smoking conditions were similar between patients and controls (Table 1). ## 3.2. Periodontal Status The comparison between patients and controls showed that both patients and controls have similar frequency of periodontitis and its severity. However, presence of P. gingivalis was more frequent in patients ($78.43\%$), and IgG titers against P. gingivalis were higher in controls (Table 2). ## 3.3. Rheumatologic Serum Biomarkers and Disease Activity Scores Regarding serum biomarkers, $62.74\%$ of eRA patients had levels of RF > 20 IU, $45.09\%$ had levels of ACPA > 20 IU, $56.86\%$ had CRP values higher than 3 mg/L, and $29.41\%$ had prolonged ESR values (>20 mm/h). The disease activity was measured using the DAS28 score: $62.74\%$ and $58.82\%$ presented a high disease activity score in DAS28-ESR and DAS28-CRP, respectively (>3.2); $58.82\%$ showed elevated disease activity according to SDAI score (>11); $47.05\%$ exhibited high disease severity (>12) according to RAPID-3 score. ## 3.4. Adipokine Profile To analyze adipokine levels in serum and perform measurements of a continuous variable, cut-off points were necessary to create three equal groups with a canonical stratification based on a percentile 33 (Table 3). The frequency of adipokine profile in patients was higher levels of leptin, vaspin, and adipsin, than in controls. Conversely, high levels of adiponectin, resistin, and IL-6 were similar between these groups. A comparison is shown in Figure 1. ## 3.5. Multivariate Analysis In the multivariate analysis, only leptin and adipsin remained statistically significant. P. gingivalis and high levels of leptin and adipsin were simultaneously present in $53.83\%$ and $37.73\%$ of the patients, respectively, with a statistical power of $98.9\%$ for leptin and $94.4\%$ to adipsin. $35.84\%$ of patients were ACPA-positive and had P. gingivalis, of which $57.89\%$ had high levels of adipsin. Of these patients, $63.64\%$ manifested high disease activity (DAS28-ESR > 3.2). Among ACPA-positive patients with P. gingivalis, $68.42\%$ and $57.89\%$ had high levels of leptin and adipsin, respectively. Of these patients $62.53\%$ for leptin and $63.64\%$ for adipsin manifested a high disease activity score (DAS28-ESR > 3.2). In general, these patients were associated with high levels of leptin (OR, 8.22; $95\%$ CI, 2.75–24.50; $$p \leq 0.001$$), high levels of adipsin (OR, 3.06; $95\%$ CI, 1.05–8.97; $$p \leq 0.041$$), and DAS28-ESR > 3.2 (OR, 2.59; $95\%$ CI, 1.46–4.58; $$p \leq 0.001$$). A total of $22.64\%$ of patients simultaneously had P. gingivalis, DAS28-ESR > 3.2, CRP > 3.0 mg/L, and BMI > 25. These patients demonstrated an association with high levels of leptin and adipsin (OR, 7.20; $95\%$ CI, 2.68–19.33; $$p \leq 0.0001$$ and OR, 2.69; $95\%$ CI, 1.00–7.28; $$p \leq 0.005$$, respectively) (Figure 2). The statistical power for this composite variable was $99.99\%$ both for leptin and adipsin. Finally, there was no association of adiponectin, resistin, vaspin, or IL-6 with periodontal markers, rheumatology disease activity, and treatment in eRA patients. ## 4. Discussion RA is a complex systemic autoimmune disease characterized by joint destruction. Environmental and lifestyle factors such as P. gingivalis-mediated periodontitis and obesity are associated with a higher risk of developing RA [2,5]. A significant BMI score > 25 was more frequently observed in patients. However, there is a high prevalence of increased BMI scores in healthy people. Kasper et al. reported the increasing rate of obesity in Colombia, correlated with female gender, age, socioeconomic status, and urban residence [22]. With obesity as a high-risk condition for developing comorbid disorders (including RA), these results support that obesity is a public health issue, which will affect the overall quality of life of the population and healthcare costs. Regarding occupational status, RA patients more frequently reported their occupational status as homemakers, independent workers, or retired. This may be the result of the disability associated with the chronic destructive evolution of RA. According to the follow-up study of Nikiphorou et al., work loss related to RA occurred especially in the first 5 years of RA, but improved over time. The gradual changes in therapies may be one explanation for the differences observed [23]. However, work disability results from a complex interaction between a clinical disease, sociodemographic variables, macroeconomic conditions, and personal factors [24]. Therefore, this condition may imply an integral approach, involving not only the clinical treatment. As periodontal conditions pose a risk for RA, there was a similar frequency of periodontitis between both groups, with moderate periodontitis the most prevalent condition, with no significant differences. Similar findings have been reported in previous studies, despite the higher detection rate of P. gingivalis in patients than in controls [21]. This may be due to the sociocultural and hygienic conditions related to the Colombian population, which could establish the same periodontal conditions in both patients and controls [25]. However, the higher P. gingivalis detection in eRA patients has a relevant impact despite no differences in the periodontal status. P. gingivalis is a microorganism that can express peptidyl arginine deiminase (PAD), which represents an important pathogenic factor of RA, likely mediating the autoimmune response through citrullination of proteins [2]. Conversely, higher titers in IgG$\frac{1}{2}$ anti-P. gingivalis antibodies were more frequent in controls than in patients. Johansson et al. detected higher anti-P. gingivalis antibody titers in RA patients than in controls, a finding suggestive of the role of P. gingivalis in the onset of RA [26]. As an extended periodontal pathogen in the general Colombian population [27], we can hypothesize that healthy individuals have an active immunity against this microorganism, thus generating a possible protection mechanism, in contrast to eRA patients because of their systemic compromise and the associated risk factors. Adipose tissue, through the production of adipokines, is emerging as one of the major drivers of systemic and local inflammation in rheumatic diseases [28]. The higher frequency of patients with BMI > 25 and their positive association with serum leptin levels observed in this study correlates with the findings reported by other authors, where obesity has been associated with higher disease activity and worse quality of life among RA patients [5]. Regarding adipsin levels, the association between BMI > 25 and eRA patients has been reported previously [16]; however, the role of adipsin in the context of obesity and RA is unclear. In periodontitis, leptin levels increase with a reduction in adiponectin levels, similar to what occurs in obesity [6]. This related mechanism may enhance periodontal disease. Zimmerman et al. found higher leptin levels in obese patients with chronic periodontitis compared to patients with chronic periodontitis and normal weight. The levels of adiponectin were similar in both groups but lower when compared to the control group [11]. Therefore, periodontitis and obesity might be related and can influence each other through adipokine secretion in eRA [7]. The periodontopathic bacteria P. gingivalis seems to be involved with adipokines. The presence of P. gingivalis was associated with eRA patients with higher levels of leptin and adipsin. The presence of P. gingivalis can be a continuous stimulus in the oral cavity, promoting the activation of TLR2 and the synthesis of pro-inflammatory cytokines such as leptin, with a reduction in adiponectin levels [29]. As the immune response is established in the organism, leptin can upregulate TLR2 expression in monocytes [30], where these cells take part in the host defense. The levels of adipsin should also rise, being also produced by monocytes in response to infection, promoting the activation of complement, and favoring the adipsin-dependent pathway to recruit more neutrophils [31,32]. Previous analyses have shown a relationship between the high levels of adiponectin and the absence of P. gingivalis [33]. However, this relationship was not found in the current study. Another additional factor that may mediate the periodontal inflammation and systemic condition in RA patients is oxidative stress. Oxidative stress is an alteration of the balance between the levels of oxidizing agents and those of antioxidants. During normal cellular metabolic processes, free radicals and reactive metabolites are continuously generated. When the rate of oxidant production exceeds the capacity of antioxidant systems to eliminate oxidizing products, oxidative stress is installed, which subsequently leads to cell and tissue damage [34]. It has been noticed that reactive species and antioxidants influence the immune system. Oxidative stress causes disruption in cell signaling, impairs arachidonic acid metabolism, and enhances airway and systemic inflammation [35]. Oxidative stress forms the basis of chronic inflammation, among other diseases [36]. Several oxidative stress biomarkers have been explored in the literature as potentially useful parameters in monitoring the progression of both periodontitis and RA. Biomarkers of oxidative stress (derived from protein damage, lipids, uric acid, and DNA oxidation) were consistently and significantly higher in value in patients with RA compared to systemically healthy individuals, this increase being observed in serum, plasma, urine, synovial fluid, and whole blood [37]. Additionally, Sezer et al. showed that patients with RA and chronic periodontitis presented a high oxidative stress index and prolidase levels [38]. Furthermore, non-surgical periodontal therapy showed an improvement in periodontal conditions and oxidative stress markers in RA patients [39]. While the focus of the current study was to establish the association of adipokines with clinical and periodontal conditions in patients with eRA, it would be interesting to investigate the relationship between adipokines and oxidative stress biomarkers in these patients. Several rheumatological variables were associated with adipokines. Of the ACPA-positive patients with P. gingivalis, $57.89\%$ and $68.42\%$ had higher levels of adipsin and leptin, respectively. More than $60\%$ of them had higher disease activity (DAS28-ESR > 3.2). The higher levels of leptin were related to DAS28-ESR > 3.2 score. Aside from clinical disease activity, inflammatory markers such as CRP seemed to be modulated by adipokines, where higher levels of leptin were related to CRP > 3 mg/L. Galiutina et al. reported the imbalance of leptin–adiponectin levels. They found that leptin levels in RA patients were 3.2 times higher and that adiponectin levels were 1.7 times lower than those in healthy individuals. In clinical activity, such imbalance was related to the increase in inflammatory markers, such as ESR and CRP. The DAS28 score was 1.6 times higher in individuals with high leptin levels. Thus, leptin levels are increased and are significantly associated not only with disease activity but also with higher levels of CRP [40]. In the case of adipsin, high levels of adipsin were associated with clinical disease (DAS28-ESR > 3.2 score). Additionally, high serum adipsin levels were positively correlated with CRP > 3 mg/L. Although there are no inflammatory properties described, some studies have shown the importance of adipsin in the development of inflammatory arthritis, promoting the migration of neutrophils into joints [32]. Additionally, higher levels of adipsin have been positively correlated with ESR and IL-6 in eRA [16]. Therefore, adipsin seems to be involved in pathological and inflammatory processes including RA, although its role is yet to be elucidated. Serum levels of vaspin were higher in eRA patients than controls, but this was not related to periodontal and/or rheumatological variables. Maijer et al. described vaspin as a molecule associated with the development of arthritis in ACPA-positive individuals [17]. This may suggest that vaspin could be involved in the development of the initial clinical manifestations. The serum adipokines leptin, IL-6, and resistin, along with CRP, have been added to a multi-biomarker test as a novel assessment for RA disease activity (MBDA). This test includes the measurement of 12 biomarkers, in which the test score was shown to be positively associated with disease activity and disease relapses in RA patients [41]. However, the “gold standard” assessment for disease activity of RA and the first choice of this study and the cited authors is the use of the DAS28 score instead of MBDA, perhaps due to the technical/economic difficulties of measuring the 12 biomarkers to carry out the MBDA test. A previous study found a similar relationship between serum adipokines with rheumatological variables of clinical activity and BMI [15]. All of these findings suggest that adipokines play a role in RA pathogenesis and that they may serve as a connection between RA and obesity. However, in this study, we also showed an association between higher leptin and adipsin levels in patients with eRA and the presence of P. gingivalis, high index of joint activity, high inflammation markers, and overweight (DAS28-ESR > 3.2, CRP > 3.0 mg/L and BMI > 25). Therefore, adipokines could be the link between obesity, P. gingivalis–related periodontitis, and eRA. We found no evidence of research on this topic in the literature reviewed; additional investigations should be encouraged. While these are important findings, this study presented some limitations. The main limitation of this study is the relatively small number of patients, given the lack of accessibility of the Colombian healthcare system for medical specialist appointments, the difficulties of recruiting patients in the early stages of the disease, and the strict inclusion criteria. In addition, the cross-sectional design of this study does not allow the establishment of the causality of the associations found. Further studies are encouraged, such as prospective follow-up studies with eRA patients, which allow the establishment of the causality of the data related to the adipokine profile and its relationship with rheumatological and periodontal conditions. ## 5. Conclusions High levels of leptin and adipsin are associated with greater clinical activity in eRA patients with overweight and periodontal infection. These adipokines could be a pathological mechanism whereby overweight and P. gingivalis infection might worsen the clinical activity in eRA patients. ## 6. Strengths and Limitations To the best of the authors’ knowledge, there are few reports evaluating the adipokine profile and its association with periodontal infection and RA activity among eRA patients. In a previous report, the evaluation included a shorter adipokine profile and radiological compromise. The current study measured six different adipokines and evaluated their association with serum rheumatological markers, and clinical and periodontal conditions, which were found to be important associations. This study provides evidence that adipokines may play an essential pathophysiological role in the clinical activity of patients with eRA, overweight, and periodontal infection. However, the cross-sectional design of this study does not allow the establishment of the causality of the associations found. Furthermore, the main limitation of this study is the relatively small number of patients, given the lack of accessibility of the Colombian healthcare system for medical specialist appointments, the difficulties of recruiting patients in the early stages of the disease, and the strict inclusion criteria applied in this study. It is important to carry out additional studies (especially prospective follow-up studies) on this topic, to find causality associations. Early stages of RA represent an important therapeutic window to use integral approaches to modulate the progression of the disease and improve the quality of life of patients. A possible bias in this study is the high BMI score in the patients, and the difficulty of recruiting controls who could be matched by sex, age, and BMI score simultaneously. Given that the frequency of overweight is higher in patients with RA, which is supported by the literature, patients and controls were selected under similar environmental and working conditions, and paired by age and sex, not by BMI. ## References 1. Malmström V., Catrina A.I., Klareskog L.. The Immunopathogenesis of Seropositive Rheumatoid Arthritis: From Triggering to Targeting. Nat. Rev. Immunol. (2017) 17 60-75. DOI: 10.1038/nri.2016.124 2. Lundberg K., Wegner N., Yucel-Lindberg T., Venables P.J.. Periodontitis in RA—The Citrullinated Enolase Connection. Nat. Rev. Rheumatol. (2010) 6 727-730. DOI: 10.1038/nrrheum.2010.139 3. Londoño J., Peláez Ballestas I., Cuervo F., Angarita I., Giraldo R., Rueda J.C., Ballesteros J.G., Baquero R., Forero E., Cardiel M.. Prevalencia de La Enfermedad Reumática En Colombia, Según Estrategia COPCORD-Asociación Colombiana de Reumatología. Estudio de Prevalencia de Enfermedad Reumática En Población Colombiana Mayor de 18 Años. Rev. Colomb. Reumatol. (2018) 25 245-256. DOI: 10.1016/j.rcreu.2018.08.003 4. Bautista-Molano W., van der Heijde D., Landewé R., Lafaurie G.I., de Ávila J., Valle-Oñate R., Romero-Sanchez C.. Is There a Relationship between Spondyloarthritis and Periodontitis? A Case–Control Study. RMD Open (2017) 3 e000547. DOI: 10.1136/rmdopen-2017-000547 5. Nikiphorou E., Norton S., Young A., Dixey J., Walsh D., Helliwell H., Kiely P.. The Association of Obesity with Disease Activity, Functional Ability and Quality of Life in Early Rheumatoid Arthritis: Data from the Early Rheumatoid Arthritis Study/Early Rheumatoid Arthritis Network UK Prospective Cohorts. Rheumatology (2018) 57 1194-1202. DOI: 10.1093/rheumatology/key066 6. Suvan J., D’Aiuto F., Moles D.R., Petrie A., Donos N.. Association between Overweight/Obesity and Periodontitis in Adults. A Systematic Review. Obes. Rev. (2011) 12 e381-e404. DOI: 10.1111/j.1467-789X.2010.00808.x 7. Deschner J., Eick S., Damanaki A., Nokhbehsaim M.. The Role of Adipokines in Periodontal Infection and Healing. Mol. Oral Microbiol. (2014) 29 258-269. DOI: 10.1111/omi.12070 8. Dodson M.V., Du M., Wang S., Bergen W.G., Fernyhough-Culver M., Basu U., Poulos S.P., Hausman G.J.. Adipose Depots Differ in Cellularity, Adipokines Produced, Gene Expression, and Cell Systems. Adipocyte (2014) 3 236-241. DOI: 10.4161/adip.28321 9. Leal V.d.O., Mafra D.. Adipokines in Obesity. Clin. Chim. Acta (2013) 419 87-94. DOI: 10.1016/j.cca.2013.02.003 10. Shimada Y., Komatsu Y., Ikezawa-Suzuki I., Tai H., Sugita N., Yoshie H.. The Effect of Periodontal Treatment on Serum Leptin, Interleukin-6, and C-Reactive Protein. J. Periodontol. (2010) 81 1118-1123. DOI: 10.1902/jop.2010.090741 11. Zimmermann G.S., Bastos M.F., Dias Gonçalves T.E., Chambrone L., Duarte P.M.. Local and Circulating Levels of Adipocytokines in Obese and Normal Weight Individuals with Chronic Periodontitis. J. Periodontol. (2013) 84 624-633. DOI: 10.1902/jop.2012.120254 12. Nokhbehsaim M., Keser S., Nogueira A.V.B., Jäger A., Jepsen S., Cirelli J.A., Bourauel C., Eick S., Deschner J.. Leptin Effects on the Regenerative Capacity of Human Periodontal Cells. Int. J. Endocrinol. (2014) 2014 180304. DOI: 10.1155/2014/180304 13. Nokhbehsaim M., Keser S., Nogueira A.V.B., Cirelli J.A., Jepsen S., Jäger A., Eick S., Deschner J.. Beneficial Effects of Adiponectin on Periodontal Ligament Cells under Normal and Regenerative Conditions. J. Diabetes Res. (2014) 2014 796565. DOI: 10.1155/2014/796565 14. Hutcheson J.. Adipokines Influence the Inflammatory Balance in Autoimmunity. Cytokine (2015) 75 272-279. DOI: 10.1016/j.cyto.2015.04.004 15. Chaparro-Sanabria J.A., Bautista-Molano W., Bello-Gualtero J.M., Chila-Moreno L., Castillo D.M., Valle-Oñate R., Chalem P., Romero-Sánchez C.. Association of Adipokines with Rheumatic Disease Activity Indexes and Periodontal Disease in Patients with Early Rheumatoid Arthritis and Their First-Degree Relatives. Int. J. Rheum. Dis. (2019) 22 1990-2000. DOI: 10.1111/1756-185X.13724 16. Klein-Wieringa I.R., van der Linden M.P.M., Knevel R., Kwekkeboom J.C., van Beelen E., Huizinga T.W.J., van der Helm-van Mil A., Kloppenburg M., Toes R.E.M., Ioan-Facsinay A.. Baseline Serum Adipokine Levels Predict Radiographic Progression in Early Rheumatoid Arthritis. Arthritis Rheum. (2011) 63 2567-2574. DOI: 10.1002/art.30449 17. Maijer K.I., Neumann E., Müller-Ladner U., Drop D.A.C.A.D., Ramwadhdoebe T.H., Choi I.Y.K., Gerlag D.M., de Hair M.J.H., Tak P.P.. Serum Vaspin Levels Are Associated with the Development of Clinically Manifest Arthritis in Autoantibody-Positive Individuals. PLoS ONE (2015) 10. DOI: 10.1371/journal.pone.0144932 18. Romero-Sanchez C., Chaparro-Sanabria J.A., Bello-Gualtero J.M., Valle-Oñate R., Chila L., Castillo D.M., Lafaurie G., Chalem Ch P., Bautista-Molano W.. OP0171 Leptin Levels, Overweight and. Ann. Rheum. Dis. (2018) 77 135-136. DOI: 10.1136/annrheumdis-2018-eular.3166 19. Aletaha D., Neogi T., Silman A.J., Funovits J., Felson D.T., Bingham C.O., Birnbaum N.S., Burmester G.R., Bykerk V.P., Cohen M.D.. 2010 Rheumatoid Arthritis Classification Criteria: An American College of Rheumatology/European League Against Rheumatism Collaborative Initiative. Arthritis Rheum. (2010) 62 2569-2581. DOI: 10.1002/art.27584 20. Papapanou P.N., Sanz M., Buduneli N., Dietrich T., Feres M., Fine D.H., Flemmig T.F., Garcia R., Giannobile W.V., Graziani F.. Periodontitis: Consensus Report of Workgroup 2 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. J. Periodontol. (2018) 89 S173-S182. DOI: 10.1002/JPER.17-0721 21. Bello-Gualtero J.M., Lafaurie G.I., Hoyos L.X., Castillo D.M., De-Avila J., Munevar J.C., Unriza S., Londoño J., Valle-Onate R., Romero-Sánchez C.. Periodontal Disease in Individuals with a Genetic Risk of Developing Arthritis and Early Rheumatoid Arthritis: A Cross-Sectional Study. J. Periodontol. (2016) 87 346-356. DOI: 10.1902/jop.2015.150455 22. Kasper N.M., Herrán O.F., Villamor E.. Obesity Prevalence in Colombian Adults Is Increasing Fastest in Lower Socio-Economic Status Groups and Urban Residents: Results from Two Nationally Representative Surveys. Public Health Nutr. (2014) 17 2398-2406. DOI: 10.1017/S1368980013003418 23. Nikiphorou E., Guh D., Bansback N., Zhang W., Dixey J., Williams P., Young A.. Work Disability Rates in RA. Results from an Inception Cohort with 24 Years Follow-Up. Rheumatology (2012) 51 385-392. DOI: 10.1093/rheumatology/ker401 24. Uhlig T.. Which Patients with Rheumatoid Arthritis Are Still Working?. Arthritis Res. Ther. (2010) 12 114. DOI: 10.1186/ar2979 25. 25. Ministerio de Salud y Protección Social IV Estudio Nacional de Salud Bucal—ENSAB IVMinisterio de Salud y Protección SocialBogotá, Colombia2014. IV Estudio Nacional de Salud Bucal—ENSAB IV (2014) 26. Johansson L., Sherina N., Kharlamova N., Potempa B., Larsson B., Israelsson L., Potempa J., Rantapää-Dahlqvist S., Lundberg K.. Concentration of Antibodies against. Arthritis Res. Ther. (2016) 18 201. DOI: 10.1186/s13075-016-1100-4 27. Lafaurie G.I., Contreras A., Barón A., Botero J., Mayorga-Fayad I., Jaramillo A., Giraldo A., González F., Mantilla S., Botero A.. Demographic, Clinical, and Microbial Aspects of Chronic and Aggressive Periodontitis in Colombia: A Multicenter Study. J. Periodontol. (2007) 78 629-639. DOI: 10.1902/jop.2007.060187 28. Gómez R., Conde J., Scotece M., Gómez-Reino J.J., Lago F., Gualillo O.. What’s New in Our Understanding of the Role of Adipokines in Rheumatic Diseases?. Nat. Rev. Rheumatol. (2011) 7 528-536. DOI: 10.1038/nrrheum.2011.107 29. Le Sage F., Meilhac O., Gonthier M.-P.P.. Mol. Cell Endocrinol. (2017) 446 102-110. DOI: 10.1016/j.mce.2017.02.022 30. Jaedicke K.M., Roythorne A., Padget K., Todryk S., Preshaw P.M., Taylor J.J.. Leptin Up-Regulates TLR2 in Human Monocytes. J. Leukoc. Biol. (2013) 93 561-571. DOI: 10.1189/jlb.1211606 31. Zhao Y., Chen C., Preedy V.R., Hunter R.J.. Adipsin as Adipokine. Adipokines (2017) 20-27 32. Li Y., Zou W., Brestoff J.R., Rohatgi N., Wu X., Atkinson J.P., Harris C.A., Teitelbaum S.L.. Fat-Produced Adipsin Regulates Inflammatory Arthritis. Cell Rep. (2019) 27 2809-2816. DOI: 10.1016/j.celrep.2019.05.032 33. Chaparro J.A., Bello-Gualtero J.M., Valle-Oñate R., Bautista-Molano W., Chila L., Castillo D.M., Lafaurie G., Chalem Ch P., Romero-Sanchez C.. AB0120 Adiponectin Could Be a Mediator of the Presence of. Ann. Rheum. Dis. (2018) 77 1254 34. Mills E.L., Kelly B., O’Neill L.A.J.. Mitochondria Are the Powerhouses of Immunity. Nat. Immunol. (2017) 18 488-498. DOI: 10.1038/ni.3704 35. Moreno-Macias H., Romieu I.. Effects of Antioxidant Supplements and Nutrients on Patients with Asthma and Allergies. J. Allergy Clin. Immunol. (2014) 133 1237-1244. DOI: 10.1016/j.jaci.2014.03.020 36. Pisoschi A.M., Pop A.. The Role of Antioxidants in the Chemistry of Oxidative Stress: A Review. Eur. J. Med. Chem. (2015) 97 55-74. DOI: 10.1016/j.ejmech.2015.04.040 37. Quinonez-Flores C.M., Gonzalez-Chavez S.A., del Rio Najera D., Pacheco-Tena C.. Oxidative Stress Relevance in the Pathogenesis of the Rheumatoid Arthritis: A Systematic Review. Biomed. Res. Int. (2016) 2016 6097417. DOI: 10.1155/2016/6097417 38. Sezer U., Erciyas K., Üstün K., Pehlivan Y., Ziya Şenyurt S., Aksoy N., Tarakçıoğlu M., Taysı S., Onat A.M.. Effect of Chronic Periodontitis on Oxidative Status in Patients with Rheumatoid Arthritis. J. Periodontol. (2013) 84 785-792. DOI: 10.1902/jop.2012.120179 39. Martu M.A., Surlin P., Lazar L., Maftei G.A., Luchian I., Gheorghe D.N., Rezus E., Toma V., Foia L.G.. Evaluation of Oxidative Stress before and after Using Laser and Photoactivation Therapy as Adjuvant of Non-Surgical Periodontal Treatment in Patients with Rheumatoid Arthritis. Antioxidants (2021) 10. DOI: 10.3390/antiox10020226 40. Galiutina O., Shevchuk S., Seheda Y., Kuvikova I.. AB0130 Serum Leptin and Adiponectin Levels in Rheumatoid Arthritis Patients, Their Association with Inflammatory Process. Ann. Rheum. Dis. (2018) 77 1257-1258 41. Curtis J.R., van der Helm-van Mil A.H., Knevel R., Huizinga T.W., Haney D.J., Shen Y., Ramanujan S., Cavet G., Centola M., Hesterberg L.K.. Validation of a Novel Multibiomarker Test to Assess Rheumatoid Arthritis Disease Activity. Arthritis Care Res. (2012) 64 1794-1803. DOI: 10.1002/acr.21767
--- title: Physical Self-Concept and Physical Activity in Children with Congenital Heart Defects—Can We Point Out Differences to Healthy Children to Promote Physical Activity? authors: - Jannos Siaplaouras - Annika Jahn - Paul Helm - Kerstin Hanssen - Ulrike Bauer - Christian Apitz - Claudia Niessner journal: Children year: 2023 pmcid: PMC10047027 doi: 10.3390/children10030478 license: CC BY 4.0 --- # Physical Self-Concept and Physical Activity in Children with Congenital Heart Defects—Can We Point Out Differences to Healthy Children to Promote Physical Activity? ## Abstract Objective: Children with congenital heart defects (CHD) are at high risk for cardiovascular disease in addition to their congenital disease, so it is important to motivate this group of patients to live a physically active lifestyle. A potential influencing determinant of younger children’s physical performance is the physical self-concept. The objective of the present study was first to evaluate the correlation between the physical self-concept (PSC) and the participation in physical activities (PA) of a representative group of children with congenital heart disease (CHD), and second to point out differences in comparison to their healthy peer group. Methods: Using the database of PA of the S-BAHn-Study we focused on physical self-concept assessed by the German version of the Physical Self-Description Questionnaire. We compare the obtained data of children with CHD to a representative age-matched sample of 3.385 participants of the Motorik Modul Study. Results: $$n = 1$.198$ complete datasets could be included in the analyses. The mean age of patients was 11.6 ± 3.1 years. For the total cohort of patients with CHD and the reference group, PA correlated significantly with a positive PSC ($p \leq 0.001$). PA was significantly reduced in all groups of patients despite the severity of their heart defect ($p \leq 0.001$). Remarkably, PSC did not differ statistically significantly in patients with simple CHD from the reference collective ($p \leq 0.24$). Conclusions: According to this representative survey, there is a clear relation between PA and PSC in the cohort of healthy children and the group of children with CHD throughout the severity of their heart defects. Although PSC did not differ in patients with simple CHD and their healthy peer group, PA was significantly reduced. This gap invites us to reflect on how we could break new ground to promote a physically active lifestyle in children with CHD regardless of the severity of their cardiac defects. ## 1. Introduction Congenital heart defects are the most frequent congenital deformation in childhood. They occur in $1.1\%$ of in newborns [1]. Today, technical progress has reached the point where medical care and surgical methods exist that allow children with CHD to grow up very well [2]. However, in affected children, as in healthy children and adolescents, cardiovascular risk factors (i.e., cardiovascular disease) increase [3,4]. Myocardial infarction could become the major cause of death in CHD patients with simple cardiac defects [5]. This highlights the need for primary prevention. Physical activity (PA) is the key to well-being and a healthy life, especially in early childhood when children’s physical abilities mainly develop [6]. In general, only $12\%$ of children and adolescents achieve the WHO-recommended physical activity of at least 60 min per day [7]. In comparison, only $8.8\%$ of children and adolescents with CHD achieve these recommendations [8]. The lack of physical activity has several causes. Bjarnason-Wehrens [9] posited the “vicious cycle” as an explanatory approach. The lack of physical activity is created by overprotection and anxiety in the family of the affected person. This is reinforced by motor deficits. Motor deficits, in turn, increase anxiety and overprotection in the family. A “vicious cycle” develops. In addition, psycho-social situations become more difficult and the family’s radius of action is limited. To address the problem of overprotection, there are recommendations for assessing the physical fitness of children with CHD and detailed recommendations for physical activity even for competitive athletes with cardiovascular abnormalities [10,11,12,13]. Regarding actual recommendations, most patients with CHD are allowed to participate in competitive and leisure sports, unrestricted like healthy children. Only patients who are about to undergo surgery or who have life-threatening findings are not allowed to do sports [10]. Otherwise, all patients are allowed to do sports. However, in patients with severe or significant (residual) findings, attention should be paid to the type and load of the sport. One factor that influences children’s physical performance (e.g., sports participation) is the physical self-concept (PSC) [14]. The knowledge of this motivational process could thus be of great importance for improving PA. Therefore, research is of great interest in children with CHD. The body of studies on physical self-concept in children with a congenital heart defect is very limited [15,16,17]. Only the studies by Chen et al. [ 16] used a questionnaire specific to the physical self-concept, the Physical Self-Description Questionnaire (PSDQ). In the other two studies [15,17], physical self-concept was asked as a subscale in a questionnaire used for general self-concept. Chen et al. [ 15,17] only made comparisons among children with CHD. Thereby, both studies found a significantly better physical self-concept among boys compared to girls. Chen et al. [ 16] analyzed the difference between school children with CHD and healthy school children. They found significantly lower physical self-concept scores in children with CHD compared to healthy children. However, due to its small sample size, the study by Chen et al. [ 16] is only of limited significance for the overall population. Using the unique database of the S-Bahn Study [8], we aimed (I) to obtain representative data regarding the physical self-concept in children with CHD and its correlation to the severity of the heart defect; (II) to correlate PSC to the amount of PA; and (III) to detect differences between a representative group of children with and without CHD. ## 2.1. Study Design Ethical approval was obtained from Charité, Berlin (Approval number $\frac{2}{034}$/17) and the Karlsruhe Institute of Technology. The study protocol complies with the ethical guidelines of the 1975 Declaration of Helsinki. This study is a cross-sectional study based on the collection of data from January to March 2018 through questionnaires. The data are composed of two studies. One study included subjects with CHD (S-BAHN study) and the other study is available as a reference group (MoMo study). Within the S-BAHN study, we contacted patients via the patient database of the NRCHD, the largest registry for CHD patients in Europe. ## 2.2. Survey Instruments The validated questionnaire of the “Motorik-Modul” Study (MoMo), which is an in-depth study of the German Health Interview and Examination Survey for Children and Adolescents (KiGGS), was used to assess PSC and PA. The MoMo study has been recording the PA of children and adolescents at regular intervals of 5 years since 2003, and thus provides an important basis for a health monitoring of children in Germany. Using the MoMo, PAQ allowed us to compare the collected data with a representative age-matched reference collective of 3385 participants from the MoMo Wave 2 study (2015–2017) [18]. Further information, i.e., results and design of the MoMo baseline and longitudinal studies, is published elsewhere [19,20,21,22]. Detailed information about the MoMo Physical Activity Questionnaire (MoMo-PAQ), such as validity, content, and details on the structure, was published in high-ranking journals [19,20,21,22]. ## 2.2.1. Physical Activity (PA) The MoMo PAQ contains questions with single or multiple answers, and consists of 28 items on frequency, duration, and intensity of physical activity to assess habitual physical activity in different domains, for example physical activity in sports clubs, leisure time activity outside sports clubs, extracurricular physical activity, playing outdoors, and active commuting to school. ## 2.2.2. Physical Self-Concept (PSC) PSC was analyzed by using 36 items of the MoMo-PAQ, which assesses physical self-description based on the German version of the Physical Self-Description Questionnaire with response categories on a 4-point Likert scale [22]. These 36 items represent the basic functions of physical performance, namely, overall sporting skills (here referred to as “skill”), strength, endurance, speed, flexibility, and coordination. ## 2.3. Statistical Analysis Values of continuous variables are reported as mean ± standard deviation. Pearson’s chi-square test was used for group comparisons including nominal data (e.g., gender and age). In order to estimate the impact of potential contributing factors on PA, analysis of variance and covariance, Pearson’s correlation, and multiple and linear regression analysis were used. IBM SPSS statistics version 25.0 (IBM Inc., Armonk, NY, USA) was used for statistical analyses. A significance level of p ≤ 0.05 was applied. ## 3.1. Patient Characteristics Of 21.354 eligible patients, the invitation was successfully delivered to 14.496 patients of whom 1.718 patients participated in the study. $$n = 1$.198$ complete datasets could be included in the analyses. The mean age of patients was 11.6 ± 3.1 years, we could include $46.2\%$ females and $53.8\%$ males (Table 1 and Table 2). The patient characteristics of the study group were consistent with the overall cohort of registered eligible patients (Table 3). In the patient group, $47\%$ report that their parents had a high-school graduation (fathers ($47\%$) and mothers ($47.4\%$)), and in the reference collective $45.8\%$ of mothers and $43.9\%$ of fathers had a high-school graduation (Table 4). In total, $57.2\%$ of patients reported living in an urban or suburban area, whereas $42.8\%$ lived in a rural environment. We used Warnes’ categorization [23] to divide study participants into simple, moderate, and complex CHD. In total, $34.3\%$ ($$n = 411$$) were classified as simple CHD, $35.3\%$ ($$n = 423$$ moderate, and $30.4\%$ ($$n = 364$$) as complex CHD. Additionally, $20.5\%$ of the patients had more than three operations or other interventions, $30.2\%$ had 1–3 operations or other interventions, and the majority ($49.3\%$) had untreated CHD. Genetic syndromes were present in $5.8\%$ ($$n = 70$$), most frequently trisomy 21 with $3.1\%$ ($$n = 37$$), and Di-George syndrome in $1.3\%$ ($$n = 15$$) (Table 3). Table 1 modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). Table 2 \| Unnamed: 0 \| Simple CHD \| Moderate CHD \| Complex CHD \| \| --- \| --- \| --- \| --- \| \| Participants (n) \| 411 \| 423 \| 364 \| \| Participants (in percent) \| 34.3 \| 35.3 \| 30.4 \| \| Mean age * \| 11.18 ± 2.92Min.: 6; Max.: 17 \| 11.84 ± 3.04Min.: 6; Max.: 17 \| 11.64 ± 3.19Min.: 6; Max.: 17 \| \| Gender (m:f) in % \| 46:54 \| 55.1:44.9 \| 61:39 \| \| Operations (n)(Min.: 0; Max.: 18) \| 0.19 ± 0.63Min.: 0; Max.: 5 \| 1.4 ± 1.65Min.: 0; Max.: 10 \| 4.26 ± 3.26Min.: 0; Max.: 17 \| Table modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022) Table 3 \| Genetic Syndrome \| % \| \| --- \| --- \| \| Down Syndrome \| 3.1 \| \| Di-George Syndrome \| 1.3 \| \| CHARGE Syndrome \| 0.3 \| \| Chromosomal anomaly \| 0.3 \| \| Noonan Syndrome \| 0.2 \| \| VACTERL Association \| 0.2 \| \| Goldenhar Syndrome \| 0.1 \| \| Turner Syndrome \| 0.1 \| \| Williams-Beuren Syndrome \| 0.1 \| \| Klinefelter Syndrome \| 0.1 \| Table modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022) Table 4 \| Unnamed: 0 \| Patients (%) \| Reference (%) \| \| --- \| --- \| --- \| \| Father without graduation \| 0.5 \| 0.7 \| \| Father completed secondary school \| 16.5 \| 16.2 \| \| Father completed comprehensive school \| 31.3 \| 35.5 \| \| Father with high-school graduation \| 47.0 \| 45.8 \| \| Father with other educational qualifications \| 4.7 \| 1.8 \| \| Mother without graduation \| 0.8 \| 0.4 \| \| Mother completed secondary school \| 8.4 \| 9.1 \| \| Mother completed comprehensive school \| 40.5 \| 43.9 \| \| Mother with high-school graduation \| 47.4 \| 44.9 \| \| Mother with other educational qualifications \| 2.9 \| 1.7 \| Table modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). ## 3.2. Physical Activity According to WHO recommendations, children and adolescents should carry out at least an average of 60 min per day of moderate-to-vigorous intensity, mostly aerobic physical activity throughout the week [24]. Compared to the reference group, the total cohort of CHD patients was outstandingly less physically active (3.4 vs. 4 days physical activity per week, $p \leq 0.001$) and reached the level of 60 min of daily PA demanded by the WHO significantly less often (8.8 vs. $12\%$; $p \leq 0.001$). Total weekly activity decreased as the severity of the heart defect increased. Children with simple CHD were physically active 3.5 days per week, with moderate CHD 3.4 days, and with complex CHD 3.3 days (Figure 1 and Figure 2). Figure 1: Total physical activity in days per week (patients vs. reference). Figure 1 modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). Figure 2: Compliance with WHO guidelines (patients vs. reference). Figure 2 modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). ## 3.3. Physical Self-Concept The total patient cohort rated their ability to perform best in the categories of coordination and flexibility ($56.6\%$ and $56\%$ respectively). The worst result was achieved in the category of endurance. Only $28.5\%$ of all patients with CHD stated a positive self-concept for this item (Figure 3). The proportion of the patients with simple CHD reached a positive self-concept in all categories and did not differ significantly in any of the categories compared to the reference group (post hoc test: $p \leq 0.24$). Remarkably, patients with simple CHD rated their athletic performance even better than the reference group in the categories of coordination ($68\%$ vs. $64.9\%$) and mobility ($62.1\%$ vs. $56.8\%$) (Table 5). The participants with moderate CHD showed notably different mean values in each category (post hoc test: $p \leq 0.05$), except for the categories of strength (post hoc test: $p \leq 0.17$) and mobility (post hoc test: $p \leq 0.57$). Patients with complex CHD consistently rated themselves as worst and achieved significantly lower mean values in all categories than all other subgroups (post hoc test: $p \leq 0.01$). They achieved a positive self-concept least often in the endurance category ($16.8\%$). ## 3.4. Physical Self-Concept and Physical Activity PA correlated significantly with the achievement of a positive PSC. Respondents with simple CHD achieved the strongest correlation coefficients between total activity and physical self-concept. The higher the total physical activity was, the higher the achieved values in the physical self-concept and the more often a positive PSC was achieved. This correlation was clearest for the dimension “endurance” (Figure 3 and Table 5). The data showed that only $28.5\%$ of patients with CHD achieved a positive PSC in the category “endurance”. This result makes it clear that low endurance is a painful limitation for patients with CHD. Not least for this reason, dynamic sports that train endurance skills are particularly recommended for children and adolescents with CHD. On further analysis of the subgroups, it was noticeable that there was a difference between the severity of heart defect in terms of PSC. While a large proportion of patients with simple CHD achieved a positive PSC in all categories and the patients with simple CHD did not differ significantly from the reference collective in any category, the patients with complex CHD differed significantly from the other subgroups in the frequency of positive PSC in all categories and achieved a positive PSC less frequently than the other subgroups in all categories. Figure 3 modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). Figure 3: Frequency of a positive self-concept in the CHD group (in percent). TABLE_PLACEHOLDER:Table 5 Table modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). Figure 4 modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation. Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). Figure 4: Relationship between a positive self-concept and physical activity in the CHD group. There is a clear relationship between a positive physical self-description (%) of basic functions of physical performance and the level of PA reported in days per week (Figure 4 and Table 6). The percentage of patients reporting a positive self-description is significantly higher for those who are physically active on more than two days per week. Table 6 \| Unnamed: 0 \| Endurance \| Coordination \| Skills \| Strength \| Flexibility \| Speed \| \| --- \| --- \| --- \| --- \| --- \| --- \| --- \| \| Complex CHD \| 0.25 * \| 0.23 * \| 0.28 * \| 0.28 * \| 0.21 * \| 0.25 * \| \| Moderate CHD \| 0.33 * \| 0.27 * \| 0.34 * \| 0.33 * \| 0.24 * \| 0.29 * \| \| Simple CHD \| 0.41 * \| 0.36 * \| 0.45 * \| 0.38 * \| 0.36 * \| 0.33 * \| \| Reference \| 0.34 * \| 0.28 * \| 0.34 * \| 0.31 * \| 0.20 * \| 0.27 * \| Table modified according to Jahn, Annika [2022]: Körperliche Aktivität und Sportverhalten bei Kindern und Jugendlichen mit angeborenen Herzfehlern—eine deutschlandweite Analyse. Universität Ulm. Dissertation Open Access Repositorium der Universität Ulm und Technischen Hochschule Ulm. Dissertation. http://dx.doi.org/10.18725/OPARU-46553 (accessed on 13 October 2022). ## 4. Discussion Who am I? This question plays an important role in adolescence. It is a crucial contributor to the developmental process not only in healthy children and adolescents but also in children and adolescents with CHD. However, due to the disease burden, these children usually face more difficult challenges in this process, such as dealing with physical limitations [23,25,26]. Adequate physical activity is important for socio-emotional and physical development and a healthy life and is crucial to prevent affluence diseases. However, to the best of our knowledge, no research has evaluated PSC and PA simultaneously in a representative cohort of healthy children and children with CHD. Our study strived to explore the relations between PSC and PA. As expected, there was a correlation between the severity of the heart defect and the PSC of the participants. PSC correlated with the amount of PA. More precisely, this effect was statistically significant in healthy children, regardless of the severity of the heart defect for children with simple, moderate, and severe CHD. Viewed from another perspective, reduced PA correlated with an impaired PSC. Jekauc et al. [ 19] described physical self-concept as one of the determinants of physical activity in children. Thus, motivational strategies to increase positive physical self-concept and strategies to increase self-awareness may be important for maintaining physical activity, especially during the transition from adolescence to adulthood. Increasing physical self-concept can be realized with the help of psychological interventions or through specific thematization within a sports intervention program. Children with complex CHD were less physically active than their peers with CHD. Unexpectedly, children with simple CHD, which means children who normally are allowed to do unrestrictive leisure and competitive sports, were also notably less physically active compared to the healthy reference group. Interestingly, children with simple CHD did not differ in any of the PSC categories significantly from the reference group and rated their physical performance in the categories of coordination and flexibility even better than the reference group. Complementary to these results, it is important to state that children with CHD enjoy physical activity on a comparable level as the reference group of healthy children as published before [8]. Psychosocial as well as physical causes could be the reason for the differing results of patients with simple and complex CHD. One psychosocial reason for the difference in physical self-concept could be that the respondents with simple CHD were more likely to participate in sports with healthy people of the same age or with their peer group. For patients with complex heart defects, sports with peers could be hampered by several factors, including increased physician-advised sports restrictions, the severity of the heart defect and the associated reduced cardiopulmonary performance, or the lack of infrastructure in the sports club that would allow the integration of patients with complex heart disease into sports clubs with peers. Returning to the initial question, according to this study we are able to point out differences between healthy children and those with CHD. Preserved PSC cannot be considered an independent explanatory factor for the gap in PA in the cohort of children with simple CHD compared to their peer group. The bottom line is that, since physical activity is multifactorial, it is not surprising that self-concept is not sufficient to predict activity. We would certainly also need to know more about environmental factors (social support, clubs, playgrounds, parental attitudes, etc.). The presented intention–behavior gap in CHD patients implies that we should additionally focus on general conditions for sporting activities, such as new concepts in school and club sports and individualized advice. However, the major implication arising from this study is to promote impaired PA irrespective of the severity of CHD and beyond the burden of the heart defect itself. ## Strength and Limitation For this study, a questionnaire was chosen to collect the data. Questionnaires are suitable for quantitative interviewing because they are associated with low effort and costs. The disadvantage of questionnaires is that they are rather general and, in contrast to interviews, less detailed and cannot respond to the individual. In return, the influence of unintentionally and unconsciously dealing with each person differently is eliminated, resulting in deviations in measurement accuracy. Furthermore, the psychological effect of social desirability can lead to falsification of the answers [27]. The CHD patients were contacted via an online questionnaire and therefore had no opportunity to ask comprehension questions directly. To counteract this, participants had the option of calling the study leader or documenting comprehension questions at the end. For more accurate physical activity measurement data, directly measured physical activity using an accelerometer, for example, would be preferable [28]. However, this would lead to a significantly higher financial outlay and a smaller sample size would be the consequence. Questionnaires are therefore a suitable means of survey for large samples for this research question. One aspect that was not surveyed in this work is BMI. Due to the high number of overweight and obesity in CHD patients (Pinto et al., 2007; Tamayo et al., 2015) and the known influence on body perception and physical self-concept in healthy children and adolescents (Fernández-Bustos et al., 2019), it is an important parameter in terms of physical self-concept but could not be considered in this work. This work is a cross-sectional analysis. Thus, causal relationships cannot be derived. However, a cross-sectional analysis is sufficient for the presentation of descriptive information. It cannot be ruled out that the reference group also includes sick children. However, since this concerns only a minority, there should be no bias from a statistical point of view. Due to the large sample size and the cooperation with the NRCHD and the Motorik Modul study, representative data for Germany are presented. ## 5. Conclusions Our results from this nationwide survey suggest that PSC and PA are deeply connected in children with CHD. An improvement in one leads to an increase in the other and vice versa. However, since PA is multifactorial, the self-concept is not sufficient enough to predict individual activity. The circumstances leading to the outlined PA gap in children with simple CHD compared to their peer group are neither caused by impaired PSC nor by reduced enjoyment in sports. Possible factors causing this gap are framework conditions for sporting activities in Germany, such as a lack of new concepts in school and club sports and individualized advice for this group of children. To avoid a sedentary lifestyle, we should focus on these matters in further studies. ## 6. Patents This section is not mandatory but may be added if there are patents resulting from the work reported in this manuscript. ## References 1. Schwedler G., Lindinger A., Lange P.E., Sax U., Olchvary J., Peters B., Bauer U., Hense H.-W.. Frequency and spectrum of congenital heart defects among live births in Germany: A study of the competence network for congenital heart defects. Clin. Res. Cardiol. (2011) 100 1111-1117. DOI: 10.1007/s00392-011-0355-7 2. Helm P.C., Kaemmerer H., Breithardt G., Sticker E.J., Keuchen R., Neidenbach R., Diller G.-P., Tutarel O., Bauer U.M.M.. Transition in patients with congenital heart disease in Germany: Results of a nationwide patient survey. Front. Pediatr. (2017) 5 115. DOI: 10.3389/fped.2017.00115 3. Pinto N.M., Marino B.S., Wernovsky G., de Ferranti S.D., Walsh A.Z., Laronde M., Hyland K., Dunn S.O., Cohen M.S.. Obesity is a common comorbidity in children with congenital and acquired heart disease. Pediatrics (2007) 120 e1157-e1164. DOI: 10.1542/peds.2007-0306 4. Tutarel O.. Acquired heart conditions in adults with congenital heart disease: A growing problem. Heart (2014) 100 1317-1321. DOI: 10.1136/heartjnl-2014-305575 5. Olsen M., Marino B.S., Kaltman J., Laursen H., Jakobsen L., Mahle W., Pearson G., Madsen N.. Myocardial infarction in adults with congenital heart disease. Am. J. Cardiol. (2017) 120 2272-2277. DOI: 10.1016/j.amjcard.2017.08.050 6. Rodriguez-Ayllon M., Cadenas-Sánchez C., Estévez-López F., Muñoz N.E., Mora-Gonzalez J., Migueles J.H., Molina-García P., Henriksson H., Mena-Molina A., Martínez-Vizcaíno V.. Role of Physical Activity and Sedentary Behavior in the Mental Health of Preschoolers, Children and Adolescents: A Systematic Review and Meta-Analysis. Sports Med. (2019) 49 1383-1410. DOI: 10.1007/s40279-019-01099-5 7. 7. World Health Organization (WHO) Global Recommendations on Physical Activity for Health GenevaWorld Health OrganizationGeneva, Switzerland2010978. Global Recommendations on Physical Activity for Health Geneva (2010) 8. Siaplaouras J., Niessner C., Helm P.C., Jahn A., Flemming M., Urschitz M.S., Sticker E., Abdul-Khaliq H., Bauer U.M., Apitz C.. Physical Activity Among Children With Congenital Heart Defects in Germany: A Nationwide Survey. Front. Pediatr. (2020) 8 170. DOI: 10.3389/fped.2020.00170 9. Bjarnason-Wehrens B., Dordel S., Schickendantz S., Krumm C., Bott D., Sreeram N., Brockmeier K.. Motor development in children with congenital cardiac diseases compared to their healthy peers. Cardiol. Young (2007) 17 487-498. DOI: 10.1017/S1047951107001023 10. Takken T., Giardini A., Reybrouck T., Gewillig M., Hövels-Gürich H.H., Longmuir P.E., McCrindle B.W., Paridon S.M., Hager A.. Recommendations for physical activity, recreation sport, and exercise training in pediatric patients with congenital heart disease: A report from the Exercise, Basic & Translational Research Section of the European Association of Cardiovascular Prevention and Rehabilitation, the European Congenital Heart and Lung Exercise Group, and the Association for European Paediatric Cardiology. Eur. J. Prev. Cardiol. (2012) 19 1034-1065. DOI: 10.1177/1741826711420000 11. Budts W., Börjesson M., Chessa M., van Buuren F., Trindade P.T., Corrado D., Heidbuchel H., Webb G., Holm J., Papadakis M.. Physical activity in adolescents and adults with congenital heart defects: Individualized exercise prescription. Eur. Heart J. (2013) 34 3669-3674. DOI: 10.1093/eurheartj/eht433 12. Baumgartner H., De Backer J., Babu-Narayan S.V., Budts W., Chessa M., Diller G.P., Lung B., Kluin J., Lang I.M., Meijboom F.. ESC Scientific Document Group. 2020 ESC Guidelines for the management of adult congenital heart disease. Eur. Heart J. (2020) 42 563-645. DOI: 10.1093/eurheartj/ehaa554 13. Van Hare G.F., Ackerman M.J., Evangelista J.A., Kovacs R.J., Myerburg R.J., Shafer K.M., Warnes C.A., Washington R.L.. American Heart Association Electrocardiography and Arrhythmias Committee of Council on Clinical Cardiology, Council on Cardiovascular Disease in Young, Council on Cardiovascular and Stroke Nursing, Council on Functional Genomics and Translational Biology, and American College of Cardiology. Eligibility and Disqualification Recommendations for Competitive Athletes With Cardiovascular Abnormalities: Task Force 4: Congenital Heart Disease: A Scientific Statement From the American Heart Association and American College of Cardiology. Circulation (2015) 132 e281-e291. DOI: 10.1161/CIR.0000000000000240 14. Marsh H.W., Redmayne R.S.. A multidimensional physical self-concept and its relations to multiple components of physical fitness. J. Sport Exerc. Psychol. (1994) 16 43-55. DOI: 10.1123/jsep.16.1.43 15. Chen C.-W., Li C.-Y., Wang J.-K.. Self-concept: Comparison between school-aged children with congenital heart disease and normal school-aged children. J. Clin. Nurs. (2005) 14 394-402. DOI: 10.1111/j.1365-2702.2004.01068.x 16. Chen C.-W., Su W.-J., Wang J.-K., Yang H.-L., Chiang Y.-T., Moons P.. Physical self-concept and its link to cardiopulmonary exercise tolerance among adolescents with mild congenital heart disease. Eur. J. Cardiovasc. Nurs. (2015) 14 206-213. DOI: 10.1177/1474515114521926 17. Chen C.-W., Chen Y.-C., Beckstead J.W., Kennel S., Evans M.E.. R1 version, self-concept in Taiwanese adolescents with congenital heart disease. Pediatr. Int. (2011) 53 168-174. DOI: 10.1111/j.1442-200X.2010.03207.x 18. Helm P.C., Koerten M.A., Abdul-Khaliq H., Baumgartner H., Kececioglu D., Bauer U.M.. Representativeness of the Germany National Register for Congenital Heart Defects: A clinically oriented analysis. Cardiol. Young (2015) 26 921-926. DOI: 10.1017/S1047951115001547 19. Jekauc D., Wagner M.O., Kahlert D., Woll A.. Reliability and Validity of MoMo-Physical-Activity-Questionnaire for Adolescents (MoMo-AFB). Diagnostica (2013) 59 100-101. DOI: 10.1026/0012-1924/a000083 20. Woll A., Albrecht C., Worth A.. Motorik-Module (MoMo)—The KiGGS Wave 2 module to survey motor performance and physical activity. J. Health Monit. (2017) 2 66-73. DOI: 10.17886/RKI-GBE-2017-104 21. Wagner M.O., Bös K., Jekauc D., Karger C., Mewes N., Oberger J., Reimers A.K., Schlenker L., Worth A., Woll A.. Cohort Profile: The Motorik-Modul (MoMo) Longitudinal Study—Physical fitness and physical activity as determinants of health development in german children and adolescents. Int. J. Epidemiol. (2013) 43 1410-1416. DOI: 10.1093/ije/dyt098 22. Schmidt S., Henn A., Albrecht C., Woll A.. Physical activity of German children and adolescents 2003–2012: The MoMo-study. Int. J. Environ. Res. Public Health (2017) 14. DOI: 10.3390/ijerph14111375 23. Sticker E.J., Schickendantz S., Lohaus A.. Angeborene Herzfehler. Chronische Erkrankungen im Kindes und Jugendalter. Psychologische und Medizinische Grundlagen (2013) 24. Warnes C.A., Williams R.G., Bashore T.M., Child J.S., Connolly H.M., Dearani J.A., del Nido P., Fasules J.W., Graham T.P., Hijazi Z.M.. ACC/AHA 2008 guidelines for the management of adults with congenital heart disease. J. Am. Coll. Cardiol. (2008) 52 e143-e263. DOI: 10.1016/j.jacc.2008.10.001 25. Tong E.M., Sparacino P.S., Messias D.K., Foote D., Chesla C.A., Gilliss C.L.. Growing up with congenital heart disease: The dilemmas of adolescents and young adults see comment. Cardiol. Young (1998) 8 303-309. DOI: 10.1017/S1047951100006806 26. Yan H., Kantawang S., Yinghua Y.. Selected factors influencing self-concept among hospitalized Chinese school-age children with a chronic illness. Int. J. Nurs. Pract. (1999) 5 38-46. DOI: 10.1046/j.1440-172x.1999.00149.x 27. Rogers R., Reybrouck T., Weymans M., Dumoulin M., Van der Hauwaert L., Gewillig M.. Reliability of subjective estimates of exercise capacity after total repair of tetralogy of Fallot. Acta Paediatr. (1994) 83 866-869. DOI: 10.1111/j.1651-2227.1994.tb13159.x 28. Voss C., Harris K.C.. Physical activity evaluation in children with congenital heart disease. Heart (2017) 103 1408-1412. DOI: 10.1136/heartjnl-2017-311340
--- title: The Perceived Influence of Neurofibromatosis Type 1(NF1) on the Parents’ Relationship authors: - Lori Wiener - Sima Zadeh Bedoya - Archita Goyal - Mallorie Gordon - Natalie Deuitch - Brigitte Widemann journal: Children year: 2023 pmcid: PMC10047031 doi: 10.3390/children10030448 license: CC BY 4.0 --- # The Perceived Influence of Neurofibromatosis Type 1(NF1) on the Parents’ Relationship ## Abstract Neurofibromatosis type 1 (NF1) is a genetic condition affecting 1 in 3000 individuals. Having a child with a chronic illness can introduce both practical and emotional challenges to a parental relationship. This cross-sectional study was administered to 50 parents of children with NF1, diagnosed between the ages of 1–24. Each participant was provided a 50-item self-report survey to complete during an inpatient or outpatient visit. The survey gathered information on the participants’ views of the spouse/partner relationship, coping mechanisms, and elements that supported emotional connections. While the majority of parental relationships were reported to remain strong, the mean relationship quality was perceived to have decreased compared to prior to the child’s diagnosis. Compassionate and open communication, shared perspective, having time alone with their partner outside of medical situations, and dyadic coping were identified as strategies that could strengthen the relationship. The identified stressors to the parental relationship during the NF1 illness trajectory can inform interventions and help guide development of a couple’s intervention. The National Cancer Institute, NIH Institutional Review Board approved this study (12-C-0206). ## 1. Introduction Neurofibromatosis type 1 (NF1) is a genetic condition affecting 1 in 3000 individuals [1,2]. Approximately $50\%$ of cases are inherited from a parent while the other half are de novo [1]. NF1 is characterized by clinical features in almost all patients, most notably, the development of disfiguring, cutaneous neurofibromas, café-au-lait spots, macules, plexiform neurofibromas, and boney manifestations such as tibial dysplasia and scoliosis. [ 3]. NF1 causes tumors to grow along the nerves and is associated with an increased risk for the development of other cancers compared to the general population. In addition, NF1 often contributes to social and behavioral problems, including learning difficulties, development of Attention Deficit Hyperactivity Disorder (ADHD), and impaired social skills [4,5]. The manifestations of NF1 vary widely, even within families, from very mild to severely debilitating and manifestations often progress over time. The presence of NF1 can have a significant impact on a child’s psychological health and quality of life [5,6,7,8]. Faced with concerns regarding their child’s appearance, tumors, social and behavioral issues, and uncertainty about the future [3,8], NF1 can be a significant source of stress for parents. These issues, along with trying to manage daily life, can also put considerable strain on couples [4]. While data support a strong, positive association between family functioning and a child’s psychological health [9,10] and a negative impact on a child’s mental health when inter-partner conflicts exist [11,12], we found no studies that explored the impact of an NF1 diagnosis on the parental relationship. Several studies have been conducted among parents who have a child with a chronic illness including spina bifida, diabetes, and more acute conditions such as cancer [13,14,15]. A literature review was performed exploring the impact of a pediatric cancer diagnosis on couple functioning found overall strong adaptation in the areas of emotional closeness and general marital adjustment though difficulties in the domain of sexual intimacy and reports on conflict [16,17]. Studies have also demonstrated that dyadic coping strategies have helped couples cope better. Dyadic coping refers to the extent to which parents deal with stress as a dyad and has been identified as having a key role in individual and relationship functioning within couples facing significant stressors [18,19]. While there are many studies that examine the impact of childhood chronic illness on the family, factors that foster or inhibit parental relationship outcomes have received considerably less research attention. This cross-sectional study was designed to explore the impact of a child’s NF1 on the parents’ relationship by examining various aspects of the relationship before and after the NF1 diagnosis was made. We also aimed to explore dyadic adjustment, participant-identified issues of greatest stress individually and in the relationship, and interest in and challenges associated with couples counseling. ## 2.1. Participants Eligibility for the study required that at least three months had elapsed since the child’s NF1 diagnosis and that the parent had been in a partnered relationship at time of diagnosis. The diagnosis of NF1 was based on established clinical criteria. Confirmation of diagnosis using genetic germline testing was performed when a diagnosis could not be clearly made based on the clinical diagnostic criteria. Eligibility was extended to both parents, regardless of their relationship status at the time of study enrollment. ## 2.2. Procedures Study enrollment occurred when the child was present at the institution for scheduled appointments. Sixty-eight parents were approached. Four did not meet eligibility criteria. Of the eligible participants, 10 consented but did not complete the survey, 1 was not interested, 2 felt unable to participate due to lack of partner participation in child’s life or felt too much time had passed since their child was first diagnosed, and 1 had discrepant responses that could not be resolved. This resulted in a total cohort of 50 participants, and a $78\%$ response rate. Each participant was provided a self-report survey to complete during an inpatient or outpatient visit after reviewing and signing the informed consent. The study was approved by the Institutional Review Board at the National Cancer Institute, National Institutes of Health. ## 2.3. Measures The study team developed a 50-item self-report survey, including both multiple choice and open-ended questions for the purposes of this study. The survey collected information on demographics, as well as the participant’s view of their partner relationship, coping mechanisms, and stressors experienced secondary to the child’s NF1 (see Supplementary Material). To assess marital/relationship issues, the survey included The Revised Dyadic Adjustment Scale (RDAS), a standard measure of marital stress that gauges aspects of marital adjustment that are tested including consensus (e.g., decision making), cohesion (e.g., working together and discussion), and marriage satisfaction (e.g., frequency of quarrels and considering separation) [20], and is a revision of the 32-item Dyadic Adjustment Scale (DAS) [21]. When compared with the DAS, this measure has good construct validity. In addition, the RDAS has good internal consistency (Cronbach α = 0.90) and excellent split-half reliability (Guttman split half = 0.94 and Spearman–Brown split half = 0.95) with estimates larger than those of the DAS. To gather information on perceived changes in the relationship, three items related to emotional connection and managing stress effectively as a couple were adapted from the Gottman-17. This is a clinical marriage screening tool [20] that has been widely validated and well utilized as a measure of marital stability and divorce prediction [22]. Clinical consensus alongside a literature review guided question development for constructs related to specific challenges throughout the illness trajectory where no appropriate validated measures existed. The survey was first administered to parents of children with cancer and later adapted to parents of children with NF1. As noted in a prior publication [13], survey questions were pilot tested with 5 parents of children undergoing cancer treatment. Investigators assessed respondent comprehension of each item, ease of recall of past experiences, and ability to report personal experience. ## 2.4. Analysis Descriptive analyses allowed for sample characterization and definition of key variables. SPSS 28 software was used to perform statistical analyses. Qualitative analyses were conducted on free-text, open-ended responses to the question, “What recommendations would you make to a couple whose child was recently diagnosed in terms of trying to keep their relationship strong”. Independent analysis was performed by three investigators (L.W., A.G., S.B.) who read and coded text, and identifying emergent themes. Common themes across each question were then compiled, modified, and finalized based on team consensus. A coding dictionary was subsequently developed and applied to all responses. Discrepancies were reconciled through group review of the coded material until consensus was reached. ## 3.1. Sample Characteristics Fifty [50] parents of children aged 6 to 24 diagnosed with NF1 participated in this study. Participants included 32 individual parents and 9 “dyads” (9 groups of 2 people who identified being a parent of the same child) and ranged in age from 27 to 72 ($M = 46.22$, SD = ±8.85). Each participant was treated as an individual (enrollment, questions, and analyses). Table 1 includes additional information, such as marital status. The majority ($78\%$) self-reported their race as Caucasian. ## 3.2. Relationship Quality Forty-two percent of participants indicated strengthened relationships secondary to their child’s NF1, while $20.0\%$ reported the relationship was challenged but still strong, and another $20.0\%$ reported no change. Since time of diagnosis, few ($4.0\%$) participants had considered separation, and $14.0\%$ had separated. Most participants ($86.0\%$) rated their relationship in the year prior to their child’s NF1 diagnosis as good/very good/excellent and $14.0\%$ rated it as poor/fair. Similarly, the majority of participants ($68.0\%$) rated the quality of their current relationship as good/very good/excellent, and $32.0\%$ rated it as poor/fair. Nearly half ($42.0\%$) of participants indicated on the aforementioned questions that their relationship had moved in a negative direction. Treated as a continuous scale, on which “poor” = 1 and “excellent” = 5, participants’ responses demonstrated a decrease in the mean relationship quality over the previous year (3.70 vs. 3.16, $t = 3.5$, $$p \leq 0.001$$, Cohen = 0.50, indicating small-to-medium effect size). ## 3.3. Emotional Connection Participant perception of their relationship pre- and post-diagnosis of their child’s NF1 was assessed using a 5-point scale (never, sometimes, often, almost always, and always), which was then dichotomized to never/sometimes and often/almost always/always for analysis. Significant differences were found with participants reporting more loneliness (mean diff = −0.58, $p \leq 0.01$), less intimacy (mean diff = 0.72, $p \leq 0.001$), less shared decision-making (mean diff = 0.54, $p \leq 0.01$), more anger (mean diff = −0.30, $p \leq .05$), and more tension/stress in the relationship (mean diff = −0.44, $p \leq 0.05$). Participants were also asked how they were currently handling six different intrapersonal aspects of their relationship. Participants either responded “not a problem”, “sometimes a problem”, or “often a problem”, and this was dichotomized to “not a problem” and “sometimes”/”often a problem”. Feeling taken for granted, spending time together, and staying emotionally in touch were reported as sometimes/often a problem in the relationship ($64\%$, $64\%$, and $60\%$, respectively). ## 3.4. Most Stressful Issues to Self, Partner, and Relationship Participants were asked about potential stressors impacting their partners, their relationship, and/or themselves. Overwhelmingly, participants reported that fear of disease outcome was most stressful to themselves, while financial issues and (to a lesser degree) lack of intimacy, were most stressful to the relationship. Additionally, participants reported financial concerns and fear of disease outcome as issues they believed to be most stressful to their partners. See Table 2. ## 3.5. Handling Stress Twenty-seven participants ($54\%$) indicated they were managing stress related to caring for their child with NF1 “somewhat” effectively with their partner, $16\%$ reported handling their stress completely effectively, and $30\%$ reported they were challenged in this area. To further clarify the experienced stressors, participants were asked to rate certain areas that could be problematic for their relationship as “not a problem”, “sometimes a problem”, or “often a problem”. Areas rated to be problematic “sometimes” or “often” by over half of the participants included helping each other reduce daily stress (sometimes: $52\%$, often: $16\%$), talking about daily stresses together (sometimes: $46\%$, often: $22\%$), talking together about stress in a helpful manner (sometimes: $42\%$, often: $20\%$), listening with understanding about each other’s stresses and worries (sometimes: $42\%$, often: $18\%$), and partner taking job or other stresses out on the other (sometimes: $38\%$, often: $20\%$). ## 3.6. What Would Strengthen Your Marriage/Relationship Optional open- and closed-ended items were included to gather perspective on what would strengthen a couple’s relationship following a child’s NF1 diagnosis. Of those who endorsed practical suggestions in the closed question, participants most frequently suggested having the partner more involved in household chores ($30\%$). Those who endorsed “other” ($28\%$) tended to note that having their partner more involved with finances would strengthen their relationship. Responses to the open-ended question, “What recommendations would you make to a couple whose child was recently diagnosed in terms of trying to keep their relationship strong” reflected three major themes reflecting the value participants placed on the interpersonal aspects of their partner relationship. The themes included open and respectfulcommunication, shared perspective, having time alone with partner outside of medical situations, and dyadic coping strategies. ## 3.7. Open and Compassionate Communication Participants believed that open communication between themselves and their partner could help them understand their partner better and make their voices heard. The need for open communication permeated the responses as the following quote illustrates: “Talk frequently about your emotions. You can’t understand each other’s actions without understanding the emotions underneath”. ## 3.8. Sharing the Same Outlook Respondents reflected on the value of a shared outlook on current and future goals and methods of reaching their shared goals whether that was through further communication, education of their child’s illness, a supportive medical team, or uniting based on faith. Examples included, “Focus on your child—that helped us to maintain a strong relationship” and “Find good doctors so they can help you make good informative decisions”. ## 3.9. Having Time Outside of Illness Participants wrote about the importance of maintaining a strong family dynamic and emphasized the importance of maintaining a relationship with their partner that did not solely revolve around their child’s NF1. Examples included, “Put [your] relationship as a very high priority”; “It’s hard not to make a child’s NF1 the whole focus of one’s own life as a parent”; and “Be thankful for every day. Try not to stray from normalcy outside of appointments. Find ways to strengthen family bonds”. ## 3.10. Dyadic Strategies Dyadic adjustment was reflected by recognizing partners’ feelings and stresses, respecting their opinions, being supportive, and understanding how the other partner experiences stresses. Illustrative examples included: “We deal with the stresses differently. Try to respect each other’s process”; “I can remember standing in the hospital waiting room crying, with him just holding us. It meant a lot”; “Finding each other approachable and that the other cares. Feeling heard and understood”; and “When we just cried together in each other’s arms”. Participants also expressed that their relationship was strengthened by a capacity to be flexible, understand and adapt new family roles. The following quote illustrates this sentiment. “ My wife has to take care of the business and home while I am in charge of taking care of our son and communications with the doctors and everyone else involved in his treatments and care. In these times we are divided but working together as a team to accomplish what needs to be done”. ## 3.11. Interest, Timing, and Location of Counseling Services Interest in couples counseling to address strategies to support/strengthen their relationship if it were offered was gauged. Most ($58\%$) responded “yes”, $12\%$ responded “no”, and $30\%$ were “not sure”. When asked about appropriate timing for services to be offered, the majority of participants responded, “soon after diagnosis” ($28\%$) or “any point after diagnosis” ($28\%$), while others said, “first year after diagnosis” ($16\%$), “only if we ask” ($14\%$), or “other” ($14\%$). Most participants preferred that this service be offered in their home community ($50\%$) or in the hospital ($26\%$). When participants were asked to list topics they felt to be most important to address in counseling, the most common responses included open communication, financial guidance, knowing what to expect, understanding their spouse’s stress, and being heard. Several barriers were reported pertaining to participating in couples counseling. These included financial concerns ($58\%$), limited time ($50\%$), difficulty of both partners to be present at the same time ($24\%$), and difficulty of talking about relationship while child is receiving treatment ($18\%$). Others believed their partner would never do it ($16\%$), believed they did not need it ($14\%$), or were not interested ($8\%$). ## 4. Discussion This study explored the effect that a child’s NF1 diagnosis has on the parental relationship. Findings indicate that most couples were able to identify some effective means of coping with their situation, given that many reported that their relationship either stayed the same or improved over time. This differs from findings among parents of children with cancer, over half of whom reported that their relationship had been challenged following their child’s diagnosis [13]. These differences suggest that the stresses and impact of an acute illness such as cancer are not necessarily reflected in more chronic conditions. For a chronic illness such as NF1, couples have a longer time to adapt to the situation (often from the child’s birth). This may allow them to become more efficient in making decisions and communicating in areas pertaining to their child’s illness and acquire effective techniques over time. Despite reports of overall relationship strength in the wake of a child’s NF1 diagnosis, findings also demonstrated a decrease in average relationship quality as compared to the year prior to the child’s diagnosis. Participants reported experiencing negative impacts on aspects of emotional and sexual intimacy after receiving their child’s diagnosis. Several factors must be considered when looking at this data. The first is to differentiate between the “strength” of the relationship versus the “quality” of the relationship, particularly the quality of the relationship prior to the diagnosis. Perhaps the “strength” of the relationship improving or being maintained is more related to working together and feeling like the relationship is a partnership, whereas the “quality” may be more related to issues such as loss of intimacy. Next, the study tried to tease out what specific factors might play a role in the perception that the relationship changed in a negative direction. The participants identified issues that posed a stress to themselves, their partner, or their relationship. Fear of disease outcome and financial concerns were reported as the greatest stress to themselves and to their partner. In addition to changes in roles, responsibilities, emotional and physical needs, and living with uncertainty, the “what ifs” associated with disease outcome can be emotionally distressing and exhausting. Participants reported financial concerns as issues they believed to be most stressful to their partners. Having a child with a chronic condition can result in economic burden due to out-of-pocket medical expenses, lost wages, and lost career opportunities [23]. Along with the impact on the parent relationships, financial hardship has implications for both the emotional and mental health of the parent, as well as the well-being of the child. Comprehensive care must include efforts to identify and address family financial hardship [24]. If and how couples communicate with each other about these concerns can be critically important for the strength and health of the relationship. Along with specific external stresses, we explored specific problems within the relationship. Over half of the participants reported problems spending time together, staying emotionally in touch, and feeling taken for granted. When considering the relationship since the child’s diagnosis of NF1, participants also reported feeling lonelier in the relationship, a reduced sense of togetherness, and a lower degree of intimacy. Several studies with parents of children with cancer have reported the diagnosis negatively affected their level of physical intimacy and sexuality [13,25,26,27,28]. As noted above, the feeling of disconnection from one another may be a mechanism underpinning the quality of their relationship, particularly while caring for a child with chronic health care needs and can be used to explore what is happening in a parent’s relationship. It can also be used to develop support strategies to enhance communication of distress and emotional needs and strengthen the emotional connection between partners. In this regard, participants provided several suggestions for what can help strengthen the parental relationship (Table 3). The majority of participants expressed an interest in couples counseling shortly after diagnosis and in their home communities, yet $74\%$ were not offered this service. To increase access, counseling interventions could be integrated into the child’s treatment plan. thereby reducing barriers such as financial burden and time away from their child. The COVID-19 pandemic has demonstrated that virtual (telehealth) sessions can not only be more convenient for families, but also effective [29,30]. In addition, parents diagnosed with NF1 who are considering having more children, may benefit from reproductive genetic counseling [31]. Genetic counselors can target specific information or intervention needs. There are currently no evidence-based couples’ interventions that target disease-specific stresses to improve parental coping skills and decrease stress resulting from chronic or acute pediatric conditions. The data from this study can be used to inform intervention strategies. One such promising intervention is the couples coping enhancement training (CCET) model. This evidence-based 3-phase cognitive–behavioral intervention that utilizes dyadic coping strategies specifically addresses couples experiencing low marital satisfaction and high marital distress [32]. Interventions utilizing dyadic coping theory and strategies have shown great importance and promise in pediatric and adult oncology populations [33]. The study is limited by the small sample size and being a single-institution study. Participants had the capacity to travel to the National Institutes of Health to participate in care for their child living with NF1 and in this study. Due to the potential impact of selection bias, the results may not fully capture the experience of having a child with NF1 on the parents’ relationship. Additionally, its cross-sectional study design and varying length of time since diagnosis makes it subject to recall bias. Both parents of only nine children participated in the study. With this small sample size of couples, we did not conduct nested analyses. Prospective, longitudinal studies and the addition of a control group can provide important insights on how the relationship is impacted at different points in time, particularly if illness severity changes and if more than one child is living with NF1. Financial issues were a reported stressor. Future studies should further explore the factors involved, possibly including lost wages from having to care for a child with a chronic illness, medication costs, or other factors. With so few parent participants being diagnosed with NF1, we were not able to explore relationship or coping differences between familial and spontaneous NF1. This would be important to explore in future studies. Other limitations include only offering the study in English, underrepresentation of racial and ethnic minorities, and overrepresentation of highly educated parents. Future studies should include perspectives from culturally and linguistically diverse parents and focus on recruiting more parent dyads. Each of these factors could influence the generalizability of the results. There are some strengths of this study to note as well. Though men are not equally represented in the sample, they comprised over a third of the participants, a value higher than most caregiver studies. Additionally, to the best of our knowledge, this is the first study that evaluates the effects of a child’s NF1 on the parents’ relationship. Information learned from this study can contribute to the development of interventions that specifically address some of the unique stresses parents who have a child with NF1 experience in their relationship. The data may also be helpful to other parents where uncertainty of prognosis is a factor. Future research focusing on the couple’s functioning as parents and as partners, based on both parents’ reports, can best inform couple-based interventions. ## 5. Conclusions A critical component of ongoing psychosocial care is helping parents cope with specific disease stressors. Worsening parental relationships have a significant impact on a child’s mental health and well-being. Interventions should be tailored to disease specific-stressors and in both acute and chronic populations. Prospective research will aid in better understanding how chronic conditions, such as NF1, affect caregiver’s partnerships throughout the child’s development. ## References 1. Evans D.G., Howard E., Giblin C., Clancy T., Spencer H., Huson S.M., Lalloo F.. Birth incidence and prevalence of tumor-prone syndromes: Estimates from a UK family genetic register service. Am. J. Med. Genet. Part A (2010.0) 152A 327-332. DOI: 10.1002/ajmg.a.33139 2. Ferner R., Huson S., Thomas N., Moss C., Willshaw H., Evans G., Upadhyaya M., Towers R., Gleeson M., Steiger C.. Guidelines for the diagnosis and management of individuals with neurofibromatosis 1. J. Med. Genet. (2007.0) 44 81-88. DOI: 10.1136/jmg.2006.045906 3. Wolkenstein P., Rodriguez D., Ferkal S., Gravier H., Buret V., Algans N., Simeoni M.-C., Bastuji-Garin S.. Impact of neurofibromatosis 1 upon quality of life in childhood: A cross-sectional study of 79 cases. Br. J. Dermatol. (2009.0) 160 844-848. DOI: 10.1111/j.1365-2133.2008.08949.x 4. Barke J., Coad J., Harcourt D.. Parents’ experiences of caring for a young person with neurofibromatosis type 1 (NF1): A qualitative study. J. Community Genet. (2016.0) 7 33-39. DOI: 10.1007/s12687-015-0247-z 5. Rietman A.B., van der Vaart T., Plasschaert E., Nicholson B.A., Oostenbrink R., Krab L.C., Descheemaeker M.J., Wit M.Y., Moll H.A., Legius E.. Emotional and behavioral problems in children and adolescents with neurofibromatosis type 1. Am. J. Med. Genet. Part B Neuropsychiatr. Genet. (2018.0) 177 319-328. DOI: 10.1002/ajmg.b.32612 6. Roy A., Roulin J.L., Gras-Le Guen C., Corbat M.L., Barbarot S.. Executive functions and quality of life in children with neurofibromatosis type 1. Orphanet J. Rare Dis. (2021.0) 16 420. DOI: 10.1186/s13023-021-02051-5 7. Hummelvoll G., Antonsen K.. Young adults’ experience of living with neurofibromatosis type 1. J. Genet. Couns. (2013.0) 22 188-199. DOI: 10.1007/s10897-012-9527-5 8. Wolters P.L., Burns K.M., Martin S., Baldwin A., Dombi E., Toledo-Tamula M.A., Dudley W.N., Gillespie A., Widemann B.C.. Pain interference in youth with neurofibromatosis type 1 and plexiform neurofibromas and relation to disease severity, social-emotional functioning, and quality of life. Am. J. Med. Genet. (2015.0) 167A 2103-2113. DOI: 10.1002/ajmg.a.37123 9. Boettcher J., Denecke J., Barkmann C., Wiegand-Grefe S.. Quality of Life and Mental Health in Mothers and Fathers Caring for Children and Adolescents with Rare Diseases Requiring Long-Term Mechanical Ventilation. Int. J. Environ. Res. Public Health (2020.0) 17. DOI: 10.3390/ijerph17238975 10. Leeman J., Crandell J.L., Lee A., Bai J., Sandelowski M., Knafl K.. Family Functioning and the Well-Being of Children with Chronic Conditions: A Meta-Analysis. Res. Nurs. Health (2016.0) 39 229-243. DOI: 10.1002/nur.21725 11. Harold G.T., Sellers R.. Annual Research Review: Interparental conflict and youth psychopathology: An evidence review and practice focused update. J. Child Psychol. Psychiatr. (2018.0) 59 374-402. DOI: 10.1111/jcpp.12893 12. Wheeler N.J., Daire A.P., Barden S.M., Carlson R.G.. Relationship Distress as a Mediator of Adverse Childhood Experiences and Health: Implications for Clinical Practice with Economically Vulnerable Racial and Ethnic Minorities. Fam. Process (2019.0) 58 1003-1021. DOI: 10.1111/famp.12392 13. Wiener L., Battles H., Zadeh S., Pelletier W., Arruda-Colli M.N.F., Muriel A.C.. The perceived influence of childhood cancer on the parents’ relationship. Psycho-Oncology (2017.0) 26 2109-2117. DOI: 10.1002/pon.4313 14. Vermaes I.P., Janssens J.M., Bosman A.M., Gerris J.R.. Parents’ psychological adjustment in families of children with spina bifida: A meta-analysis. BMC Pediatr. (2005.0) 5. DOI: 10.1186/1471-2431-5-32 15. Loomba L.A., Lansing A.H., Cortez J.N., Welch K., Solowiejczyk J.N., Ghetti S., Styne D.M., Glaser N.S.. Parental marital relationship satisfaction is associated with glycemic outcomes in children with type 1 diabetes. J. Diabetes Metab. Disord. (2022.0) 21 1479-1489. DOI: 10.1007/s40200-022-01084-2 16. Van Schoors M., Caes L., Alderfer M.A., Goubert L., Verhofstadt L.. Couple functioning after pediatric cancer diagnosis: A systematic review. Psychooncology (2017.0) 26 608-616. DOI: 10.1002/pon.4204 17. Alba-Suarez J., Davidson S.L., Priebe C., Patel P., Greenspahn E., Boucher L., Rodríguez E.M.. Patient and sibling adjustment to pediatric cancer: The roles of parental relationship adjustment and depressive symptoms. J. Psychosoc. Oncol. (2021.0) 39 613-628. DOI: 10.1080/07347332.2020.1819932 18. Weitkamp K., Feger F., Landolt S.A., Roth M., Bodenmann G.. Dyadic Coping in Couples Facing Chronic Physical Illness: A Systematic Review. Front. Psychol. (2021.0) 12 722740. DOI: 10.3389/fpsyg.2021.722740 19. Bodenmann G., Revenson T.A., Kayser K., Bodenmann G.. Dyadic coping and its significance for marital functioning. Decade of Behavior. Couples Coping with Stress: Emerging Perspectives on Dyadic Coping (2005.0) 20. Spanier G.B.. Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. J. Marriage Fam. (1976.0) 38 15-28. DOI: 10.2307/350547 21. Busby D.M., Christensen C., Crane D.R., Larson J.H.. A revision of the Dyadic Adjustment Scale for use with distressed and nondistressed couples: Construct hierarchy and multidimensional scales. J. Marital Fam. Ther. (1995.0) 21 289-308. DOI: 10.1111/j.1752-0606.1995.tb00163.x 22. Gotten 17Areas Scale 23. Genereaux D., van Karnebeek C.D., Birch P.H.. Costs of caring for children with an intellectual developmental disorder. Disabil. Health J. (2015.0) 8 646-651. DOI: 10.1016/j.dhjo.2015.03.011 24. Pelletier W., Bona K.. Assessment of financial burden as a standard of care in pediatric oncology. Pediatr. Blood Cancer (2015.0) 62 619-631. DOI: 10.1002/pbc.25714 25. Greenberg H.S., Meadows A.T.. Psychosocial impact of cancer survival on school-age children and their parents. J. Psychosoc. Oncol. (1992.0) 9 43-56. DOI: 10.1300/J077v09n04_03 26. Enskar K., Carlsson M., Golsater M., Hamrin E., Kreuger A.. Parental reports of changes and challenges that result from parenting a child with cancer. J. Pediatr. Oncol. Nurs. (1997.0) 14 156-163. DOI: 10.1016/S1043-4542(97)90051-5 27. Ferrell B.R., Rhiner M., Shapiro B., Dierkes M.. The experience of pediatric cancer pain, Part I: Impact of pain on the family. J. Pediatr. Nurs. (1994.0) 9 368-379. PMID: 7837055 28. Lavee Y., Mey-Dan M.. Patterns of change in marital relationships among parents of children with cancer. Health Soc. Work (2003.0) 28 255-263. DOI: 10.1093/hsw/28.4.255 29. Orlowski E.W., Friedlander M.L., Megale A., Peterson E.K., Anderson S.R.. Couple and family therapists’ experiences with Telehealth during the COVID-19 pandemic: A phenomenological analysis. Contemp. Fam. Ther. (2022.0) 44 101-114. DOI: 10.1007/s10591-022-09640-x 30. Arem H., Moses J., Cisneros C., Blondeau B., Nekhlyudov L., Killackey M., Pratt-Chapman M.L.. Cancer Provider and Survivor Experiences with Telehealth during the COVID-19 Pandemic. JCO Oncol. Pract. (2022.0) 18 e452-e461. DOI: 10.1200/OP.21.00401 31. Radtke H.B., Bergner A.L., Goetsch L., McGowan C., Panzer K., Cannon A.. Genetic counseling for Neurofibromatosis 1, Neurofibromatosis 2, and Schwannomatosis–Practice Resource of the National Society of Genetic Counselors. J. Genet. Couns. (2020.0) 29 692-714. DOI: 10.1002/jgc4.1303 32. Bodenmann G., Shantinath S.D.. The couples coping enhancement training (ccet): A new approach to prevention of marital distress based upon stress and coping. Fam. Relat. (2004.0) 53 477-484. DOI: 10.1111/j.0197-6664.2004.00056.x 33. Van Schoors M., Loeys T., Goubert L., Berghmans G., Ooms B., Lemiere J., Norga K., Verhofstadt L.L.. Couples Dealing with Pediatric Blood Cancer: A Study on the Role of Dyadic Coping. Front. Psychol. (2019.0) 10 402. DOI: 10.3389/fpsyg.2019.00402
--- title: Sestrin2 Mediates Metformin Rescued the Age-Related Cardiac Dysfunctions of Cardiorenal Syndrome Type 3 authors: - Migdalia Iglesias - Hao Wang - Meredith Krause-Hauch - Di Ren - Linda Ines Zoungrana - Zehui Li - Jie Zhang - Jin Wei - Nikita Yadav - Kshama Patel - Mohammad Kasim Fatmi - Ruisheng Liu - Edward J. Lesnefsky - Ji Li journal: Cells year: 2023 pmcid: PMC10047033 doi: 10.3390/cells12060845 license: CC BY 4.0 --- # Sestrin2 Mediates Metformin Rescued the Age-Related Cardiac Dysfunctions of Cardiorenal Syndrome Type 3 ## Abstract Acute kidney injury (AKI) leads to acute cardiac injury and dysfunction in cardiorenal syndrome Type 3 (CRS3) through oxidative stress (OS). The stress-inducible Sestrin2 (Sesn2) protein reduces reactive oxygen species (ROS) accumulation and activates AMP-dependent protein kinase (AMPK) to regulate cellular metabolism and energetics during OS. Sesn2 levels and its protective effects decline in the aged heart. Antidiabetic drug metformin upregulates Sesn2 levels in response to ischemia–reperfusion (IR) stress. However, the role of metformin in CRS3 remains unknown. This study seeks to explore how the age-related decrease in cardiac Sesn2 levels contributes to cardiac intolerance to AKI-induced insults, and how metformin ameliorates CRS3 through Sesn2. Young (3–5 months) and aged (21–23 months) C57BL/6J wild-type mice along with cardiomyocyte-specific knockout (cSesn2−/−) and their wild type of littermate (Sesn2f/f) C57BL/6J mice were subjected to AKI for 15 min followed by 24 h of reperfusion. Cardiac and mitochondrial functions were evaluated through echocardiograms and seahorse mitochondria respirational analysis. Renal and cardiac tissue was collected for histological analysis and immunoblotting. The results indicate that metformin could significantly rescue AKI-induced cardiac dysfunction and injury via Sesn2 through an improvement in systolic and diastolic function, fibrotic and cellular damage, and mitochondrial function in young, Sesn2f/f, and especially aged mice. Metformin significantly increased Sesn2 expression under AKI stress in the aged left-ventricular tissue. Thus, this study suggests that Sesn2 mediates the cardioprotective effects of metformin during post-AKI. ## 1. Introduction Acute kidney injury (AKI) consists of an abrupt decrease in or cessation of kidney function, characterized by increased levels of plasma creatinine (PCr) [1]. AKI occurs in approximately $20\%$ of adults and $33\%$ of children hospitalized with acute illness, and $67\%$ of ICU patients [1,2]. Cardiorenal syndrome Type 3 (CRS3) is a subtype of cardiorenal syndrome (CRS) in which AKI leads to acute cardiac injury and dysfunction. AKI can either directly or indirectly lead to cardiac abnormalities in CRS3 [1]. The direct mechanisms through which AKI interacts with the heart include cellular apoptosis, the activation of the sympathetic nervous and renin–angiotensin–aldosterone systems, and oxidative stress [3]. Direct effects on cardiac function can consist of alterations in cardiomyocyte contractility, myocardial infarctions, and vasoconstriction [4], while indirect mechanisms include electrolyte imbalance, and the accumulation of fluid and uremic toxins [3]. The indirect cardiac effects of AKI can be hypertension, arrhythmias, and pericarditis [4]. Oxidative stress (OS) is the imbalance between the amount of reactive oxygen species (ROS) and antioxidants. OS can damage DNA, lipids, and proteins, leading to conditions such as cancer, and neurodegenerative and cardiovascular diseases [5,6]. Renal ischemia-reperfusion injury (IRI) is a common way via which AKI is induced, and is a risk factor for the development of CRS3 [5]. Renal IRI leads to ROS accumulation in both the kidneys and heart, which promotes cellular damage. Cellular and tissue damage induces an inflammatory response. Increased levels of proinflammatory factors such as tumor necrosis factors (TNFs) and interleukins (ILs) are found in the heart after renal IRI and during AKI [3]. Persistent exposure to these proinflammatory factors can incite cardiac functional and structural abnormalities, such as reduced fractional shortening, increased collagen volume fraction, and the presence of fibrosis [3,7]. These abnormalities are exacerbated by aging. Aging is a risk factor for CRS3, and is associated with ROS accumulation and low levels of cardiac Sestrin2 (Sesn2) [8]. Sesn2 is a stress-inducible protein that acts as a metabolic regulator by activating AMP-dependent protein kinase (AMPK) that, in turn, inhibits the mammalian target of rapamycin complex 1 (mTORC1) [9]. Sesn2 also displays antioxidant properties by reducing ROS accumulation, and providing cardioprotective effects against oxidative and ischemia–reperfusion (IR) stresses. Additionally, Sesn2 can repress cardiac proinflammatory signals through the downregulation of IL-17 signaling cascades [9]. Antidiabetic drug metformin contributes to the upregulation of Sesn2. Metformin can decrease ATP concentration in a cell through the AMPK-activated downregulation of mTORC1, leading to Sesn2 upregulation in response to energetic and ischemia–reperfusion (IR) stress [7]. During IR stress, metformin attenuates left-ventricular (LV) and mitochondrial dysfunction, inflammatory response, and ROS accumulation [10]. As discussed above, although the cardioprotective effects of Sesn2 and its upregulation through metformin administration in IRI have been reported, these effects in IR-induced AKI are not well-studied. This study hypothesizes that the aging-related decrease in cardiac Sesn2 renders the aged heart more vulnerable to injury and dysfunction following IR-induced AKI, and that metformin can help in ameliorating such cardiac stress via Sesn2. ## 2.1. Experimental Animals Young (3–5 months) C57BL/6J wild-type mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Aged (21–23 months) C57BL/6J wild-type mice were provided by the National Institute of Aging. Sesn2f/f mice with a C57BL/6J background were bred in our laboratory. Mice with a cardiomyocyte-specific knockout of Sestrin2 (cSesn2−/−) were generated in our laboratory from the breeding of Sesn2f/f mice and transgenic Cre mice (purchased from The Jackson Laboratory). Transgenic mice have an autosomally integrated Cre gene driven by the cardiac-specific alpha-myosin heavy chain promoter (αMHC). This animal protocol was approved by the Institutional Animal Care and Use Committee of the University of South Florida. ## 2.2. AKI Surgery and Sample Collection Male and female mice were randomly assigned to the sham, sham with metformin, AKI, and AKI with metformin groups. Metformin and vehicle group mice received a 21 µg/g metformin intraperitoneal (IP) injection and saline IP injection, respectively, 30 min before surgery. Mice were anesthetized with $1.5\%$ isoflurane and underwent a laparotomy followed by either a sham surgery or AKI surgery. AKI surgery consisted of clamping both renal arteries and veins for 15 min, followed by 24 h of reperfusion. The mice were then anesthetized as described above and euthanized via the rapid excision of the heart (IACUC # 9408R). The left ventricle (LV) and kidney of all mice were isolated and freeze-clamped in liquid nitrogen for the acquisition of total protein or placed in $4\%$ paraformaldehyde in PBS for histological examination (Figure 1). ## 2.3. Immunoblotting LV total protein underwent SDS-PAGE and electrotransfer to polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA) [6,9]. Rabbit Sesn2 antibody (Cat #10795-1-AP, Cat #21346-1-AP) from Proteintech® (Chicago, IL, USA) and Ab178518 from Abcam (Waltham, MA), were used following the manufacturer’s protocol. Additionally, rabbit GAPDH (Cat # 2118S) was obtained from Cell Signaling Technology (Danvers, MA, USA). Bands were detected with SuperSignal West Femto (Thermo Fisher Scientific, Waltham, MA, USA) and the ChemiDoc XRS+ Gel Imaging System (Bio-Rad Laboratories, Hercules, CA, USA). Intensity values were quantified and analyzed with Image Lab™ Software and expressed relative to GAPDH. ## 2.4. Mitochondrial Respiration Analysis The oxygen consumption rate (OCR) levels of isolated cardiomyocytes suspended in DMEM media was measured using the Mitochondrial Stress Test in the Seahorse X96 analyzer. Measurements were obtained under basal condition and with the addition of 2 µL of oligomycin, an ATP synthase inhibitor; 6 µL carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), an OXPHOS uncoupler that induces maximal respiration; 3 µL antimycin A, a Complex I inhibitor. The results were graphed with the Seahorse software and normalized to 4000 cells per well. ## 2.5. Echocardiography After the 24 h reperfusion period, the echocardiograms of all mice were performed using the Vevo 3100 imaging system (VisualSonics Inc., Toronto, ON, Canada) to evaluate cardiac function as previously described [11]. Simpson’s measurements were performed to obtain the systolic function values consisting of an LV ejection fraction (LVEF) and fractional shortening (LVFS) and the diastolic function assessed through the early-to-late ventricular filling velocity ratio (E/A) [12]. ## 2.6. Histopathology Renal and LV tissues fixed in $4\%$ paraformaldehyde in PBS were stained with hematoxylin and eosin (H&E) to assess the cellular morphology. LV tissues were stained with Masson’s Trichrome to evaluate the presence of collagenous tissue. The slides were imaged in a blinded fashion using the Keyence BZ-X710 All-in-One Fluorescence Microscope with 20X magnification. The percent collagen volume fraction (CVF) by area was calculated with Fiji ImageJ Version 2.3.0 (ImageJ Software, Madison, WI, USA). ## 2.7. Kidney Function Assessment To determine plasma creatinine concentration (PCr), blood samples were collected through the tail vein, and PCr was measured via HPLC at the O’Brien Center Core of the University of Alabama, Birmingham [13,14,15]. ## 2.8. Statistical Analysis Statistical analysis was performed using GraphPad Prism 9 (GraphPad Software, Inc., San Diego, CA). One- or two-way ANOVA was used to determine differences among two or more groups, with $p \leq 0.05$ deemed statistically significant. Data are expressed as means ± standard error of the mean (SEM). ## 3.1. Metformin Protects Renal Structure and Function following AKI Surgery An abrupt change in plasma creatinine concentration (PCr) is a strong indicator of acute kidney injury and dysfunction. The AKI surgery performed in this study significantly increased PCr levels, especially in the aged mice, compared to the sham (Figure 2A). Moreover, metformin was able to significantly reduce PCr levels in both aged and young mice with AKI. The AKI surgery also significantly increased the percentage of necrotic tubules in renal medullae and cortices in all four genotypes, especially in aged mice (Figure 2B–D). Metformin significantly decreased the percentage of necrotic tubules present in all mice after AKI surgery except in cSesn2−/− mice. These results indicate that the IR-AKI surgery successfully induced AKI conditions in all mice, with a more severe effect on the aged mice, and the known protective effects of metformin during IRI could also be seen during IR-AKI. ## 3.2. Metformin Preserves Cardiac Function under AKI Stress Once the successful induction of AKI conditions in the murine kidneys had been confirmed, the effects of CRS3 were evaluated through an assessment of cardiac function. To evaluate the effects of aging on cardiac function following AKI and metformin administration, echocardiograms were performed on young and aged wild-type (WT) mice (Figure 3A). As shown with the E/A ratio, the diastolic function of young WT mice, unlike that of their aged WT littermates, significantly deteriorated under the AKI condition compared to the sham condition. Systolic function, as assessed through LVEF and LVFS, significantly declined in both young and aged WT mice under the AKI condition as compared to the sham. However, a significant decrease in LVFS was not observed in the aged AKI mice when compared to the aged sham mice. Nevertheless, both genotypes showed a significant improvement in post-AKI systolic and diastolic function with metformin administration. To examine metformin’s ability to protect cardiac function via Sesn2, echocardiograms were also performed on Sesn2f/f and cSesn2−/− mice (Figure 3B). Although systolic and diastolic function significantly declined in both genotypes under the AKI condition, metformin was only able to significantly rescue systolic and diastolic function in the Sesn2f/f mice. Such results show that metformin could protect cardiac systolic and diastolic function through Sesn2. ## 3.3. Metformin Preserves Sesn2 Levels in the Heart under AKI Condition Immunoblotting was performed to assess Sesn2 expression in the left ventricle under the AKI condition and metformin treatment. Data indicated that AKI did not significantly affect Sesn2 expression levels in both the young and the aged group when compared to sham (Figure 4). However, the metformin injection significantly increased Sesn2 levels in aged but not in young hearts under both sham and AKI conditions (Figure 4). These data reveal that metformin administration can rescue impaired cardiac Sesn2 levels under physiological and pathological conditions in aging. However, metformin administration did not have a significant effect on cardiac Sesn2 levels in young mice. ## 3.4. Metformin Provides Protection from Myocardial Cellular and Fibrotic Damage The accumulation of ROS and inflammatory cytokines does not only affect cardiac function, but also leads to cellular damage and fibrosis. The presence of fibrosis in LV tissue was quantified through the percent collagen volume fraction (CVF) in all mice (Figure 5A). Aged, Sesn2f/f, and cSesn2−/− mice with AKI had significantly more fibrosis than that of their sham counterparts, with aged mice having the highest average CVF. AKI and metformin did not have a significant effect on the presence of fibrosis in young mice. Meanwhile, metformin significantly reduced fibrosis in the aged and Sesn2f/f mice with AKI, but not in the cSesn2−/− mice. These results suggest that AKI can lead to the development of myocardial fibrosis, with aged mice being more vulnerable, and that metformin is able to upregulate Sesn2 to save myocardial tissue from such abnormality. The health and integrity of myocardial tissue was further assessed through H&E staining (Figure 5B). Myocardial cells in the aged and cSesn2−/− mice under AKI stress were notably more damaged and experienced a greater loss of their ordered physiological organization as compared to their sham counterparts and to Sesn2f/f and young LV tissue under AKI stress. Such findings illustrate that the lower levels of Sesn2 in aged and cSesn2−/− mice failed to protect the LV from AKI-induced cellular damage. Despite these lower levels, metformin was still able to noticeably rescue the myocardial cells of both genotypes from AKI-induced damage. ## 3.5. Metformin Preserves Mitochondrial Function in the Heart under AKI Stress ROS accumulation causes dysfunctions in the electron transport chain (ETC) during IR stress. Therefore, the mitochondrial respiration of isolated cardiomyocytes was measured to assess the effects of IR-AKI and metformin on mitochondrial function. In aged (Figure 6A) and Sesn2f/f mice (Figure 6B), basal OCR levels significantly decreased after AKI, indicating a cardiomyocyte response to stress conditions and a significant decline in mitochondrial function. This decrease in basal OCR levels could also indicate low ATP demand, and the inhibition of ATP synthase and the ETC overall [16]. These levels significantly increased in the presence of metformin in both genotypes. The basal OCR levels of young (Figure 6A) and cSesn2−/− (Figure 6B) mice were not significantly affected by AKI or metformin administration. In young and cSesn2−/− mice, maximal OCR levels significantly decreased after AKI suggesting deterioration of the structural integrity of the ETC and mitochondria. Metformin did not influence maximal respiration after AKI in any of the four genotypes. ## 4. Discussion The results of this study suggest that the age-related decline in cardiac Sesn2 levels is attributed to more detrimental cardiac outcomes in CRS3 and that metformin can provide cardiac structural and functional protection against AKI stress through Sesn2. Fifteen minutes of ischemia-induced AKI followed by 24 h of reperfusion deteriorated renal structure and function through the augmentation of the percentage of necrotic tubules and levels of plasma creatinine, both of which were rescued by metformin. Additionally, metformin attenuated systolic and diastolic dysfunction, myocardial cell damage and fibrosis, and the loss of function in the mitochondria of cardiomyocytes in mice affected by AKI stress. These benefits were especially observed in aged mice and absent in cSesn2−/− mice. Overall, this study revealed that metformin could significantly improve cardiac functional and structural integrity during AKI stress via Sesn2. As previously mentioned, renal IRI induces an inflammatory response in the heart. Elevated levels of cytokines such as ILs and TNFs are associated with indicators of cardiac dysfunction such as a reduction in left ventricular ejection fraction and fractional shortening [1]. The detrimental effects of ischemic-AKI on cardiac systolic function could also be seen in this study through the significant decline in LVEF and LVFS in mice with AKI. While there was not a significant decrease in E/A and LVFS in aged mice with AKI compared to sham, aging alone contributed to a decline in systolic and diastolic function. Therefore, AKI only seemed to make a small contribution to the already existing cardiac dysfunction in aged mice. A study by Jo et al. found that, in rats, metformin can improve left ventricular systolic function after IRI by significantly increasing LVEF and LVFS and diastolic function through an increase in E/E′ ratio [17]. The present study also found that metformin can significantly increase LVEF and LVFS values and diastolic function after IR-AKI stress. The inability of metformin to do so in the cSesn2−/− mice is likely attributed to the decreased Sesn2 levels in this genotype, meaning that more Sesn2 is likely needed for metformin to provide its protective effects. The ongoing activation of inflammatory pathways can also incite the development of myocardial fibrosis. Myocardial fibrosis arises due to the excessive production of collagen from fibroblasts caused by aging, injury, and disease [18,19]. IRI leads to the replacement of damaged tissue with fibrotic scars originating from fibroblasts differentiated into myofibroblasts [20,21]. These myofibroblasts overproduce extracellular matrix proteins such as collagen causing fibrosis [21]. Here, a high amount of cardiac fibrotic tissue was found in aged, Sesn2f/f and cSesn2−/− mice with AKI as compared to sham conditions. A significant development of fibrosis in the young mice with AKI was not seen, while the aged mice with AKI had the highest CVF. This correlates with previous findings highlighting the increased collagen accumulation and fibrosis in aging animal and human models [22]. Increased fibrosis following IR-AKI was linked to hypertrophic cardiomyopathy and a loss of myocardial elasticity that decreases LVEF [23,24]. Here, a decrease in LVEF was indeed observed in the genotypes affected by fibrosis including aged, Sesn2f/f, and cSesn2−/− mice with AKI. Metformin rescues mice from fibrosis and inflammation following IRI [25]. The present study also reports a significant reduction in cardiac fibrotic tissue in IR-AKI with metformin treatment. AKI generates ROS accumulation in the heart. During ischemia, the electron transport chain (ETC) is in a reduced state; however, during reperfusion, the ETC reacts with oxygen causing the formation of ROS [19]. These ROS species can increase mitochondrial pore permeability, causing mitochondrial dysfunction and in turn more ROS production [26]. Mitochondrial ROS can trigger inflammatory cascades, and lead to cell death and cardiac dysfunction [19]. The mitochondrial respiration analysis in the present study showed that mitochondrial function was indeed impaired with IR-AKI compared to sham conditions. Metformin reduces mitochondrial ROS production by inhibiting reverse electron flow through complex 1 [27]. Metformin also decreases ATP concentration in the cell, which upregulates Sesn2 as a response [28]. Additionally, Ren et al. reported that Sesn2 offers a protective effect against mitochondrial damage in mice hearts after IR stress [29]. Consequently, given the metformin-driven decrease in ROS accumulation and upregulation of Sesn2, mitochondrial function should be significantly rescued by metformin following IRI stress. Indeed, this conclusion held true in the present study, where basal OCR levels were significantly increased in the presence of metformin in aged and Sesn2f/f mice but not in cSesn2−/− mice. In the present study, metformin did not influence maximal respiration OCR levels. Although few studies are available on the effects of metformin on the maximal respiration levels of isolated cardiomyocytes, other studies reported its effects on other cell types. Studies by Orang et al., and Geng et al. reported that metformin decreased maximal mitochondrial respiration in colorectal and hepatocellular carcinomas, respectively [30,31]. Immunoblotting results indicate that Sesn2 expression is significantly elevated in both sham and AKI conditions with metformin administration in Sesn2f/f and aged mice. Under AKI stress, cSesn2−/−mice also showed significantly elevated Sesn2 expression with metformin administration. Sesn2 knockout in cSesn2−/− mice is also specific only to cardiomyocytes. Therefore, Sesn2 is still present in other myocardial cell types such as endothelial cells, smooth muscle cells, fibroblasts, and immune cells [7]. Consequently, the observed expression and metformin-induced elevation of Sesn2 in cSesn2−/− tissue can be attributed to this fact. For future studies, performing immunoblotting with mice with a global Sesn2 knockout or with isolated Sesn2-knockout cardiomyocytes could yield different results. This study demonstrates that aging aggravates the negative effects of AKI on the heart, and that metformin can provide cardioprotective effects during IR-AKI through Sesn2. Furthermore, through AMPK activation, metformin improves LV function, incite anti-inflammatory action, and improve mitochondrial respiration and ATP generation [10]. Therefore, an additional investigation into the Sesn2-AMPK pathway could lead to better comprehension of the mechanism via which Sesn2 offers its cardioprotective effects during CRS3. Furthermore, under OS, Sesn2 is upregulated causing nuclear factor erythroid 2-related factor 2 (Nrf2) to travel from the cytoplasm to the nucleus, where it can offer protection against OS through the expression of Sesn2 [6,8,32]. Given this positive feedback loop, future studies into the connection between Nrf2 and Sesn2 during CRS3 could be beneficial. Additionally, given the importance of inflammatory responses on cardiac systolic and diastolic function and the development of myocardial fibrosis during CRS3, further investigation into the cardiac inflammatory cytokines present during CRS3 and their relationship to Sesn2 could also provide insights into the mechanism by which Sesn2 offers protection. Altogether, this study demonstrates that metformin’s ability to provide cardioprotective effects through Sesn2 could render it a potential target for the treatment of CRS3. ## References 1. Goyal A., Daneshpajouhnejad P., Hashmi M.F., Bashir K.. Acute Kidney Injury. StatPearls (2022.0) 2. Rewa O., Bagshaw S.M.. Acute kidney injury—Epidemiology, outcomes and economics. Nat. Rev. Nephrol. (2014.0) 10 193-207. DOI: 10.1038/nrneph.2013.282 3. Clementi A., Virzì G.M., Brocca A., De Cal M., Pastori S., Clementi M., Granata A., Vescovo G., Ronco C.. Advances in the Pathogenesis of Cardiorenal Syndrome Type 3. Oxidative Med. Cell. Longev. (2015.0) 2015 148082. DOI: 10.1155/2015/148082 4. Di Lullo L., Bellasi A., Barbera V., Russo D., Russo L., Di Iorio B., Cozzolino M., Ronco C.. Pathophysiology of the Cardio-Renal Syndromes Types 1–5: An Uptodate. Indian Heart J. (2017.0) 69 255-265. DOI: 10.1016/j.ihj.2017.01.005 5. Caio-Silva W., da Silva Dias D., Junho C.V.C., Panico K., Neres-Santos R.S., Pelegrino M.T., Pieretti J.C., Seabra A.B., De Angelis K., Carneiro-Ramos M.S.. Characterization of the Oxidative Stress in Renal Ischemia/Reperfusion-Induced Cardiorenal Syndrome Type 3. BioMed Res. Int. (2020.0) 2020 1605358. DOI: 10.1155/2020/1605358 6. Krause-Hauch M., Fedorova J., Zoungrana L.I., Wang H., Fatmi M.K., Li Z., Iglesias M., Slotabec L., Li J.. Targeting on Nrf2/Sesn2 Signaling to Rescue Cardiac Dysfunction during High-Fat Diet-Induced Obesity. Cells (2022.0) 11. DOI: 10.3390/cells11162614 7. Bates L., Fedorova J., Le V., Yang Y., Murphy J., Davitt K., Logan C., Ren D., Li J.. Abstract 12111: Metformin Maintains Mitochondrial Integrity and Preserve Heart from Ischemia Reperfusion Injury Through Sestrin2-AMPK Signaling. Circulation (2021.0) 144 A12111. DOI: 10.1161/circ.144.suppl_1.12111 8. Pasha M., Eid A.H., Eid A.A., Gorin Y., Munusamy S.. Sestrin2 as a Novel Biomarker and Therapeutic Target for Various Diseases. Oxidative Med. Cell. Longev. (2017.0) 2017 3296294. DOI: 10.1155/2017/3296294 9. Ren D., Quan N., Fedorova J., Zhang J., He Z., Li J.. Sestrin2 Modulates Cardiac Inflammatory Response through Maintaining Redox Homeostasis during Ischemia and Reperfusion. Redox Biol. (2020.0) 34 101556. DOI: 10.1016/j.redox.2020.101556 10. Lu Q., Li X., Liu J., Sun X., Rousselle T., Ren D., Tong N., Li J.. AMPK Is Associated with the Beneficial Effects of Antidiabetic Agents on Cardiovascular Diseases. Biosci. Rep. (2019.0) 39 BSR20181995. DOI: 10.1042/BSR20181995 11. Li X., Liu J., Hu H., Lu S., Lu Q., Quan N., Rousselle T., Patel M.S., Li J.. Dichloroacetate Ameliorates Cardiac Dysfunction Caused by Ischemic Insults Through AMPK Signal Pathway—Not Only Shifts Metabolism. Toxicol. Sci. (2019.0) 167 604-617. DOI: 10.1093/toxsci/kfy272 12. Gao S., Ho D., Vatner D.E., Vatner S.F.. Echocardiography in Mice. Curr. Protoc. Mouse Biol. (2011.0) 1 71-83. DOI: 10.1002/9780470942390.mo100130 13. Zhang J., Wang X., Cui Y., Jiang S., Wei J., Chan J., Thalakola A., Le T., Xu L., Zhao L.. Knockout of Macula Densa Neuronal Nitric Oxide Synthase Increases Blood Pressure in db/db Mice. Hypertension (2021.0) 78 1760-1770. DOI: 10.1161/HYPERTENSIONAHA.121.17643 14. Wei J., Zhang J., Wang L., Jiang S., Fu L., Buggs J., Liu R.. New mouse model of chronic kidney disease transitioned from ischemic acute kidney injury. Am. J. Physiol.-Ren. Physiol. (2019.0) 317 F286-F295. DOI: 10.1152/ajprenal.00021.2019 15. Zhang J., Wang X., Wei J., Wang L., Jiang S., Xu L., Qu L., Yang K., Fu L., Buggs J.. A two-stage bilateral ischemia-reperfusion injury-induced AKI to CKD transition model in mice. Am. J. Physiol.-Ren. Physiol. (2020.0) 319 F304-F311. DOI: 10.1152/ajprenal.00017.2020 16. Hill B.G., Benavides G.A., Lancaster J.R., Ballinger S., Dell’Italia L., Zhang J., Darley-Usmar V.M.. Integration of cellular bioenergetics with mitochondrial quality control and autophagy. Biol. Chem. (2012.0) 393 1485-1512. DOI: 10.1515/hsz-2012-0198 17. Jo W., Kang K.-K., Chae S., Son W.-C.. Metformin Alleviates Left Ventricular Diastolic Dysfunction in a Rat Myocardial Ischemia Reperfusion Injury Model. Int. J. Mol. Sci. (2020.0) 21. DOI: 10.3390/ijms21041489 18. Hinderer S., Schenke-Layland K.. Cardiac Fibrosis—A Short Review of Causes and Therapeutic Strategies. Adv. Drug Deliv. Rev. (2019.0) 146 77-82. DOI: 10.1016/j.addr.2019.05.011 19. Moris D., Spartalis M., Spartalis E., Karachaliou G.-S., Karaolanis G.I., Tsourouflis G., Tsilimigras D.I., Tzatzaki E., Theocharis S.. The role of reactive oxygen species in the pathophysiology of cardiovascular diseases and the clinical significance of myocardial redox. Ann. Transl. Med. (2017.0) 5 326. DOI: 10.21037/atm.2017.06.27 20. Talman V., Ruskoaho H.. Cardiac fibrosis in myocardial infarction—From repair and remodeling to regeneration. Cell Tissue Res. (2016.0) 365 563-581. DOI: 10.1007/s00441-016-2431-9 21. Kalogeris T., Baines C.P., Krenz M., Korthuis R.J.. Cell Biology of Ischemia/Reperfusion Injury. Int. Rev. Cell Mol. Biol. (2012.0) 298 229-317. DOI: 10.1016/B978-0-12-394309-5.00006-7 22. Aging and Cardiac Fibrosis—PMC 23. Ho C.Y., López B., Coelho-Filho O.R., Lakdawala N.K., Cirino A.L., Jarolim P., Kwong R., González A., Colan S.D., Seidman J.G.. Myocardial Fibrosis as an Early Manifestation of Hypertrophic Cardiomyopathy. N. Engl. J. Med. (2010.0) 363 552-563. DOI: 10.1056/NEJMoa1002659 24. Smiley D., Smith M.A., Carreira V., Jiang M., Koch S.E., Kelley M., Rubinstein J., Jones W.K., Tranter M.. Increased Fibrosis and Progression to Heart Failure in MRL Mice Following Ischemia/Reperfusion Injury. Cardiovasc. Pathol. (2014.0) 23 327-334. DOI: 10.1016/j.carpath.2014.06.001 25. Loi H., Kramar S., Laborde C., Marsal D., Pizzinat N., Cussac D., Roncalli J., Boal F., Tronchere H., Oleshchuk O.. Metformin Attenuates Postinfarction Myocardial Fibrosis and Inflammation in Mice. Int. J. Mol. Sci. (2021.0) 22. DOI: 10.3390/ijms22179393 26. Rubattu S., Mennuni S., Testa M., Mennuni M., Pierelli G., Pagliaro B., Gabriele E., Coluccia R., Autore C., Volpe M.. Pathogenesis of Chronic Cardiorenal Syndrome: Is There a Role for Oxidative Stress?. Int. J. Mol. Sci. (2013.0) 14 23011-23032. DOI: 10.3390/ijms141123011 27. Apostolova N., Iannantuoni F., Gruevska A., Muntane J., Rocha M., Victor V.M.. Mechanisms of Action of Metformin in Type 2 Diabetes: Effects on Mitochondria and Leukocyte-Endothelium Interactions. Redox Biol. (2020.0) 34 101517. DOI: 10.1016/j.redox.2020.101517 28. Ben-Sahra I., Dirat B., Laurent K., Puissant A., Auberger P., Budanov A., Tanti J.-F., Bost F.. Sestrin2 integrates Akt and mTOR signaling to protect cells against energetic stress-induced death. Cell Death Differ. (2013.0) 20 611-619. DOI: 10.1038/cdd.2012.157 29. Ren D., He Z., Fedorova J., Zhang J., Wood E., Zhang X., Kang D.E., Li J.. Sestrin2 Maintains OXPHOS Integrity to Modulate Cardiac Substrate Metabolism during Ischemia and Reperfusion. Redox Biol. (2021.0) 38 101824. DOI: 10.1016/j.redox.2020.101824 30. Orang A., Ali S.R., Petersen J., Mckinnon R.A., Aloia A.L., Michael M.Z.. A functional screen with metformin identifies microRNAs that regulate metabolism in colorectal cancer cells. Sci. Rep. (2022.0) 12 2889. DOI: 10.1038/s41598-022-06587-9 31. Geng Y., Hernández Villanueva A., Oun A., Buist-Homan M., Blokzijl H., Faber K.N., Dolga A., Moshage H.. Protective Effect of Metformin against Palmitate-Induced Hepatic Cell Death. Biochim. Biophys. Acta BBA-Mol. Basis Dis. (2020.0) 1866 165621. DOI: 10.1016/j.bbadis.2019.165621 32. Deshmukh P., Unni S., Krishnappa G., Padmanabhan B.. The Keap1–Nrf2 pathway: Promising therapeutic target to counteract ROS-mediated damage in cancers and neurodegenerative diseases. Biophys. Rev. (2017.0) 9 41-56. DOI: 10.1007/s12551-016-0244-4
--- title: Disparities in Access to Thoracic Surgeons among Patients Receiving Lung Lobectomy in the United States authors: - Sean J. Halloran - Christine E. Alvarado - Anuja L. Sarode - Boxiang Jiang - Jillian Sinopoli - Philip A. Linden - Christopher W. Towe journal: Current Oncology year: 2023 pmcid: PMC10047038 doi: 10.3390/curroncol30030213 license: CC BY 4.0 --- # Disparities in Access to Thoracic Surgeons among Patients Receiving Lung Lobectomy in the United States ## Abstract Objective: Lung lobectomy is the standard of care for early-stage lung cancer. Studies have suggested improved outcomes associated with lobectomy performed by specialized thoracic surgery providers. We hypothesized that disparities would exist regarding access to thoracic surgeons among patients receiving lung lobectomy for cancer. Methods: The Premier Hospital Database was used to identify adult inpatients receiving lung lobectomy from 2009 to 2019. Patients were categorized as receiving their lobectomy from a thoracic surgeon, cardiovascular surgeon, or general surgeon. Sample-weighted multivariable analysis was performed to identify factors associated with provider type. Results: When adjusted for sampling, 121,711 patients were analyzed, including 71,709 ($58.9\%$) who received lobectomy by a thoracic surgeon, 36,630 ($30.1\%$) by a cardiovascular surgeon, and 13,373 ($11.0\%$) by a general surgeon. Multivariable analysis showed that thoracic surgeon provider type was less likely with Black patients, Medicaid insurance, smaller hospital size, in the western region, and in rural areas. In addition, non-thoracic surgery specialty was less likely to perform minimally-invasive (MIS) lobectomy (cardiovascular OR 0.80, $p \leq 0.001$, general surgery OR 0.85, $$p \leq 0.003$$). Conclusions: In this nationally representative analysis, smaller, rural, non-teaching hospitals, and certain regions of the United States are less likely to receive lobectomy from a thoracic surgeon. Thoracic surgeon specialization is also independently associated with utilization of minimally invasive lobectomy. Combined, there are significant disparities in access to guideline-directed surgical care of patients receiving lung lobectomy. ## 1. Introduction Studies across multiple areas of surgery have found a relationship between fellowship-trained surgeons and improved outcomes of complex surgical procedures [1,2,3]. In patients receiving lung lobectomy for lung cancer, specialty training has been found to be an independent predictor of improved morbidity and mortality [4,5]. Furthermore, multiple studies have shown significant benefits to utilization of minimally invasive surgery (MIS) approaches for lung lobectomy in terms of decreased complications, hospital length of stay, and mortality [6,7,8]. A nationally representative study by Blasberg et al. found significantly higher utilization of MIS for lobectomy amongst thoracic surgeons in comparison to non-thoracic providers [8]. That study also showed significant geographic variation in practice. Blasberg noted that “VATS adoption appears to have slowed in specific regions of the country, where VATS lobectomy rates remain less than $40\%$”. [ 8] Regional variation has also been demonstrated in other surgical procedures in the United States [9]. Similar findings have been demonstrated within the field of thoracic surgery. Over $75\%$ of thoracic surgeons are employed through either hospital-based or academic/university-based practices, but the majority of thoracic procedures done in the community setting are performed by general surgeons [10]. Healthcare disparities have been well-documented within the United States [11,12,13]. The United States is approaching a critical shortage of surgeons, with resultant decreased patient access to surgical specialists [14]. Using data from the American Board of Thoracic Surgery and US Census Bureau, Moffatt-Bruce et al. expect the number of cardiothoracic procedures to increase by $61\%$ and the caseload of the average surgeon to increase by $121\%$ from 2010 to 2035 [15]. We believe that this trend towards increased cardiothoracic caseload will expose shortages in specialty care [16,17]. Specialty care is particularly important for thoracic oncology, where thoracic surgeons play a critical role in the work-up and evaluation of patients with lung cancer. The Nation Comprehensive Cancer Network recommends that decisions about lung cancer surgery “should be performed by thoracic surgeons” [18]. Despite these recommendations, many patients who receive lung resection are not receiving care by a thoracic surgeon. Regional and demographic factors associated with this disparity in access to thoracic surgeons are unknown. The purpose of this study was to evaluate historical disparities in access to both thoracic surgeons and minimally invasive approaches to surgery among patients receiving lung lobectomy in the United States. We queried a nationally representative database to look for social, racial, and regional differences that may impact how and by whom patients are receiving lung lobectomy. We hypothesize that disparities will be prevalent regarding access to both thoracic surgeons and the minimally invasive approach among lung lobectomy patients, a treatment modality also endorsed by current treatment guidelines [18]. ## 2.1. Data Source This study used the Premier Healthcare Database to analyze disparities and access to thoracic surgeon specialization among patients receiving lobectomy. The Premier Healthcare Database is a nationally representative database that contains de-identified clinical data from more than a thousand participating hospitals, capturing patient billing records, costs, and coding histories. It is comprised of data from more than one billion patient encounters, which equates to approximately twenty-five percent of all inpatient admissions in the United States. The database is maintained by Premier, Inc. (Washington, DC, USA) and contains hospital admissions (patient demographic characteristics), hospital characteristics, surgeon characteristics, payer information, diagnosis-related groups (DRGs), primary and secondary International Classification of Diseases (ICD) diagnosis and procedure codes, current procedural terminology codes, and resource utilization (hospital length-of-stay and in-hospital mortality). ## 2.2. Patient Selection The Premier Healthcare Database was queried for all adult inpatients age ≥ 18 years who received an elective lung lobectomy for lung cancer. Procedure codes and diagnosis codes were determined using ICD-9 and -10 version coding. All adult inpatient admissions between 2009 and 2019 were included. Patients were excluded if provider type was unknown, if discharge status was unknown, if a patient had a non-elective admission type, or if the patient’s visit status was not inpatient. Patients were categorized by the provider specialty performing the lobectomy: thoracic surgery, cardiovascular surgery, or general surgery. If patients had multiple provider specialties listed, the most specialized provider category was used (thoracic > cardiovascular > general surgery). For the purposes of this study, provider type was analyzed as thoracic vs. non-thoracic. The Elixhauser comorbidity score was generated from ICD-10 coding to estimate comorbidities using software from the Healthcare Cost and Utilization Project (HCUP). ## 2.3. Outcome Measures The outcome of interest was whether the lobectomy provider was categorized as a thoracic surgery specialist. The secondary outcome measure was whether the lobectomy procedure was performed MIS (defined as video-assisted thoracoscopy or robotic-assisted thoracoscopy) vs. open. ## 2.4. Statistical Analysis Survey methodology was used to correct for sampling such that patient level weighting derived from the Premier Healthcare Database was used to estimate a nationally representative sample. Hospital and patient characteristics associated with provider specialization were compared using bivariable analysis. Categorical variables were compared using the survey weight-adjusted Pearson’s χ2 test. Explanatory variables from the bivariable analysis that were significant were included in a survey-weighted multivariable logistic regression analysis to determine whether there were demographic and regional differences among provider specializations performing lobectomy. Lastly, given an a priori assumption that MIS is the preferred approach to surgery for lung cancer, we performed a multivariable analysis of factors associated with MIS that included provider specialization. Statistical analysis was performed using STATA MP (Version 17.0, Statacorp, College Station, TX, USA). Statistical significance was set at a p value ≤ 0.05. Since all patient-related data in the Premier Healthcare Database is aggregated, de-identified, and HIPAA-compliant, this study was determined to be exempt from institution review board review. ## 3. Results During the study period, the Premier Healthcare Database included 26,999 patients who met inclusion criteria representing an estimated population size of 121,711 lung lobectomy patients. Among them, 71,709 ($58.9\%$) had their surgery performed by a thoracic surgeon while the remaining 50,003 ($41.1\%$) were performed by non-thoracic providers: 36,630 ($30.1\%$) by a cardiovascular surgeon and 13,373 ($11.0\%$) by a general surgeon. These percentages and values are displayed in Figure 1 and Figure 2. Figure 3 represents trends in lobectomy by provider type during the study period. The proportion of lobectomies performed by thoracic surgeons decreased from 2009 to 2019 ($71.5\%$ vs. $54.6\%$). Unadjusted analysis revealed several differences between the two groups (Table 1). Patients receiving lobectomy from a thoracic surgeon were more likely to have private insurance ($24\%$ vs. $21\%$, $p \leq 0.001$), be treated at a hospital with >500 beds ($62\%$ vs. $38\%$, $p \leq 0.001$), be in an urban setting ($61\%$ vs. $40\%$, $p \leq 0.001$), and be treated at a teaching hospital ($58\%$ vs. $42\%$, $p \leq 0.001$). Multivariable analysis (Table 2) showed that thoracic surgeon provider type was less likely in Black patients (odds ratio (OR) 0.84, $p \leq 0.001$), Medicaid insurance (OR 0.84, $$p \leq 0.003$$), smaller hospital size (<300 beds: OR 0.86, $p \leq 0.001$, 300–499 beds: OR 0.80, $p \leq 0.001$), western region of the U.S. (OR 0.54, $p \leq 0.001$), and in rural areas (OR 0.38, $p \leq 0.001$). To determine factors associated with the MIS approach to lobectomy, a multivariable analysis was performed, which demonstrated similar disparities (Table 3). MIS was less likely in rural settings (OR 0.75, $p \leq 0.001$), non-teaching hospitals (OR 0.87, $$p \leq 0.0001$$), and in the western region (OR 0.49, $p \leq 0.001$). A non-thoracic surgeon specialist was also less likely to perform MIS lobectomy (cardiovascular OR 0.80, $p \leq 0.001$, general surgery OR 0.85, $$p \leq 0.003$$). In contrast, non-White patients were more likely to receive MIS lobectomy (Asian OR 5.62, $p \leq 0.001$, Black OR 1.75, $p \leq 0.001$). ## 4. Discussion Lung lobectomy is a complex procedure used to treat both benign and malignant lung disease and has been shown to have superior outcomes when performed by a thoracic surgeon [19]. Technological advancements have resulted in the increased adoption of minimally invasive techniques that have demonstrated superior outcomes in comparison to the traditional, open approach [6,7,8]. This study found multiple social, racial, and regional factors that significantly affected whether a patient receiving a lung lobectomy for cancer would be treated by a thoracic provider. Factors that were associated with a decreased likelihood of receiving care from a thoracic provider included non-White race, treatment in the western U.S., lower socioeconomic status, and rural hospital setting. These findings highlight disparities that exist within our current healthcare system regarding patient access to specialized surgical providers for lung lobectomy. In addition to these disparities observed for access to thoracic surgeons, similar demographic factors were also associated with a decreased likelihood of MIS utilization. Furthermore, non-thoracic providers were found to have decreased MIS utilization, emphasizing the importance of appropriate access to specialty surgical providers to improve patient outcomes. Bringing awareness to these disparities will help to facilitate discussion, strategies, and hopefully solutions that will increase the accessibility of specialized surgical providers, particularly among these vulnerable patient populations, in order to provide all lung cancer patients with the highest standard of surgical care. Race has long been recognized as a surrogate for other disparities in the medical field. Byrd et al. demonstrated significant disparities in the prevalence of racial and ethnic minorities that became apparent as early as the preschool years for these groups [20]. They found minority status to have influence on factors such as diet, physical activity, psychological factors, stress, income, and discrimination [20]. These findings have been found to hold true in the field of oncology, as well. Shavers et al. found evidence of racial disparities in receipt of definitive primary therapy, conservative therapy, and adjuvant therapy for patients with cancer [21]. Our study corroborates these disparities given our finding of non-White patients being less likely to receive care from a thoracic provider. Interestingly, our study found patients of black race to be more likely to undergo lung lobectomy via a minimally invasive approach; which goes against our original hypothesis. A plausible explanation for this revolves around the geographical distribution of non-white patients in the United States. Non-white patients make up the majority of the population in urban settings [22]. Similarly, the majority of academic medical centers are primarily located in densely populated urban centers. This combination could be the driving force behind increased adoption of minimally invasive lobectomy among non-white patients given that academic medical centers are more likely to utilize minimally invasive approaches [8]. The disparities found in this study have also been observed in other surgical sub-specialties, such as urology and obstetrics and gynecology [23,24]. Boyd et al. found that women of racial minorities who were eligible to receive minimally invasive hysterectomies were significantly more likely to have the procedure done via an open approach and were subject to increased adverse outcomes as a result [23]. In addition, a recently published study among lung and colorectal cancer patients examining the influence of race, insurance status, and rurality found that uninsured status was the largest predictor of receipt of surgery [25]. The gap in access to care is growing larger each year and will only continue to worsen the existing, non-modifiable disparities affecting patients. Increasing both patient access to specialized care through greater outreach and the number of practicing specialized providers are some of the only means to alleviate these disparities. Previous studies suggest that specialty training improves outcomes in patients undergoing lobectomy [4,5]. In addition, numerous studies show fewer complications, shorter length of hospital stay, and improved mortality when lobectomies are performed via a MIS approach [6,7,8]. In an analysis of Medicare patients, Farjah et al. found that, when adjusting for other patient, hospital, and surgeon factors, specialty training in general thoracic surgery was associated with a significantly decreased risk of death after pulmonary resection for cancer [4]. Based on these findings, referral to specialized thoracic surgeons remains a best practice for the surgical management of lung cancer. Surgical volume is another key factor to consider in terms of improving lobectomy outcomes [26]. Blasberg et al. found a significant association between MIS utilization and surgeon volume independent of surgeon specialty [8]. In our study, we found that, averaged over the five-year study period, thoracic surgeons had the highest annual rate of lobectomy performance in the country. However, we also found that the proportion of lobectomies performed by thoracic surgeons has decreased over the same five-year period, which may reflect the beginning of a decline in provider specialists. The findings of this study naturally beg the question of “what can be done to address these disparities?” Given the current fragmented state of the United States healthcare system, there is no simple solution to address these disparities. In the case of thoracic surgery, a plausible solution would be to implement regionalization of specialized thoracic surgery care. Although this approach may seem radical, regionalization of thoracic surgery is not an unprecedented policy. In 2007, Ontario, Canada implemented a policy to regionalize lung cancer surgery to 14 designated hospitals in the province [27]. This policy shift required significant support through government funding, but has resulted in shorter hospital length of stay and decreased mortality amongst certain populations undergoing thoracic procedures [27,28]. Within the United States, the Kaiser Permanente Northern California medical system implemented regionalization of higher complexity thoracic procedures such as lobectomy, bilobectomy, and pneumonectomy to five of the region’s 21 hospitals in 2014 [29]. Over a 3-year period, this hospital system demonstrated significant increases in pulmonary resection volume, adoption of a video-assisted thoracoscopic approach, and found regionalization to be independently associated with significant reductions in length of stay and morbidity. This system also demonstrated decreases in 1-year, 3-year, and overall mortality rates following implementation of regionalization [30]. These studies provide encouraging results to support the regionalization of thoracic care to address the disparities highlighted in our study. This study proves to be a timely contribution to the literature, given that our data summarizes the current state of access to thoracic surgeons for lung lobectomies up until the onset of the COVID-19 pandemic. Nation-wide lockdown resulted in a temporary halt to all elective procedures, including lung lobectomy, for an extended period of time. The long-term effects of the pandemic on access to care have yet to be properly characterized. However, Nguyen et al. found rebound increases in surgical volumes following staged reopening of their thoracic oncology program in response to significant drops in volume at the height of the pandemic [31]. Future studies are warranted to properly evaluate the impact of the pandemic on access to specialized thoracic providers. Our study has several limitations. This is a retrospective review of a large national database. As such, the Premier Healthcare Database lacks granular data surrounding clinical staging, neoadjuvant treatment, and intraoperative and surgical data, which therefore did not allow us to compare important oncologic and survival outcomes between the two groups. Given the lack of short- and long-term mortality data in this study, we hope this study may serve as the basis of future studies evaluating the impact of surgeon specialty on both short- and long-term mortality rates for specialized thoracic procedures such as lung lobectomy. Additionally, it is possible that the database lacks granularity in the administrative coding of race. PHD only represents roughly $25\%$ of all admissions in the United States. Thus, it is possible the database does not truly capture our patient population in the most homogenous manner. Another notable limitation of our study is that it may underestimate the disparities identified in this paper. Patients who did not have access to a specialty provider may not have been offered surgery, which is inferior to non-surgical care for early-stage lung cancer patients. ## 5. Conclusions In this study of a nationally representative database, factors associated with a decreased likelihood of patients receiving care from a thoracic provider for lung lobectomy included non-White race, lower socioeconomic status, western region, and receipt of care in a rural hospital setting. These factors, with the exception of race, were also associated with a decreased likelihood of a MIS approach to lobectomy. Since it has been demonstrated that both thoracic training and MIS result in better outcomes for lung lobectomy, it is necessary that the appropriate steps are taken to address these disparities and provide easier access to thoracic surgeons, providing all patients with the highest standard of care. ## References 1. Chowdhury M.M., Dagash H., Pierro A.. A Systematic Review of the Impact of Volume of Surgery and Specialization on Patient Outcome. Br. J. Surg. (2007) 94 145-161. DOI: 10.1002/bjs.5714 2. Cataneo J.L., Veilleux E., Lutfi R.. Impact of Fellowship Training on Surgical Outcomes after Appendectomies: A Retrospective Cohort Study. Surg. Endosc. (2021) 35 4581-4584. DOI: 10.1007/s00464-020-07879-9 3. Rhee D., Papandria D., Yang J., Zhang Y., Ortega G., Colombani P.M., Chang D.C., Abdullah F.. Comparison of Pediatric Surgical Outcomes by the Surgeon’s Degree of Specialization in Children. J. Pediatr. Surg. (2013) 48 1657-1663. DOI: 10.1016/j.jpedsurg.2012.12.048 4. Farjah F., Flum D.R., Varghese T.K., Symons R.G., Wood D.E.. Surgeon Specialty and Long-Term Survival after Pulmonary Resection for Lung Cancer. Ann. Thorac. Surg. (2009) 87 995-1004. DOI: 10.1016/j.athoracsur.2008.12.030 5. Schipper P.H., Diggs B.S., Ungerleider R.M., Welke K.F.. The Influence of Surgeon Specialty on Outcomes in General Thoracic Surgery: A National Sample 1996 to 2005. Ann. Thorac. Surg. (2009) 88 1566-1573. DOI: 10.1016/j.athoracsur.2009.08.055 6. Hernandez-Vaquero D., Vigil-Escalera C., Pérez-Méndez I., Gutiérrez A., Avanzas P., Wei Y., Diaz R., Silva J., Moris C., Pascual I.. Survival after Thoracoscopic Surgery or Open Lobectomy: Systematic Review and Meta-Analysis. Ann. Thorac. Surg. (2021) 111 302-313. DOI: 10.1016/j.athoracsur.2020.05.144 7. Paul S., Altorki N.K., Sheng S., Lee P.C., Harpole D.H., Onaitis M.W., Stiles B.M., Port J.L., D’Amico T.A.. Thoracoscopic Lobectomy Is Associated with Lower Morbidity than Open Lobectomy: A Propensity-Matched Analysis from the STS Database. J. Thorac. Cardiovasc. Surg. (2010) 139 366-378. DOI: 10.1016/j.jtcvs.2009.08.026 8. Blasberg J.D., Seder C.W., Leverson G., Shan Y., Maloney J.D., Macke R.A.. Video-Assisted Thoracoscopic Lobectomy for Lung Cancer: Current Practice Patterns and Predictors of Adoption. Ann. Thorac. Surg. (2016) 102 1854-1862. DOI: 10.1016/j.athoracsur.2016.06.030 9. Birkmeyer J.D., Reames B.N., McCulloch P., Carr A.J., Campbell W.B., Wennberg J.E.. Understanding of regional variation in the use of surgery. Lancet (2013) 382 1121-1129. DOI: 10.1016/S0140-6736(13)61215-5 10. Ikonomidis J.S., Boden N., Atluri P.. The Society of Thoracic Surgeons Thoracic Surgery Practice and Access Task Force-2019 Workforce Report. Ann. Thorac. Surg. (2020) 110 1082-1090. DOI: 10.1016/j.athoracsur.2020.04.004 11. Haider A.H., Scott V.K., Rehman K.A., Velopulos C., Bentley J.M., Cornwell E.E., Al-Refaie W.. Racial Disparities in Surgical Care and Outcomes in the United States: A Comprehensive Review of Patient, Provider, and Systemic Factors. J. Am. Coll. Surg. (2013) 216 482-492.e12. DOI: 10.1016/j.jamcollsurg.2012.11.014 12. Doherty R., Cooney T.G., Mire R.D., Engel L.S., Goldman J.M.. Envisioning a Better U.S. Health Care System for All: A Call to Action by the American College of Physicians. Ann. Intern. Med. (2020) 172 S3-S6. DOI: 10.7326/M19-2411 13. Levine A.A., de Jager E., Britt L.D.. Perspective: Identifying and Addressing Disparities in Surgical Access: A Health Systems Call to Action. Ann. Surg. (2020) 271 427-430. DOI: 10.1097/SLA.0000000000003572 14. Oslock W.M., Satiani B., Way D.P., Tamer R.M., Maurer J., Hawley J.D., Sharp K.L., Williams T.E., Pawlik T.M., Ellison E.C.. A Contemporary Reassessment of the US Surgical Workforce through 2050 Predicts Continued Shortages and Increased Productivity Demands. Am. J. Surg. (2022) 223 28-35. DOI: 10.1016/j.amjsurg.2021.07.033 15. Moffatt-Bruce S., Crestanello J., Way D.P., Williams T.E.. Providing Cardiothoracic Services in 2035: Signs of Trouble Ahead. J. Thorac. Cardiovasc. Surg. (2018) 155 824-829. DOI: 10.1016/j.jtcvs.2017.09.135 16. Slatore C.G., Au D.H., Gould M.K.. An Official American Thoracic Society Systematic Review: Insurance Status and Disparities in Lung Cancer Practices and Outcomes. Am. J. Respir. Crit. Care Med. (2010) 182 1195-1205. DOI: 10.1164/rccm.2009-038ST 17. Asokan S., Sridhar P., Qureshi M.M., Bhatt M., Truong M.T., Suzuki K., Mak K.S., Litle V.R.. Presentation, Treatment, and Outcomes of Vulnerable Populations with Esophageal Cancer Treated at a Safety-Net Hospital. Semin. Thorac. Cardiovasc. Surg. (2020) 32 347-354. DOI: 10.1053/j.semtcvs.2019.12.008 18. Ettinger D.S., Wood D.E., Aisner D.L., Akerley W., Bauman J.R., Bharat A., Bruno D.S., Chang J.Y., Chirieac L.R., D’Amico T.A.. NCCN Guidelines Insights: Non-Small Cell Lung Cancer, Version 2.2021. J. Natl. Compr. Cancer Netw. (2021) 19 254-266. DOI: 10.6004/jnccn.2021.0013 19. Howington J.A., Blum M.G., Chang A.C., Balekian A.A., Murthy S.C.. Treatment of Stage I and II Non-Small Cell Lung Cancer: Diagnosis and Management of Lung Cancer, 3rd Ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest (2013) 143 e278S-e313S. DOI: 10.1378/chest.12-2359 20. Byrd A.S., Toth A.T., Stanford F.C.. Racial Disparities in Obesity Treatment. Curr. Obes. Rep. (2018) 7 130-138. DOI: 10.1007/s13679-018-0301-3 21. Shavers V.L., Brown M.L.. Racial and ethnic disparities in the receipt of cancer treatment. J. Natl. Cancer Inst. (2002) 94 334-357. DOI: 10.1093/jnci/94.5.334 22. Mitchell T.. 1. Demographic and Economic Trends in Urban, Suburban and Rural Communities. Pew Research Center’s Social & Demographic Trends Project. (2018) 23. Boyd B.A.J., Winkelman W.D., Mishra K., Vittinghoff E., Jacoby V.L.. Racial and Ethnic Differences in Reconstructive Surgery for Apical Vaginal Prolapse. Am. J. Obstet. Gynecol. (2021) 225 e1-e405. DOI: 10.1016/j.ajog.2021.05.002 24. Yamoah K., Lee K.M., Awasthi S., Alba P.R., Perez C., Anglin-Foote T.R., Robison B., Gao A., DuVall S.L., Katsoulakis E.. Racial and Ethnic Disparities in Prostate Cancer Outcomes in the Veterans Affairs Health Care System. JAMA Netw. Open (2022) 5 e2144027. DOI: 10.1001/jamanetworkopen.2021.44027 25. Leech M.M., Weiss J.E., Markey C., Loehrer A.P.. Influence of Race, Insurance, Rurality, and Socioeconomic Status on Equity of Lung and Colorectal Cancer Care. Ann. Surg. Oncol. (2022) 29 3630-3639. DOI: 10.1245/s10434-021-11160-1 26. Smith C.B., Wolf A., Mhango G., Wisnivesky J.P.. Impact of Surgeon Volume on Outcomes of Older Stage I Lung Cancer Patients Treated via Video-Assisted Thoracoscopic Surgery. Semin. Thorac. Cardiovasc. Surg. (2017) 29 223-230. DOI: 10.1053/j.semtcvs.2017.01.013 27. Bendzsak A.M., Baxter N.N., Darling G.E., Austin P.C., Urbach D.R.. Regionalization and Outcomes of Lung Cancer Surgery in Ontario, Canada. J. Clin. Oncol. (2017) 35 2772-2780. DOI: 10.1200/JCO.2016.69.8076 28. Sundaresan S., McLeod R., Irish J., Burns J., Hunter A., Meertens E., Langer B., Stern H., Sherar M.. Early results after regionalization of thoracic surgical practice in a single-payer system. Ann. Thorac. Surg. (2013) 95 472-478. DOI: 10.1016/j.athoracsur.2012.10.001 29. Ely S., Jiang S.F., Patel A.R., Ashiku S.K., Velotta J.B.. Regionalization of Lung Cancer Surgery Improves Outcomes in an Integrated Health Care System. Ann. Thorac. Surg. (2020) 110 276-283. DOI: 10.1016/j.athoracsur.2020.02.019 30. Ely S., Jiang S.F., Dominguez D.A., Patel A.R., Ashiku S.K., Velotta J.B.. Effect of thoracic surgery regionalization on long-term survival after lung cancer resection. J. Thorac. Cardiovasc. Surg. (2022) 163 769-777. DOI: 10.1016/j.jtcvs.2021.03.050 31. Nguyen D.M., Kodia K., Szewczyk J., Alnajar A., Stephens-McDonnough J.A., Villamizar N.R.. Effect of COVID-19 on the delivery of care for thoracic surgical patients. JTCVS Open (2022) 10 456-468. DOI: 10.1016/j.xjon.2021.11.015
--- title: Transarterial Embolization of Ruptured Pancreaticoduodenal Artery Pseudoaneurysm Related to Chronic Pancreatitis authors: - Lucian Mărginean - Adrian Vasile Mureșan - Emil Marian Arbănași - Cătălin Mircea Coșarcă - Eliza Mihaela Arbănași - Eliza Russu - Rares Cristian Filep - Réka Kaller journal: Diagnostics year: 2023 pmcid: PMC10047043 doi: 10.3390/diagnostics13061090 license: CC BY 4.0 --- # Transarterial Embolization of Ruptured Pancreaticoduodenal Artery Pseudoaneurysm Related to Chronic Pancreatitis ## Body Figure 1: Aneurysmal dilatation of visceral arteries occurs in less than 1% of arterial aneurysm cases [1,2], primarily affecting the splenic artery, hepatic artery, and superior mesenteric artery; in only 1–2% of visceral aneurysms, it is found in the pancreaticoduodenal or gastroduodenal artery [1], with a high mortality rate in untreated cases [2,3,4]. Arterial aneurysmal dilatation develops as a result of damage to the collagen structure at the arterial wall caused by atherosclerosis or hypertension, and is found with the highest incidence at the level of the abdominal aorta [5,6,7,8,9]. In contrast, arterial pseudoaneurysms are connected with trauma, surgical treatments, or the presence of malignancies, but in the case of the pancreaticoduodenal artery, the main reason is the association with chronic pancreatitis [9,10,11,12,13,14,15,16]. Although it is a rare pathology, it is lethal in the absence of intervention. Endovascular treatment is currently the primary approach in the event of these pathologies, with a high success rate and excellent patient progression [4,9,13,14,15,16,17,18,19,20,21,22,23]. A 67-year-old woman presented with lightheadedness, diaphoresis, and acute epigastric and right hypochondrium pain. Along with other antecedents, her past medical history includes stage 2 essential hypertension, chronic ischemic cardiomyopathy, and class 1 obesity. Furthermore, three weeks before her current admission, she experienced a SARS-CoV-2 infection. She has no history of alcohol abuse, smoking, or abdominal trauma. An abdominal contrast-enhanced CT scan showed an extensive hematoma (3 × 4 cm2 in size) located intra-abdominally, adjacent to the duodenojejunal area, with hyperdensity around the duodenum and inferior to the pancreas (30–59 HU). Moreover, the CT scan also revealed an enhancing lesion as a pseudoaneurysm of the inferior pancreaticoduodenal artery measuring 5 × 8 × 8 mm3 with active bleeding and an associated hematoma. Additionally, the pancreas showed multiple hyperdense hemorrhagic structures anteriorly and superiorly under the aspect of possible acute hemorrhagic-ulcerative pancreatitis. Following these investigations of the abdominal area, a decision was made to proceed with an endovascular intervention within the interventional radiology department. (a) Transverse CT angiogram image shows a 12 mm enhancing lesion (white arrow) and a large intraperitoneal hematoma (arrowheads); (b) Sagittal MIP image depicts the enhancing lesion as a pseudoaneurysm (white arrow) of the inferior pancreaticoduodenal artery, responsible for the bleeding. Figure 2: With the patient under conscious sedation, via a right common femoral artery approach, the superior mesenteric artery was catheterized with a 5F Cobra catheter and a Terumo 35 guidewire (Figure 2a). While injecting the contrast agent to obtain a better working projection, the pseudoaneurysm ruptured, and acute extravasation of the contrast agent was noted (Figure 2b). A Direxion microcatheter (Boston Scientific, Marlborough, MA, USA) with a 14 guidewire was selectively advanced in the inferior pancreaticoduodenal artery and another small pancreatic branch (Figure 2c–e), followed by injection of a mixture of 1 mL Glubran 2 (GEM, Viareggio, Italy) with 2 mL Lipiodol (Guerbet, Villepinte, France) until complete obliteration of the pseudoaneurysm was obtained (Figure 2f). The patient was hemodynamically stable at the end of the procedure and was discharged 6 days later in a good condition without any active bleeding signs. (a) Superior mesenteric artery contrast injection reveals an oval-shaped pseudoaneurysm; (b) Acute contrast extravasation during contrast injection; (c) Residual filling of the pseudoaneurysm through collaterals, despite IPD artery occlusion with glue; (d) Collateral branch responsible for residual filling; (e) Microcatheter in the inferior pancreaticoduodenal artery; (f) Complete obliteration of the pseudoaneurysm with the glue cast visible. Given the high morbidity and fatality rates associated with pancreaticoduodenal pseudoaneurysm rupture [23], immediate treatment is vital. Several recent studies in the literature demonstrate promising outcomes for endovascular embolization. Suzuki et al. [24] showed a series of seven patients embolized using micro-coils, with a 100% success rate and no long-term (28 months average, range 5–65 months) recurrence of symptoms or bleeding. Waguri et al. [23] also reported effective embolization of an anterior inferior pancreaticoduodenal artery pseudoaneurysm that perforated in the portal system. Furthermore, Krishna et al. [20] described a 60-year-old female patient who presented with melena and hematemesis caused by a rupture of the lower pancreaticoduodenal artery pseudoaneurysm at the level of the intestinal wall. Ren et al. [21] reported on their endovascular treatment of 159 patients with visceral artery aneurysms and pseudoaneurysms, with 96.9% of patients successfully treated and a 1.9% mortality rate at 30 days. Similar to our case, Mitrovic et al. [13,14,15] reported successful endovascular resolution of a posterior inferior pancreaticoduodenal artery pseudoaneurysm using the sandwich technique, in which the pseudoaneurysm inflow and outflow were embolized with coils. Jang et al. [16], on the other hand, embolized the pseudoaneurysm with an N-Butyl-Cyanoacrylate-lipiodol composition. Gurala et al. [17] also reported an effective embolization of a pseudoaneurysm from the gastro-duodenal artery, using coils. Inferior pancreaticoduodenal artery pseudoaneurysms may be diagnosed using ultrasound, CT, or visceral angiography with sensitivities of 50%, 67%, and 100%, respectively. An abdominal contrast-enhanced CT scan is usually sufficient for the appropriate identification of visceral pseudoaneurysms, but it should be confirmed through angiography and treated if necessary [13,14]. Treatments may involve direct thrombin injections, occlusive balloon catheters, surgical ligation, or percutaneous transcatheter embolization with coils or synthetic particles [9]. Vascular embolization with NBCA should proceed with extensive and careful evaluation of the vascular anatomy and close attention to technical details. NBCA is typically mixed with iodized oil (Lipiodol) for visualization under X-rays and to adjust the polymerization rate. Interventional radiologists should become more familiarized with these types of embolic agents, as they can be used as first-line treatment for various peripheral abnormalities [25]. Pseudoaneurysms of the visceral arteries are identified in up to 60% of cases at the level of the splenic artery, 25% of cases at the level of the renal arteries, and up to 20% of cases at the level of the hepatic artery. Additionally, with a considerably lower frequency of just under 5%, they are seen at the level of the celiac trunk, the mesenteric arteries, and in the smallest proportion at the level of the gastroduodenal and pancreaticoduodenal arteries [26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42]. Each patient with a visceral artery aneurysm should be treated individually and followed by a multidisciplinary team involving vascular and general surgeons, gastroenterologists, and radiologists for the best management choice. Endovascular therapy is the first option for these patients, based on our expertise and research published in the literature. ## Abstract We presented a 67-year-old woman with lightheadedness, diaphoresis, and acute epigastric and right hypochondrium pain, with a past medical history including stage 2 essential hypertension, chronic ischemic cardiomyopathy, and class 1 obesity. An abdominal contrast-enhanced CT scan showed an extensive hematoma (3 × 4 cm2 in size) located intra-abdominally, adjacent to the duodenojejunal area, with hyperdensity around the duodenum, positioned inferior to the pancreas (30–59 HU). Moreover, the CT scan also revealed an enhancing lesion as a pseudoaneurysm of the inferior pancreaticoduodenal artery, measuring 5 × 8 × 8 mm3 with active bleeding and associated hematoma. Following these investigations of the abdominal area, a decision was made to proceed with an endovascular intervention within the interventional radiology department. With the patient under conscious sedation, via a right common femoral artery approach, the superior mesenteric artery was catheterized. While injecting the contrast agent to obtain a better working projection, the pseudoaneurysm ruptured, and acute extravasation of the contrast agent was noted, followed by injection of a mixture of 1 mL Glubran 2 with 2 mL Lipiodol until complete obliteration of the pseudoaneurysm was obtained. The patient was hemodynamically stable at the end of the procedure and was discharged 6 days later in a good condition without active bleeding signs. ## References 1. Jesinger R.A., Thoreson A.A., Lamba R.. Abdominal and Pelvic Aneurysms and Pseudoaneurysms: Imaging Review with Clinical, Radiologic, and Treatment Correlation. Radiogr. Rev. Publ. Radiol. Soc. N. Am. Inc. (2013) 33 E71-E96. DOI: 10.1148/rg.333115036 2. Bradley S., Quenzer F., Wittler M.. Ruptured Visceral Artery Aneurysms: A Deadly Cause of Epigastric Pain. Clin. Pract. Cases Emerg. Med. (2019) 3 132-136. DOI: 10.5811/cpcem.2019.1.41057 3. Kiviluoto T., Kivisaari L., Kivilaakso E., Lempinen M.. Pseudocysts in Chronic Pancreatitis. Surgical Results in 102 Consecutive Patients. Arch. Surg. Chic. Ill 1960 (1989) 124 240-243. DOI: 10.1001/archsurg.1989.01410020114019 4. Novacic K., Vidjak V., Suknaic S., Skopljanac A.. Embolization of a Large Pancreatic Pseudoaneurysm Converted from Pseudocyst (Hemorrhagic Pseudocyst). JOP J. Pancreas (2008) 9 317-321 5. Arbănași E.M., Mureșan A.V., Coșarcă C.M., Arbănași E.M., Niculescu R., Voidăzan S.T., Ivănescu A.D., Hălmaciu I., Filep R.C., Mărginean L.. Computed Tomography Angiography Markers and Intraluminal Thrombus Morphology as Predictors of Abdominal Aortic Aneurysm Rupture. Int. J. Environ. Res. Public Health (2022) 19. DOI: 10.3390/ijerph192315961 6. Russu E., Mureșan A.V., Kaller R., Toma L., Coșarcă C.M., Chibelean C.B., Arbănași E.M., Arbănași E.M.. Innovative Technical Solution Using the Renal Artery Stump after Nephrectomy as an Inflow Artery for Lower Limb Revascularization—A Case Report. Front. Surg. (2022) 9 864846. DOI: 10.3389/fsurg.2022.864846 7. Arbănași E.-M., Russu E., Mureșan A.V., Arbănași E.-M.. Late Rupture of a Thrombosed Aortic Abdominal Aneurysm—A Case Report. J. Cardiovasc. Emerg. (2021) 7 84-87. DOI: 10.2478/jce-2021-0012 8. Kaller R., Mureșan A.V., Popa D.G., Arbănași E.-M., Russu E.. Fatal Aortoduodenal Fistula Caused by a Ruptured Abdominal Aortic Aneurysm—A Case Report. J. Cardiovasc. Emerg. (2021) 7 129-132. DOI: 10.2478/jce-2021-0015 9. Dallara H., Habboushe J.. Spontaneous Inferior Pancreaticoduodenal Artery Pseudoaneurysm Rupture. Intern. Emerg. Med. (2017) 12 1319-1321. DOI: 10.1007/s11739-017-1641-9 10. Carr S.C., Pearce W.H., Vogelzang R.L., McCarthy W.J., Nemcek A.A., Yao J.S.T.. Current Management of Visceral Artery Aneurysms. Surgery (1996) 120 627-634. DOI: 10.1016/S0039-6060(96)80009-2 11. Fukami Y., Saito T., Arikawa T., Osawa T., Komatsu S., Kaneko K., Suzuki K., Takahashi E., Tsuzuki T., Sano T.. Inferior Pancreaticoduodenal Artery Pseudoaneurysm Rupture Caused by the Pancreatic Neuroendocrine Tumor. Surgery (2020) 168 e1-e2. DOI: 10.1016/j.surg.2020.02.019 12. Pitton M.B., Dappa E., Jungmann F., Kloeckner R., Schotten S., Wirth G.M., Mittler J., Lang H., Mildenberger P., Kreitner K.-F.. Visceral Artery Aneurysms: Incidence, Management, and Outcome Analysis in a Tertiary Care Center over One Decade. Eur. Radiol. (2015) 25 2004-2014. DOI: 10.1007/s00330-015-3599-1 13. Mitrovic M., Tadic B., Jankovic A., Rankovic I., Kovac J.D.. Fatal Gastrointestinal Bleeding Associated with Acute Pancreatitis as a Complication of COVID-19: A Case Report. J. Int. Med. Res. (2022) 50 3000605221098179. DOI: 10.1177/03000605221098179 14. Mitrovic M., Dugalic V., Kovac J., Tadic B., Milosevic S., Lukic B., Lekic N., Cvetic V.. Successful Embolization of Posterior Inferior Pancreaticoduodenal Artery Pseudoaneurysm on the Grounds of Chronic Pancreatitis—Case Report and Literature Review. Medicina (2020) 56. DOI: 10.3390/medicina56110617 15. Mitrovic Jovanovic M., Tadic B., Jankovic A., Stosic K., Lukic B., Cvetic V., Knezevic D.. Endovascular Treatment of a Pseudoaneurysm of the Posterior Inferior Pancreaticoduodenal Artery as a Complication of Chronic Pancreatitis: A Case Report. J. Int. Med. Res. (2022) 50 03000605221083441. DOI: 10.1177/03000605221083441 16. Jang J.Y., Kim J.H., Kim T.U., Ryu H., Lee T.B., Ryu J.H., Jeon U.B.. Percutaneous N-Butyl-Cyanoacrylate Embolization for Treating Ruptured Pancreaticoduodenal Aneurysm: A Case Report. Medicina (2022) 58. DOI: 10.3390/medicina58101320 17. Gurala D., Polavarapu A.D., Idiculla P.S., Daoud M., Gumaste V.. Pancreatic Pseudoaneurysm from a Gastroduodenal Artery. Case Rep. Gastroenterol. (2019) 13 450-455. DOI: 10.1159/000503895 18. Ushijima T., Izumo A., Matsumoto T., Taniguchi K., Uchida T.. Pancreaticoduodenal Artery Pseudoaneurysm Caused by Segmental Arterial Mediolysis: A Case Report of Surgical Treatment. Ann. Vasc. Dis. (2013) 6 198-201. DOI: 10.3400/avd.cr.12.00094 19. George J., Besant S., Cleveland T., Al-Mukhtar A.. Pseudoaneurysm of the Inferior Pancreaticoduodenal Artery Due to Cholecystitis. Y Rep. (2019) 2. DOI: 10.3390/reports2010011 20. Kasturi K.S., Mummadi R.R., Sugumar A.. Inferior PancreaticoDuodenal Artery Pseudoaneurysm: A Rare Cause of Massive GI Hemorrhage: 464. Off. J. Am. Coll. Gastroenterol. ACG (2007) 102 S286. DOI: 10.14309/00000434-200709002-00464 21. Ren Y., Wang Z., Zhu L., Kan X., Chen L., Liu Y., Song S., Guo X., Dong X., Zheng C.. Endovascular Repair of Visceral Artery Aneurysms and Pseudoaneurysms in 159 Patients: Twelve Years’ Experience of Clinical Technique. Abdom. Radiol. (2022) 47 443-451. DOI: 10.1007/s00261-021-03326-y 22. Xu Q.-D., Gu S.-G., Liang J.-H., Zheng S.-D., Lin Z.-H., Zhang P.-D., Yan J.. Inferior Pancreaticoduodenal Artery Pseudoaneurysm in a Patient with Calculous Cholecystitis: A Case Report. World J. Clin. Cases (2019) 7 2851-2856. DOI: 10.12998/wjcc.v7.i18.2851 23. Waguri N., Azumi R., Osaki A., Sato M., Aiba T., Furukawa K.. Ruptured Pancreaticoduodenal Pseudoaneurysm Causing Arterioportal Fistula: Combined Transarterial and Transportal Embolization. CVIR Endovasc. (2020) 3 72. DOI: 10.1186/s42155-020-00165-8 24. Suzuki K., Tachi Y., Ito S., Maruyama K., Mori Y., Komada T., Matsushima M., Ota T., Naganawa S.. Endovascular Management of Ruptured Pancreaticoduodenal Artery Aneurysms Associated with Celiac Axis Stenosis. Cardiovasc. Intervent. Radiol. (2008) 31 1082-1087. DOI: 10.1007/s00270-008-9343-3 25. Comby P.-O., Guillen K., Chevallier O., Lenfant M., Pellegrinelli J., Falvo N., Midulla M., Loffroy R.. Endovascular Use of Cyanoacrylate-Lipiodol Mixture for Peripheral Embolization: Properties, Techniques, Pitfalls, and Applications. J. Clin. Med. (2021) 10. DOI: 10.3390/jcm10194320 26. Zhang W., Fu Y.-F., Wei P.-L., Bei E., Li D.-C., Xu J.. Endovascular Repair of Celiac Artery Aneurysm with the Use of Stent Grafts. J. Vasc. Interv. Radiol. (2016) 27 514-518. DOI: 10.1016/j.jvir.2015.12.024 27. Coskun B., Rozanes I., Erkan M., Ates S., Akpek S.. Giant Celiac Arteri Aneurysm Treated with a Flow-Divertıng Multilayer Stent: Early Rupture as a Fatal Complication. J. Vasc. Interv. Radiol. (2017) 28 468-470. DOI: 10.1016/j.jvir.2016.11.036 28. Gorsi U., Agarwal V., Nair V., Kang M., Kalra N., Sreedhara B.C., Gupta R., Rana S.S., Dutta U., Sandhu M.S.. Endovascular and Percutaneous Transabdominal Embolisation of Pseudoaneurysms in Pancreatitis: An Experience from a Tertiary-Care Referral Centre. Clin. Radiol. (2021) 76 314.e17-314.e23. DOI: 10.1016/j.crad.2020.12.016 29. Yamamoto S., Hirota S., Maeda H., Achiwa S., Arai K., Kobayashi K., Nakao N.. Transcatheter Coil Embolization of Splenic Artery Aneurysm. Cardiovasc. Intervent. Radiol. (2008) 31 527-534. DOI: 10.1007/s00270-007-9237-9 30. Wojtaszek M., Lamparski K., Wnuk E., Ostrowski T., Maciąg R., Rix T., Maj E., Milczarek K., Korzeniowski K., Rowiński O.. Selective Occlusion of Splenic Artery Aneurysms with the Coil Packing Technique: The Impact of Packing Density on Aneurysm Reperfusion Correlated between Contrast-Enhanced MR Angiography and Digital Subtraction Angiography. Radiol. Med. (2019) 124 450-459. DOI: 10.1007/s11547-019-00993-2 31. Yoon T., Kwon T., Kwon H., Han Y., Cho Y.. Transcatheter Arterial Embolization of Splenic Artery Aneurysms: A Single-Center Experience. Vasc. Spec. Int. (2014) 30 120-124. DOI: 10.5758/vsi.2014.30.4.120 32. Kawasaki R., Miyamoto N., Oki H., Okada T., Yamaguchi M., Gomi T., Higashino T., Washio T., Maruta T., Sugimura K.. Flow-Preserved Coil Embolization Using a Side-Holed Indwelling Catheter for Common Hepatic Artery Pseudoaneurysm: Report of Three Cases. Surg. Today (2015) 45 772-776. DOI: 10.1007/s00595-014-0976-3 33. Erben Y., De Martino R.R., Bjarnason H., Duncan A.A., Kalra M., Oderich G.S., Bower T.C., Gloviczki P.. Operative Management of Hepatic Artery Aneurysms. J. Vasc. Surg. (2015) 62 610-615. DOI: 10.1016/j.jvs.2015.03.077 34. Kim S.K., Lee J., Duncan J.R., Picus D.D., Darcy M.D., Sauk S.. Endovascular Treatment of Superior Mesenteric Artery Pseudoaneurysms Using Covered Stents in Six Patients. Am. J. Roentgenol. (2014) 203 432-438. DOI: 10.2214/AJR.13.11644 35. Ozaki T., Kimura M., Yoshimura N., Hori Y., Takano T., Kamura T., Yamamoto S., Sasai K.. Endovascular Treatment of Spontaneous Isolated Dissecting Aneurysm of the Superior Mesenteric Artery Using Stent-Assisted Coil Embolization. Cardiovasc. Intervent. Radiol. (2006) 29 435-437. DOI: 10.1007/s00270-005-0067-3 36. Kim S.H., Lee M.-S., Han H.Y., Lee H.M.. Endovascular Management of Ruptured Middle Colic Artery Aneurysm and Review of the Literature. Ann. Vasc. Surg. (2019) 59 310.e13-310.e16. DOI: 10.1097/SLA.0000000000002512 37. Gandini R., Abrignani S., Perrone O., Lauretti D.L., Merolla S., Scaggiante J., Vasili E., Floris R., Cioni R.. Retrograde Endovascular Stenting of Preocclusive Celiac Artery Stenosis with Loop Technique Associated with Pancreaticoduodenal Artery Aneurysm Embolization. J. Vasc. Interv. Radiol. (2017) 28 1607-1609. DOI: 10.1016/j.jvir.2017.06.035 38. Sadek M., Rockman C.B., Berland T.L., Maldonado T.S., Jacobowitz G.R., Adelman M.A., Mussa F.F.. Coil Embolization of a Gastroduodenal Artery Pseudoaneurysm Secondary to Cholangitis: Technical Aspects and Review of the Literature. Vasc. Endovasc. Surg. (2012) 46 550-554. DOI: 10.1177/1538574412457082 39. Sharma M., Somani P., Sunkara T., Prajapati R., Talele R.. Endoscopic Ultrasound-Guided Coil Embolization and Thrombin Injection of a Bleeding Gastroduodenal Artery Pseudoaneurysm. Endoscopy (2019) 51 E36-E37. DOI: 10.1055/a-0790-8134 40. Bains L., Kori R., Sharma R., Kaur D.. Replaced Right Hepatic Artery Pseudoaneurysm Managed with Coil Embolisation. BMJ Case Rep. CP (2019) 12 e227921. DOI: 10.1136/bcr-2018-227921 41. Nagarajan K., Sunilkumar D., Ramakrishnaiah V.P.N., Amuthabarathi M.. Left Gastric Pseudoaneurysm in a Case of Chronic Pancreatitis: A Case Report with Review of Literature. Vasc. Endovasc. Surg. (2021) 55 73-76. DOI: 10.1177/1538574420954309 42. Vargas H.A., Cousins C., Higgins J.N., See T.C.. Left Gastric Artery Aneurysm: Successful Embolization with Ethylene Vinyl Alcohol Copolymer (Onyx). Cardiovasc. Intervent. Radiol. (2008) 31 418-421. DOI: 10.1007/s00270-007-9119-1
--- title: A Comparative Study of Metabolic Syndrome Using NCEP—ATP III and IDF Criteria in Children and Its Relationship with Biochemical Indicators in Huatusco, Veracruz, Mexico authors: - Eduardo Rivadeneyra-Domínguez - Joel Jahaziel Díaz-Vallejo - Aurora Guadalupe Prado-Bobadilla - Juan Francisco Rodríguez-Landa journal: Children year: 2023 pmcid: PMC10047056 doi: 10.3390/children10030473 license: CC BY 4.0 --- # A Comparative Study of Metabolic Syndrome Using NCEP—ATP III and IDF Criteria in Children and Its Relationship with Biochemical Indicators in Huatusco, Veracruz, Mexico ## Abstract Metabolic syndrome includes a set of metabolic alterations associated with overweight and obesity. The criteria for its diagnosis are heterogeneous, and there have been few studies about prevalence in the pediatric population. The aim of this study was to describe how the estimated prevalence of metabolic syndrome varies by International Diabetes Federation (IDF) vs. National Cholesterol Education Program—Adult Treatment Panel III (NCEP—ATP) criteria. We conducted a cross-sectional study in which anthropometric information, triglyceride, cholesterol, glycemia, and insulin levels, among others, were collected. We compared the group potentially misclassified by IDF with the group classified without metabolic syndrome by NCEP—ATP with respect to weight status and biomarkers. Statistical analysis included linear regression, Mann–Whitney U test, Fisher´s exact test, and odds ratio calculation. The IDF criteria missed the association with obesity, although the undetected group differed significantly from the nonmetabolic syndrome group in terms of IBM and weight. The associated biomarkers were ultrasensitive C-reactive protein (Hs-CRP), alanine aminotransferase (ALT) enzyme, insulin, triglycerides, and high-density lipoprotein (HDL) cholesterol. Waist circumference was the parameter with the strongest association for presenting metabolic syndrome, with an odds ratio of 18.33. The results of this study showed the estimated prevalence of MS varies by criteria, due to cutoff points, and how the high prevalence of MS strongly associated with obesity. ## 1. Introduction Metabolic syndrome (MS) involves a set of metabolic disturbances including insulin resistance, abdominal obesity, increased fasting glucose, hypertension, hypertriglyceridemia, hypercholesterolemia, proinflammatory state, and low high-density lipoprotein (HDL) concentration, among others [1]. The global prevalence of MS in the adult population is estimated to be 20–$40\%$ [2]. It is one of the major risk factors for the development of type 2 diabetes mellitus and cardiovascular disease, and it is associated with the global epidemic of overweight and obesity [3]. As in adults, the prevalence of overweight and obesity in children and adolescents has been increasing. According to the World Health Organization (WHO), it increased from $8\%$ in 1975 to $18\%$ in 2016 [4,5]. In Mexico, according to the results of the 2021 National Health and Nutrition Survey, the combined prevalence of overweight and obesity was $37.4\%$ in children aged 5 to 11 years and $42.9\%$ in adolescents aged 12 to 19 years [6], making it one of the main public health problems. The pathogenesis of MS is complex, but insulin resistance and obesity increase the risk of developing it, so the presence of obesity and overweight in children and adolescents is a risk factor for this condition. Furthermore, overweight in childhood and adolescence is associated with an increased risk and earlier onset of type 2 diabetes [7]. The diagnosis of MS has generated controversy due to a lack of consensus on its definition for children and adolescents worldwide, which is why the International Diabetes Federation (IDF) adapted a classification for this population group [8]. Another widely accepted classification is that proposed by the National Cholesterol Education Program—Adult Treatment Panel III (NCEP—ATP) [9,10], see Table 1. However, there are no standardized criteria worldwide. This has led several authors to modify the criteria and cutoff points for the diagnosis of MS. In Mexico, there have been few studies of MS in children and adolescents, which is why information is limited and it has been difficult to establish criteria between age groups and BMI for diagnosing obesity and MS. The aim of this study was to describe how the estimated prevalence of MS varies by IDF vs. NCEP—ATP criteria and how the relationship between MS and obesity and chronic disease risk factors varies by IDF vs. NCEP—ATP criteria in a sample of school-aged children and adolescents from a community in Veracruz, Mexico. ## 2.1. Study Design and Participants A cross-sectional study was conducted in Mexican children and adolescents aged 8–15 years with a random sample of students from two public educational institutions in the town of Huatusco, Veracruz, Mexico. The schools are urban and low-income, and they were selected for convenience of access and location. The school enrollment of third to sixth graders in each school was 92 and 87 children, respectively. For the selection of participants, a sample size calculation was performed with an expected proportion of $10\%$ SM according to previous studies [11]. Individuals were randomly selected according to the list of students from both schools, with the same probability of selection, receiving a direct invitation to participate and the subsequent voluntary acceptance of the child´s guardian. From the sample, a total of 91 children were eligible, two children were excluded because of a previous medical diagnosis of diabetes mellitus and heart disease; 17 children refused to participate who did not differ from the characteristics of the children included in the study. In total, 72 participants were included. Interviews and measurements were conducted in the months of September 2020 to March 2021. ## 2.2.1. Anthropometric Characteristics A brief survey was conducted on sociodemographic characteristics and family history for data collection. The survey was designed by a pediatric physician and a nutritionist. Anthropometric measurements were taken in the presence of the parents and/or child´s guardian, and American Union criteria were used. Weight was measured using a scale and SECA® stadiometer while barefoot and wearing light clothing, and abdominal circumference was measured with an inextensible tape measure. Waist circumference was measured at the lower edge of the last rib and the upper edge of the iliac crest; the standard error for weight was less than 0.02 kg, and that for circumference was less than 0.01 cm, as indicated by the American Diabetes Association. Body mass index (BMI) was calculated for nutritional status using the Z-score with its respective cutoff points according to the WHO [12]. Trained personnel measured blood pressure with a sphygmomanometer (Welch Allyn, Ciudad de Mexico, Mexico) and stethoscope (3MTMLittmann, Ciudad de Mexico, Mexico). ## 2.2.2. Clinical Samples and Biomarker Measurement Sample extraction and analysis were carried out at the Huatusco Clinical Specialties Laboratory (Av. 2 no. 616 between street 4 and street 6, CP. 94100 Huatusco de Chicuellar, Veracruz, México), the same place where, owing to the COVID-19 contingency, anthropometric measurements and phlebotomies were performed under sanitary conditions in the presence of parents and/or child´s guardian. Blood samples were obtained via venous puncture (cephalic, mid-ulnar, and basilic vein) with a prior fasting period of 10–12 h; the volume collected per child was approximately 8–10 mL. For the determination of blood chemistry and liver function, a Wiener Lab CM250 automatic analyzer (version 4.2 MT-Promedt Consulting GmbHaltenhofstr. 80D-66386 St. Ingbert/Germany, colorimetric method) was used. For alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT) determinations, enzymatic methods were used. For the analysis of insulin and ultrasensitive C-reactive protein (hs-CRP), we used the IMMULITE 1000 (Siemens Healthcare Diagnostics Inc., Flanders, NJ. 078369657, USA) automated analyzer. ## Definitions Insulin Resistance: insulin resistance was assessed with the HOMA-IR (Homeostatic Assessment Model) method using the following formula: fasting insulin (μU/mL) × basal glucose (mg/dL)/405, with a cutoff point at values equal to or greater than 2.86, according to previous studies in the Mexican population [13]. For Metabolic Syndrome: the criteria adapted for children and adolescents from NCEP—ATP III (National Cholesterol Education Program—Adult Treatment Panel III) [9,10] and IDF (International Diabetes Federation) [8] were used. MS was considered when three or more conditions were present. These conditions are shown in Table 1. ## 2.3. Statistical Analysis The prevalence of MS was estimated using the NCEP—ATP and IDF criteria. The agreement between classifications was determined. We compared the group potentially misclassified by IDF with the group classified as without MS by NCEP—ATP with respect to weight status and biomarkers. Tests for the normality of variables were performed, and descriptive statistics were used for quantitative variables according to distribution. Mann–Whitney U test for median comparisons was used. Fisher´s exact test was used to compare categorical variables. We used linear regression to analyze differences between groups. Mean and standard deviation were calculated. Odds ratios (ORs) with $95\%$ confidence intervals ($95\%$ CIs) were estimated. The significance level was set as $p \leq 0.05.$ The Statistical Package for Social Sciences (SPSS-IBM) version 25 was used. ## 3. Results A total of 72 students participated in the study, with a median age of 10 years. They were $54.2\%$ male ($$n = 39$$) and $45.8\%$ female ($$n = 33$$). Table 2 shows the characteristics of the study population; differences were found in the anthropometric variables of weight, BMI, and age, according to criteria for MS. In the analysis by age group and sex, significant differences were found in the 8–9 years age group between girls and boys for weight (27.16 and 40.60 kg, respectively; $$p \leq 0.001$$), waist circumference (59 and 78 cm, respectively; $$p \leq 0.001$$), and BMI, which was lower in girls. Regarding biomarkers, there was a significant difference in basal glucose, which was higher in girls (94 mg/dL) than in boys (87.5 mg/dL). For Hs-CRP, the concentration was higher in boys (0.91 mg/dL) than in girls (0.15 mg/dL). In the 10–15 years age group, no significant differences were found. Regarding genetic factors, $48.6\%$ of the children had no family history of diabetes and $51.4\%$ mentioned having a history of type 2 diabetes mellitus in at least one member of their family, which was corroborated by their child´s guardian. On the other hand, $19.12\%$ of the children reported having a family history of hypertension, $79.5\%$ reported having no family history of hypertension, and $1.38\%$ did not know this information. A family history of obesity was reported at a frequency of $83.34\%$, and no associations were found with respect to MS. The combined prevalence of overweight and obesity was $54.2\%$, with a significantly higher rate of obesity in boys than in girls, as shown in Table 3. The prevalence of MS according to ATP III criteria was $22.2\%$ (12.4–$32.1\%$, $95\%$ CI), and no sex differences were found, with $23.1\%$ of males (9.2–$36.9\%$, $95\%$ CI) and $21.2\%$ of females (6.5–$35.9\%$). Using the IDF criteria, the prevalence was found to be $13.9\%$ (5.7–$22.1\%$, $95\%$ CI), with no sex difference. Of the participants, $23.6\%$ had a waist circumference >90th percentile (P90), $26.4\%$ had blood pressure > P90, $47.2\%$ had triglycerides > 110 mg/dL, $51.4\%$ had HDL < 40 mg/dL, $9.7\%$ had basal blood glucose > 110 mg/dL, and $37.5\%$ had insulin resistance according to HOMA. Table 4 presents the nutritional status according to BMI and the components of MS corresponding to ATP III and IDF criteria, showing significant differences in the prevalence of MS, which was higher in the obese group ($44.4\%$) compared with the overweight ($16.7\%$) and normal weight ($6.1\%$) groups in the case of ATP III criteria; using the IDF criteria, this difference was not significant. A significant association was found between obesity and MS by the NCEP—ATP III criteria (OR: 8.2, $95\%$CI: 2.3–29.4), which was not detected using the IDF criteria (OR: 2.9, $95\%$CI: 0.7–11.5). No statistically significant differences were found when analyzed by sex. Table 5 shows the significant differences found between the groups with and without metabolic syndrome with respect to biomarkers; uric acid, ALT, insulin, and Hs-CRP levels were higher in the MS group. Table 6 shows the estimated ORs; with respect to biomarkers, an association was found between Hs-CRP above the normal value and metabolic syndrome, with OR = 3.66 (1.11–12.02, CI $95\%$) ($$p \leq 0.027$$). ## 4. Discussion The prevalence of MS has varied in different studies carried out in the pediatric population due to the criteria and cutoff points used by the authors, which makes comparisons between studies difficult. In our study, the prevalence varied when using the NCEP—ATP III and IDF criteria, as we obtained a lower prevalence with the latter. Previous studies in Mexico have reported a higher prevalence of MS. For example, a study conducted in a primary care medical unit in a population aged 6 to 16 years, using the NCEP—ATP III criteria, estimated a prevalence of $33\%$ [14]. A higher prevalence of $40\%$ was reported in school children aged 6 to 12 years, and a prevalence of $54.6\%$ was reported in children with obesity [15]; in the present study, the prevalence in the obese population was $44.4\%$. Regarding nutritional status, a significant difference was observed in the prevalence of MS, being higher in children with obesity when using the ATP III criteria. With the IDF criteria, despite also estimating a higher prevalence in children with obesity, it was not statistically significant as compared with the other nutritional statuses. The IDF definition failed to identify the remaining $8\%$ that were identified with the ATP III criteria. This is because the IDF criteria increase their cutoff points for blood pressure and triglycerides that normally apply to adults, so that children with high values but not above the cutoff point, are discarded and inappropriately classified, possibly underestimating prevalence. It is important to mention that in the present study, insulin resistance was found in $37\%$ of the obese group, which was lower than the $65\%$ previously reported in Mexican obese children [16]. The authors of that study also found a high percentage of insulin resistance in children without a diagnosis of MS. In our study, we found that $33.9\%$ of the children, despite not yet meeting the ATP III criteria for MS classification, were in a possible predisposing state for the development of the disease. In addition, obesity is defined as abnormal or excessive fat accumulation that increases the risk of developing a secondary disease. Adipose tissue was previously considered as a static tissue (reservoir for energy). Studies have referred to adipose tissue as a dynamic tissue (metabolically active organ) [17,18,19]. The morphophysiological change of adipose tissue during obesity induces a chronic low-grade inflammatory state, also referred to as parainflammation (intermediate state between basal and inflammatory) or metainflammation (metabolically triggered inflammation) [20,21,22]. Many studies report that, during this inflammatory process, there is excessive segregation of inflammatory factors known as adipokines, bioactive molecules involved in the etiology of inflammation and insulin resistance associated with obesity [23], segregated by adipocytes that include TNF, IL-6, IFN-, plasminogen activator inhibitor (PAI-1), monocyte chemoattractant protein-1 (MCP1), IL-1, IL-8, IL-10, IL-15, leukemia inhibitory factor (LIF), hepatocyte growth factor (HGF), serum apolipoprotein amyloid A3 serum (SAA3), macrophage migration inhibitory factor (MIF), potent inflammatory modulators (such as leptin, adiponectin, resistin), and high sensitive C-reactive protein (hs-CRP), and these maintain both negative and positive effects, such as the maintenance of oxidative stress, changes in autophagy patterns, and tissue necrosis, principally [24,25]. On the other hand, highly sensitive C-reactive protein is a reactant, that is, a plasmatic protein that undergoes alterations in phases of inflammation, which is synthesized by the liver and deposited in sites with inflammatory processes. Recent studies have shown that the intima of arteries with atherogenesis as cardiovascular disease (CVD) even before any clinical manifestation, accompanied by tumor necrosis factor and interleukins [26,27]. hs-CRP, which is used as a biomarker for the diagnosis of MS, was found to be associated with an OR of 3.66 (1.11–12.02, $95\%$ CI). In this regard, several studies have reported the usefulness of this biomarker as an indicator of an inflammatory process related to the development of MS and diabetes mellitus [28,29]. Regarding serum transaminases, only ALT with values above the biological reference ranges showed an association with MS, with an OR of 3.57 (1.07–11.89 CI $95\%$). Other studies have suggested that elevated transaminase concentrations are associated with obesity and may be early markers of MS [30,31]. Therefore, it is important to incorporate components that have been associated with the risk of developing MS such as insulin resistance, hs-CRP, and transaminases, and to assess the possibility of improving the prognosis of the condition, during treatment. The mean serum levels of other biomarkers, such as creatinine (0.69 mg/dL), were found to be within the biological reference ranges. In the case of uric acid, levels below the cutoff point were found without any association with MS, unlike in other studies where hyperuricemia has been observed as an alteration in MS [32]. It is important to highlight some limitations of the present study. First, the incorporation of variables related to healthy lifestyles is lacking, which could be used to perform more in-depth analyses of the associations. Second, the small sample size, which was smaller than that reported in other studies; data collection was conducted during the COVID-19 pandemic and some householders did not accept the invitation for their children to participate in the present study. The small sample size may have influenced the fact that some biomarkers were not found to be related to MS, as previously reported. Furthermore, owing to the cross-sectional design of the study, causality could not be inferred. A further limitation was the lack of biomarkers that evidence low-grade systemic inflammation such as tumor necrosis factor alpha (TNF-α) and interleukins (IL-1,6,17), which are frequently used in clinical research, but are not routinely included in the clinical laboratory due to their high cost. Although our population and age cutoff may have some limitations, it gives rise to consideration in future studies for including a population with a larger number of individuals and a wider age range according to the WHO classification of infants and adolescents, with the aim of expanding and looking for other associated factors. An advantage of the study was to work with data from school children, from which it was possible to estimate the prevalence of MS in different nutritional states, where children could be considered healthy, unlike research conducted in a hospital environment, where there could be a bias and overestimation of prevalence because children who are brought by their child´s guardian may already have some pathology and even in advanced stages of the disease due to the lack of a culture of prevention. The results of the present study provide insight into the most relevant components associated with MS and which biomarkers could be used for early diagnosis. We believe that the estimates obtained in this study warrant further work to determine how to infer a child’s risk of developing MS from weight status and take action to prevent metabolic syndrome. The prevalence of metabolic syndrome varies across studies [11], depending on the criteria used to define MS. Failure to detect children with preclinical risk factors misses intervention opportunity in the window of prevention. To inform choice of MS criteria to reduce chronic disease risks for children in Mexico, this paper explored the difference between MS classification based on IDF and NCEP—ATP III. ## 5. Conclusions The results of this study showed the estimated prevalence of MS varies by criteria, owing to cutoff points, and how the high prevalence of MS strongly associated with obesity. Children in this study with obesity and overweight have a high percentage risk of developing cardiovascular disease and type 2 diabetes mellitus in adulthood. Therefore, the use of biomarkers in the pediatric population for the timely diagnosis of MS may be a useful tool when adopting strict measures to prevent the development of other conditions. Additional studies need to be conducted to identify other possible factors. It is also necessary to design school-based preventive strategies through early detection, promotion of healthy lifestyles, and physical activity. MS is a condition caused by multiple factors, most of which are preventable, such as obesity. Therefore, it is necessary to implement comprehensive public policies to prevent obesity in children and adolescents at home, school, and community. ## References 1. Wang H.H., Lee D.K., Liu M., Portincasa P., Wang D.Q.-H.. Novel Insights into the Pathogenesis and Management of the Metabolic Syndrome. Pediatr. Gastroenterol. Hepatol. Nutr. (2020.0) 23 189. DOI: 10.5223/pghn.2020.23.3.189 2. Rojas-Martínez R., Aguilar-Salinas C.A., Romero-Martínez M., Castro-Porras L., Gómez-Velasco D., Mehta R.. Trends in the Prevalence of Metabolic Syndrome and Its Components in Mexican Adults, 2006–2018. Salud Publica Mex. (2021.0) 63 713-724. DOI: 10.21149/12835 3. Després J.-P., Lemieux I., Bergeron J., Pibarot P., Mathieu P., Larose E., Rodés-Cabau J., Bertrand O.F., Poirier P.. Abdominal Obesity and the Metabolic Syndrome: Contribution to Global Cardiometabolic Risk. Arterioscler. Thromb. Vasc. Biol. (2008.0) 28 1039-1049. DOI: 10.1161/ATVBAHA.107.159228 4. Obesity and Overweight 5. Abarca-Gómez L., Abdeen Z.A., Hamid Z.A., Abu-Rmeileh N.M., Acosta-Cazares B., Acuin C., Adams R.J., Aekplakorn W., Afsana K., Aguilar-Salinas C.A.. Worldwide Trends in Body-Mass Index, Underweight, Overweight, and Obesity from 1975 to 2016: A Pooled Analysis of 2416 Population-Based Measurement Studies in 128·9 Million Children, Adolescents, and Adults. Lancet (2017.0) 390 2627-2642. DOI: 10.1016/S0140-6736(17)32129-3 6. Shamah-Levy T., Romero-Martínez M., Barrientos-Gutiérrez T., Cuevas-Nasu L., Bautista-Arredondo S., Colchero M., Gaona-Pineda E., Lazcano-Ponce E., Martínez-Barnetche J., Alpuche-Arana C.. Encuesta Nacional de Salud y Nutrición 2021 Sobre Covid-19. Resultados Nacionales (2022.0) 7. Olaiya M.T., Knowler W.C., Sinha M., Kobes S., Nelson R.G., Baier L.J., Muller Y.L., Hanson R.L.. Weight Tracking in Childhood and Adolescence and Type 2 Diabetes Risk. Diabetologia (2020.0) 63 1753-1763. DOI: 10.1007/s00125-020-05165-w 8. Zimmet P., Alberti K.G.M., Kaufman F., Tajima N., Silink M., Arslanian S., Wong G., Bennett P., Shaw J., Caprio S.. The Metabolic Syndrome in Children and Adolescents-an IDF Consensus Report. Pediatr. Diabetes (2007.0) 8 299-306. DOI: 10.1111/j.1399-5448.2007.00271.x 9. Cook S., Weitzman M., Auinger P., Nguyen M., Dietz W.H.. Prevalence of a Metabolic Syndrome Phenotype in Adolescents. Arch. Pediatr. Adolesc. Med. (2003.0) 157 821-827. DOI: 10.1001/archpedi.157.8.821 10. Fernández J.R., Redden D.T., Pietrobelli A., Allison D.B.. Waist Circumference Percentiles in Nationally Representative Samples of African-American, European-American, and Mexican-American Children and Adolescents. J. Pediatr. (2004.0) 145 439-444. DOI: 10.1016/j.jpeds.2004.06.044 11. Reisinger C., Nkeh-Chungag B.N., Fredriksen P.M., Goswami N.. The Prevalence of Pediatric Metabolic Syndrome—A Critical Look on the Discrepancies between Definitions and Its Clinical Importance. Int. J. Obes. (2021.0) 45 12-24. DOI: 10.1038/s41366-020-00713-1 12. De Onis M.. Development of a WHO Growth Reference for School-Aged Children and Adolescents. Bull. World Health Organ. (2007.0) 85 660-667. DOI: 10.2471/BLT.07.043497 13. Piña-Aguero M.I., Zaldivar-Delgado A., Salas-Fernández A., Martínez-Basila A., Bernabe-Garcia M., Maldonado-Hernández J.. Optimal Cut-Off Points of Fasting and Post-Glucose Stimulus Surrogates of Insulin Resistance as Predictors of Metabolic Syndrome in Adolescents According to Several Definitions. J. Clin. Res. Pediatr. Endocrinol. (2018.0) 10 139-146. DOI: 10.4274/jcrpe.4873 14. Bautista-Samperio L., Saldaña C., Hernández-Pérez J.. Síndrome Metabólico en Población de 6 a 16 Años de Edad en Primer Nivel de Atención. Rev Fac Med. UNAM (2010.0) 53 3-8 15. Ávila-Curiel A., Galindo-Gómez C., Juárez-Martínez L., Osorio-Victoria M.L.. Metabolic Syndrome in Children Aged 6 to 12 Years with Obesity in Public Schools of Seven Municipalities in the State of Mexico. Salud Publica Mex. (2018.0) 60 395. DOI: 10.21149/8470 16. Ortega-Cortés R., Hurtado-López E., López-Beltrán A., Trujillo-Trujillo X., Tlacuilo-Parra J., Colunga-Rodríguez C.. Caracterización de Niños Obesos con y sin Diagnóstico de Síndrome Metabólico en un Hospital Pediátrico. Rev. Med. Inst. Mex. Seguro. Soc. (2014.0) 52 548-556 17. Schwarzenberg S.J., Sinaiko A.R.. Obesity and inflammation in children. Paediatr. Respir. Rev. (2006.0) 7 239-246. DOI: 10.1016/j.prrv.2006.08.002 18. Perez Mayorga M.. El Adipocito Como Órgano Endocrino: Implicaciones Fisiopatológicas y Terapéuticas. Rev. Med. (2007.0) 152 225-242 19. Stolarczyk E.. Adipose tissue inflammation in obesity: A metabolic or immune response?. Curr. Opin. Pharmacol. (2017.0) 37 35-40. DOI: 10.1016/j.coph.2017.08.006 20. Monteiro R., Azevedo I.. Chronic inflammation in obesity and the metabolic syndrome. Mediat. Inflamm. (2010.0) 2010 289645. DOI: 10.1155/2010/289645 21. Krüger K., Mooren F.C., Eder K., Ringseis R.. Immune and Inflammatory Signaling Pathways in Exercise and Obesity. Am. J. Lifestyle Med. (2016.0) 10 268-279. DOI: 10.1177/1559827614552986 22. Castro A.M., Macedo-de la Concha L.E., Pantoja-Meléndez C.A.. Low-grade inflammation and its relation to obesity and chronic degenerative diseases. Rev. Médica Hosp. Gen. México (2017.0) 80 101-105. DOI: 10.1016/j.hgmx.2016.06.011 23. Pereira S.S., Alvarez-Leite J.I.. Low-Grade Inflammation, Obesity, and Diabetes. Curr. Obes. Rep. (2014.0) 3 422-431. DOI: 10.1007/s13679-014-0124-9 24. Domínguez-Amorocho O., Patiño-Cuervo D.. Proteína C reactiva ultrasensible (PCR-us) como marcador de riesgo de enfermedad cardiovascular. Med. Lab. (2008.0) 14 457-478 25. Sáinz N., Barrenetxe J., Moreno-Aliaga M.J., Martínez J.A.. Leptin resistance and diet-induced obesity: Central and peripheral actions of leptin. Metabolism (2015.0) 64 35-46. DOI: 10.1016/j.metabol.2014.10.015 26. Iantorno M., Campia U., Di Daniele N., Nistico S., Forleo G., Cardillo C., Tesauro M.. Obesity, inflammation and endotelial dysfunction. J. Biol. Regul. Homeost. Agents (2014.0) 282 169-176 27. Raskin Erusan R., Nalini D., Manohar G., Malathi R.. Correlation between Obesity and Inflammation in Cardiovascular Diseases—Evaluation of Leptin and Inflammatory Cytokines. Open J. Endocr. Metab. Dis. (2012.0) 2 7-15. DOI: 10.4236/ojemd.2012.22002 28. Jeong H., Baek S.-Y., Kim S.W., Park E.-J., Lee J., Kim H., Jeon C.H.. C Reactive Protein Level as a Marker for Dyslipidaemia, Diabetes and Metabolic Syndrome: Results from the Korea National Health and Nutrition Examination Survey. BMJ Open (2019.0) 9 e029861. DOI: 10.1136/bmjopen-2019-029861 29. Vu J.D., Vu J.B., Pio J.R., Malik S., Franklin S.S., Chen R.S., Wong N.D.. Impact of C-Reactive Protein on the Likelihood of Peripheral Arterial Disease in United States Adults With the Metabolic Syndrome, Diabetes Mellitus, and Preexisting Cardiovascular Disease. Am. J. Cardiol. (2005.0) 96 655-658. DOI: 10.1016/j.amjcard.2005.04.038 30. Valle-Martos R., Valle M., Martos R., Cañete R., Jiménez-Reina L., Cañete M.D.. Liver Enzymes Correlate With Metabolic Syndrome, Inflammation, and Endothelial Dysfunction in Prepubertal Children With Obesity. Front. Pediatr. (2021.0) 9 629346. DOI: 10.3389/fped.2021.629346 31. Sookoian S.. Alanine and Aspartate Aminotransferase and Glutamine-Cycling Pathway: Their Roles in Pathogenesis of Metabolic Syndrome. World J. Gastroenterol. (2012.0) 18 3775-3781. DOI: 10.3748/wjg.v18.i29.3775 32. Castillo-Durán C., Sepúlveda A.C., Espinoza G.A., Rebollo G.M.J., Le Roy O.C.. Hiperuricemia y Componentes Del Síndrome Metabólico En Niños y Adolescentes Obesos. Rev. Chil. Pediatría (2016.0) 87 18-23. DOI: 10.1016/j.rchipe.2015.07.022
--- title: ZnT8 Loss of Function Mutation Increases Resistance of Human Embryonic Stem Cell-Derived Beta Cells to Apoptosis in Low Zinc Condition authors: - Lina Sui - Qian Du - Anthony Romer - Qi Su - Pauline L. Chabosseau - Yurong Xin - Jinrang Kim - Sandra Kleiner - Guy A. Rutter - Dieter Egli journal: Cells year: 2023 pmcid: PMC10047077 doi: 10.3390/cells12060903 license: CC BY 4.0 --- # ZnT8 Loss of Function Mutation Increases Resistance of Human Embryonic Stem Cell-Derived Beta Cells to Apoptosis in Low Zinc Condition ## Abstract The rare SLC30A8 mutation encoding a truncating p.Arg138* variant (R138X) in zinc transporter 8 (ZnT8) is associated with a $65\%$ reduced risk for type 2 diabetes. To determine whether ZnT8 is required for beta cell development and function, we derived human pluripotent stem cells carrying the R138X mutation and differentiated them into insulin-producing cells. We found that human pluripotent stem cells with homozygous or heterozygous R138X mutation and the null (KO) mutation have normal efficiency of differentiation towards insulin-producing cells, but these cells show diffuse granules that lack crystalline zinc-containing insulin granules. Insulin secretion is not compromised in vitro by KO or R138X mutations in human embryonic stem cell-derived beta cells (sc-beta cells). Likewise, the ability of sc-beta cells to secrete insulin and maintain glucose homeostasis after transplantation into mice was comparable across different genotypes. Interestingly, sc-beta cells with the SLC30A8 KO mutation showed increased cytoplasmic zinc, and cells with either KO or R138X mutation were resistant to apoptosis when extracellular zinc was limiting. These findings are consistent with a protective role of zinc in cell death and with the protective role of zinc in T2D. ## 1. Introduction Zinc transporter family (ZnTs) members are zinc exporters located on cell and organelle membranes and regulate the flow of zinc from the cytosol to granules and extracellularly. Zinc transporter 8 (ZnT8), encoded by SLC30A8, functions as a channel on insulin granules for transporting zinc ion into the granules to enable formation of stable insulin/zinc hexamers [1,2,3]. ZnT8 has been considered as an appealing target for preserving beta cell function in T2D because its loss of function mutations, including a pArg138* truncating variant, are associated with a more than $50\%$ lower risk of developing T2D [4,5]. Several efforts have been made to study the mechanisms of the protective effect. In rodents, deletion of Slc30A8 (SLC30A8) globally or specifically in beta cells reduces intracellular zinc content, measured with Zinquin [6], dithizone [7], or with the recombinant cytosol-targeted probe, eCALWY4 [8,9], but does not affect cytosolic free Zn2+ measured by FluoZin-3, or insulin content [7,10,11,12,13]. These mouse models display variable alterations in glucose-stimulated insulin secretion (GSIS) and glucose metabolism; both increases and decreases in insulin secretion have been reported. A more recent mouse model carrying the human SLC30A8 p.Arg138* allele shows enhanced insulin secretion under a high fat but not normal chow diet [14]. In recent studies by Dwivedi et al., heterozygous carriers of the protective p.Arg138* allele show greater insulin secretion during the oral glucose tolerance test (OGTT), but glucose excursions during a test meal do not differ between carriers of variant and wild type alleles [5]. Further studies using the human beta cell model EndoC-βH1, an immortalized beta cell line [15], by siRNA mediated knockdown of SLC30A8 reveal reduced overall zinc content in beta cells as measured using the zinc-specific fluorescent dye Zinpyr-1, whereas no significant effect on GSIS was observed, though basal insulin secretion was increased [5]. Thus, studies in rodent models and immortalized human beta cell lines have revealed a range of phenotypes, and the consequences of ZnT8 deficiency for beta cell function and the mechanism of a protective effect of ZnT8 loss of function remain a matter of debate. Zinc is an important trace element involved in numerous processes, including the regulation of apoptosis [2,16]. Reduced serum zinc levels have been detected in patients with type 2 diabetes (T2D), especially in those with poor glycemic control [17,18]. Additionally, plasma zinc is significantly decreased in older adults compared to that of young adults [19]. Furthermore, in mice on a limiting zinc diet, degeneration and apoptosis in the pancreas islet have been observed [20]. Here, we used human embryonic stem cell-derived beta cells (sc-beta cells) as a model system to understand the molecular and cellular consequences of ZnT8 loss of function by introducing a null mutation (KO), and homo-or heterozygous versions of p.Arg138* (R138X) into the SLC30A8 locus. We demonstrate that ZnT8 deficient sc-beta cells display normal differentiation and function in vitro and in vivo, but reduced transport of zinc from the cytosol to the insulin granule, and experience lowered apoptosis under limiting zinc conditions. ## 2.1. Human Embryonic Stem Cells with SLC30A8 Mutations Exhibit Normal Differentiation Potential towards Insulin-Producing Cells To explore whether mutations in SLC30A8 locus affect the development of insulin-producing cells, we first used CRISPR/Cas9 to introduce R138X and KO mutation into the SLC30A8 locus of a human embryonic stem cell line MEL1, which carries an insulin-GFP reporter (Figure 1A). Two cell lines with KO mutation (KO23: c.65delA/p.Lys22fsX1 and KO2: c.63delC/p.Ala21AlafsX2) (Figure 1A and Figure S1A) and two homozygous cell lines for R138X (C101: c.265C > T/c.265C > T and C121: c.265C > T/c.265C > T) (Figure 1A and Figure S1B) were generated and verified by Sanger sequencing. All engineered cell lines displayed normal chromosome integrity determined by karyotyping (Figure S1C). Differentiation efficiency towards insulin-producing cells was evaluated on wild type control (WT) ($$n = 5$$ independent replicates), R138X ($$n = 3$$ independent replicates for C101; $$n = 5$$ independent replicates for C121), and KO mutants ($$n = 3$$ independent replicates for KO2; $$n = 5$$ independent replicates for KO23). Upon differentiation, ~$80\%$ PDX1 and NKX6.1 double-positive pancreatic progenitors were generated by day 12 from all three genotypes (Figure 1B). On day 27, islet-like clusters were formed, and they expressed high levels of insulin as indicated by GFP expression (Figure 1C). Correspondingly, all cell lines stained positive for C-peptide and co-expressed NKX6.1, while KO and R138X sc-beta cells did not express detectable ZnT8 (Figure 1D). No significant difference in the percentage of C-peptide-positive cells was detected (WT: 61.8 ± $5.6\%$; KO2: 61.0 ± $5.3\%$; KO23: 57.6 ± $4.3\%$; C101: 60.0 ± $4\%$; C121: 59.4 ± $5.1\%$), and comparable percentages of C-peptide and NKX6.1 double-positive cells were observed out of the total C-peptide-positive cells from all five cell lines (WT: 54.8 ± $5.2\%$; KO2: 61.0 ± $5.0\%$; KO23: 60.4 ± $6.9\%$; C101: 63.3 ± $1.2\%$; C121: 60.2 ± $9.6\%$) (Figure 1D–F and Figure S1D,E). Polyhormonal cells positive for C-peptide and glucagon were also detected in the derived islet clusters (WT: 25 ± $4.4\%$; KO23: 20 ± $5.3\%$; C121:19.7 ± $2.1\%$) (Figure 1D,G). MEL1 with an R138X heterozygous mutation (C89), as found in human subjects, differentiated into insulin-producing cells with comparable differentiation efficiency to WT, R138X, and KO mutants (Figure S1B,C,F). Thus, loss of function mutations at the SLC30A8 locus in either the homozygous or the heterozygous state do not impair beta cell differentiation from human pluripotent stem cells. ## 2.2. Sc-Beta Cells Carrying SLC30A8 Mutations Have Zinc-Depleted Secretory Granules ZnT8 is responsible for transporting zinc into insulin granules, thus forming stable insulin-Zn2+ hexamers [22]. To investigate the consequences of ZnT8 deficiency, sc-beta cell clusters derived from all genotypes were stained with dithizone to assess the presence of zinc in the insulin granules [23]. WT sc-beta cell clusters stained positive with dithizone, whereas no strongly dithizone-positive cells were observed in sc-beta cell clusters derived from R138X and KO mutants. This result indicates that they were depleted of zinc in insulin secretory granules (Figure 2A). Next, the ultra-structure of insulin granules was examined in sc-beta cells after 8 months of transplantation using electron microscopy (Figure 2B). In the grafted cells after isolation, nearly $80\%$ of the insulin granule in the WT sc-beta cells are crystalized with a dense core, whereas almost all insulin granules in R138X (99.69 ± $0.62\%$) and KO (97.97 ± $0.86\%$) displayed an enlarged and diffused granule core (Figure 2D). The degree of granule density, indicated by the grayscale value (black = 0, white = 250), was also analyzed. Granules that are denser and darker are presented with lower grayscale value. It was found that insulin granules in KO (113.25 ± $27.23\%$) and R138X (78.29 ± $33.80\%$) were both significantly less dense with higher grayscale value than that in WT (35.86 ± $30.85\%$) cells (Figure 2E). Interestingly, granules in the KO sc-beta cells are significantly more diffused than those of the R138X mutant sc-beta cells. The diffused granules were also observed in mouse islets segregating for the KO and R138X transgenic mice (mice generation described previously in [14]). In mice, insulin granules displayed no dark cores in the explanted mouse pancreas with R138X mutation under the normal chow diet and with KO mutation under the high fat diet. Whereas, in WT control mice, a dark and dense core was formed in each insulin granule under the normal chow diet. ( Figure 2C). ## 2.3. Transcriptome Analysis Shows Discrepant Expression of Genes Involved in Zinc Ion Homeostasis in SLC30A8 Mutants We next examined the effects of SLC30A8 mutants on gene expression in sc-beta cell clusters on day 27 of differentiation. Based on a profile of cell specific markers, 10 endocrine cell populations and 1 non-endocrine population using cells from all conditions were identified (Figure 3A and Figure S2A). All identified endocrine populations expressed a high level of CHGA and different level of insulin with the highest expression in identified sc-beta cells (Figure S2A). Note that among these populations, SC-beta. PP cells are characterized by expressing both SC-beta cell genes and SC-PP cell genes, and SC-beta.zinc cells express not only SC-beta cell genes but also express high levels of SLC39A5 and metallothionine genes. Other endocrine cells have been distinguished based on the lineage specific markers, but their co-expression with insulin indicated that they were polyhormonal (Figure S2B). Pancreatic beta cells and alpha cells are major cell types that express SLC30A8 [10,24]. First, the percentage of sc-beta and sc-alpha cells was quantified in WT, KO, R138X derived clusters (Figure 3B). We found that the introduced mutations in SLC30A8 did not alter the percentage of sc-beta cells and sc-alpha cells in the sc-islet like clusters (Figure 3C), indicating that SLC30A8 is not essential for the fate determination of pancreatic endocrine cells. Double hormone positive cells, which have been shown to become sc-alpha cells [25], are also not different between genotypes. Next, genes differentially expressed between mutants and WT were identified in sc-beta cells (Figure 3D). KO and R138X shared similar gene profiles, and both were distinct from the gene profile of the WT. As expected, the expression of SLC30A8 together with metallothionein genes involved in the maintenance of zinc ion homeostasis were consistently downregulated in R138X and KO mutants (Figure 3E). Metal responsive transcriptional factor 1 (MTF-1) is a zinc dependent transcription factor regulating the transcription of genes involved in zinc transport and chelation. It also regulates the transcription of metallothionein which depends on zinc ions [26]. The lower expression of MT genes in both KO and R138X mutant sc-beta cells indicates the intracellular disruption of zinc homeostasis (Figure 3E). In contrast, genes involved in insulin-containing vesicles and insulin secretion were upregulated in R138X and KO beta cells (Figure 3F). Taken together, these results indicated that the loss of function KO and R138X mutations cause relatively minor transcriptional changes overall compared to wild type cells. Those changes that are shared in both KO and R138X point to possible molecular mechanisms through which pancreatic endocrine function may be altered: control of insulin hormone secretion and zinc ion homeostasis, each of which we examine separately in the following chapters. ## 2.4. Insulin Secretion and Glucose Regulation of SLC30A8 Mutant Sc-Beta Cell Are Comparable to WT As we noted an enrichment in the expression of hormone secretion-related genes in both KO and R138X sc-beta cells (Figure S4), we next sought to address whether loss of function mutations in the SLC30A8 locus have an effect on insulin secretion and beta cell function. Sc-beta cell clusters derived from R138X mutant ($$n = 15$$ including 11 mice transplanted with clone C121 from 3 batches of independent differentiation and 4 mice with clone C101 from 1 batch of differentiation), KO mutant ($$n = 13$$ including 9 mice with KO23 from 3 batches of independent differentiation and 4 mice with KO2 from 1 batch of differentiation), and WT ($$n = 14$$ from 4 batches of independent differentiation) cell lines with equivalent number of ~2 million cells were transplanted into the leg muscle of immunodeficient mice. Human C-peptide secretion in the fed state was monitored to evaluate insulin production of grafted sc-beta cells after transplantation. Human C-peptide secretion increased over time in mice transplanted with cell clusters of WT and mutants (Figure 4A). We compared human C-peptide and insulin secretion at a 6-month time point after transplantation, and no significant differences among WT, KO, and R138X cell lines were noticed (Figure 4B,C). Interestingly, mouse C-peptide was suppressed when human C-peptide was secreted at a high level, even prior to the application of streptozotocin (STZ) (Figure 4D). This shows that the human stem cell derived grafts functionally take over blood glucose regulation from the endogenous mouse pancreas. To examine the function of sc-beta cells after 26 weeks of maturation in vivo, mice were treated with streptozotocin to eliminate endogenous mouse beta cells. Mouse beta cells were successfully eliminated, as confirmed by barely detectable levels of mouse C-peptide in blood (Figure S3A). Overall, 14 mice transplanted with 2 R138X cell lines, 11 mice transplanted with 2 KO cell lines, and 10 mice transplanted with a WT cell line could maintain glucose homeostasis after mouse beta cells were ablated (Figure S3B). We also transplanted 2 mice with an R138X heterozygous mutation (C89) cell line, and they successfully maintained normal glucose levels after STZ treatment while the non-graft mice experienced hyperglycemia after the same treatment (Figure 4E and Figure S3B). Glucose tolerance tests (ipGTT) were performed on the mice with normal levels of blood glucose after STZ treatment. Mice transplanted with sc-derived clusters of all four genotypes, including WT, KO, R138X homozygous and heterozygous SLC30A8 mutant cells, were able to normalize blood glucose (Figure S3C) after glucose injection with increased insulin secretion at comparable levels (Figure S3D). We noticed that both R138X and KO mutants tended to clear blood glucose more efficiently after the glucose challenge, although the differences compared to WT cell-bearing animals were not statistically different (Figure S3C). We then repeated the ipGTT assay by including a 15 min measurement of blood glucose. With the increased mice number in each group, we found that mice transplanted with KO cell lines exhibited reduced glucose levels at both 15 min and 30 min (Figure 4F). Mice with R138X transplants also showed lower 15 min glucose level than that of the WT, though this difference was not significant. In terms of overall glucose regulation, we analyzed the area under the curve (AUC) integrated from 0 min to 60 min and observed significantly lower glucose excursion and more tolerance in the KO compared to the WT (Figure 4G). Noting that, consistent with a modest increase of insulin secretion, we found an increase of insulin secretion in vitro when cells were incubated with 2 mM glucose solution for 1 h compared to WT control (Figure S3E). This difference was not due to variability in insulin content, which was comparable among KO, R138X, and WT (Figure S3F). To rule out the possibility that differences in insulin secretion in vivo may be caused by alterations in the number of insulin-producing cells for each cell line, we measured the size of the whole graft in vivo (Figure S3G) and quantified the insulin-GFP positive area in the graft after isolation using bioluminescence imaging (Figure S3H). We observed similar graft size within the region of interest (ROI) in the GFP-positive area from each mouse grafted with WT, KO, and R138X sc-beta cells (Figure S3I). Insulin expression was confirmed in the grafted cells after isolation, and the grafted cells formed islet-like structures containing monohormonal insulin and glucagon-positive cells (representative pictures for KO and R138X are shown in Figure 4H). This shows that the grafting of cells is not negatively affected by the loss of ZnT8 function. To determine if the modest differences in ipGTT or insulin secretion are functionally meaningful in regulating blood glucose levels, HbA1C levels were measured in mice transplanted with sc-beta cells derived from WT and mutants. Regardless of the genotypes of sc-beta cells, all transplanted mice could maintain normal HbA1C levels. As a result, no apparent improvement in blood glucose regulation in the KO was noticed (Figure 4I), and all these transplanted mice have significantly lower HbA1C level than mice without any graft. To determine if proinsulin processing is altered, we examined the proinsulin and insulin ratio in R138X, KO, and WT sc-beta cells before and after transplantation. Each line demonstrated a comparable proinsulin to insulin production ratio (Figure S3J and Figure 4J), indicating that proinsulin processing was comparable to WT in SLC30A8 mutants. We also calculated insulin/C-peptide and proinsulin/insulin content ratio in the isolated grafted cells and observed no difference in the ratio between WT, KO, and R138X sc-beta cells (Figure S3K,L). In summary, we conclude that insulin processing, secretion and sc-beta cell function are not compromised due to the mutation introduced in the SLC30A8 locus. ## 2.5. mTOR Activity Is not Influenced by SLC30A8 Expression We found that a panel of ribosome genes were significantly down in SLC30A8 mutants in the sc-beta population compared to WT in single cell RNA-seq analysis (Figure S4A). Ribosome biogenesis is associated with the mTOR signaling pathway [28]. Intracellular free zinc has been shown to correlate with the mTOR signaling pathway through phosphorylation of p70 S6 kinase [29]. To determine if reduced expression of ribosome genes in R138X and KO is due to the decreased mTOR activity, we isolated insulin-positive cells based on GFP expression with flow cytometry and evaluated the phosphorylation of mTOR, downstream effectors S6 kinase and S6 ribosome protein by Western blot. WT and mutant cells expressed similar levels of phosphorylated form of mTOR, S6 kinase, and S6 ribosome from the total form of individuals (Figure S4B). These data suggest that mTOR activity is not affected by the expression of SLC30A8, contrasting with the knockdown of ZnT8 in EndoC-betaH1 cells which increased mTOR activity [5]. ## 2.6. SLC30A8 Mutant Sc-Beta Cells Are Resistant to Apoptosis in Low Zinc Conditions Transcriptome analysis as well as the reduction of crystalline insulin granules establish a disruption of intracellular zinc homeostasis in mutant cells, consistent with the molecular function of SLC30A8. To further explore whether free cytosolic Zn2+ concentrations are altered in SLC30A8 mutants, FRET (fluorescence resonance energy transfer)-based cytoplasmic free Zn2+ content analysis [30] was performed on insulin GFP-positive cells derived from R138X homozygous, R138X heterozygous, KO, and WT cell lines. The maximum and minimal ratios were, respectively, obtained upon intracellular zinc chelation with Tetrakis-(2-pyridylmethyl) ethylenediamine (TPEN) and zinc saturation with ZnCl2 in the presence of the Zn2+ ionophore, pyrithione (Figure 5A). Cytosolic zinc levels in SLC30A8 mutant sc-beta cells were comparable to or higher than those in WT cells. Notably, free cytosolic zinc was significantly elevated in cells carrying the KO mutation compared to other cell lines (Figure 5B). The difference between KO mutant and R138X mutant may be due to the expression of a partially-active SLC30A8 protein which retains two transmembrane domains in R138X mutant [14] (Figure 1A). A functional difference between the two alleles was also suggested by the higher percentage of diffuse insulin granules detected in the KO than in the R138X (Figure 2E). Zinc is a regulator of programmed cell death required for cell survival, and its depletion induces cell death (reviewed in [31,32]). Normally, there are around $0.47\%$ of beta cells that undergo apoptosis in control non-diabetic human islets [33]; while, in autopsies of T2D islets, there are an average of ~$7\%$ beta cells that undergo apoptosis [34]. To investigate whether the elevated cytosolic zinc in SLC30A8 mutants can protect cells from apoptosis after short-term extracellular zinc depletion, extracellular zinc provided by the culture medium was reduced by 5 µM TPEN, a chelator that has a high affinity for zinc and removes free zinc in the medium, and cell apoptosis was examined by staining the untreated and TPEN-treated cells with TUNEL. After 48 h of treatment, the percentage of TUNEL-positive cells was significantly increased after TPEN treatment compared to untreated control in WT (WT: 3.71 ± $2.65\%$ with TPEN; 0.52 ± $1.17\%$ without TPEN). In contrast, in SLC30A8 mutants, no significant increase in apoptotic cells was detected after TPEN treatment (KO: 0.64 ± $0.82\%$ with TPEN; 0.14 ± $0.55\%$ without TPEN) (R138X: 0.80 ± $0.95\%$ with TPEN; 0.28 ± $0.63\%$ without TPEN) (Figure 5C,D). Taken together, the above findings reveal that both R138X and KO loss of function mutations in SLC30A8 impair the crystallization of insulin due to the failure of channeling zinc into insulin granules. This can result in an elevation of free cytoplasmic zinc in the latter case, conferring resistance to apoptosis after extracellular zinc depletion. ## 3. Discussion In this study, we used human embryonic stem cells as a model to investigate the differentiation and function of SLC30A8 p.Arg138* variant, a patient mutation that is protective against T2D. Our results demonstrate that loss of function mutations in SLC30A8 locus do not influence differentiation toward the pancreatic lineage and to insulin-producing cells. This finding is consistent with observations in SLC30A8 KO mice which have normal beta cell mass and function [7,10,14]. A previous study has reported low differentiation efficiency of induced pluripotent stem cells (iPSCs) carrying mutations in SLC30A8 to insulin-producing cells [5]. However, this phenotype is not readily reconciled with the protective effect of the mutation, and the molecular mechanisms that might be responsible are not known. The role of other confounding factors in this discrepancy cannot be excluded, as iPSC lines can show variable differentiation competence [35]. The effect of SLC30A8 mutations on beta cells was evident by a lack of dense-core zinc positive insulin granules. Zinc is essential for the formation of crystalized insulin with dense core granules and regulated insulin secretion. Our study shows that both complete loss of function mutations (KO) and mutations in p.Arg138* variant (R138X) both possess reduced crystalized insulin granules and reduced granule zinc content while sc-beta cells from KO has lost crystalized granule more completely. A recent study showed that genes related to beta cell maturation are upregulated in SLC30A8 KO sc-derived beta cells [36], providing a potential mechanism for the protective effect of SLC30A8 mutations. In our scRNA-seq, we also observed upregulation of PCSK2, ONECUT2, and IAPP in SLC30A8 KO cells, concordant with reported findings [36]. Nevertheless, in our human embryonic stem cell model of ZnT8 loss of function mutations, R138X and KO sc-beta cell clusters were neither compromised nor improved in insulin processing or glucose-stimulated secretion. A previous study has shown that after transplantation, sc-beta cells are essentially equivalent to mature pancreatic beta cells [37], which is why we chose cell grafting into mice for functional testing of insulin secretion. After transplantation in vivo, the response to glucose of cells derived from R138X and KO stem cells is comparable with WT. In the ipGTT functional test, we observed a small improvement of glucose clearance in mice with SLC30A8 KO mutant at both 15 and 30 min after glucose injection. However, analysis of HbA1C levels did not reveal differences between mice dependent on glucose regulation by grafts of WT or SLC30A8 mutant cells; thus the functionally protective relevance of a potentially improved insulin secretion was not evident over the time frame of the mouse experiments performed here. Improved insulin secretion profiles were reported in a prior study [cite reference]. Our studies are not inconsistent with this report, but any effect is very small. The increased expression of genes involved in beta cell maturation or insulin secretion observed here as well as in a prior study (REF), is not inconsistent with functional improvements, but may also be a compensatory response to the uncrystallized insulin granules. We thus also evaluated other mechanisms of a protective effect of SLC30A8 mutations on beta cell function based on changes to zinc homeostasis. We find that cytoplasmic free zinc content remains the same in R138X, and is higher in KO cells than in controls. This indicates that loss function of ZnT8 influences the transportation of zinc from the cytoplasm into insulin granules, and that the disruption of transport of cytosolic zinc to the granule can increase the availability of cytosolic zinc. We find that the consequences of ZnT8 loss of function lead to significantly more resistance to cell apoptosis when zinc in culture media is reduced through chelation. A recent study also found that isolated islets from SLC30A8 KO mice are protected against hypoxia-induced cell death when compared to wildtype islets [38]. Zinc supplement was reported to have protective modulation against high glucose induced apoptosis in renal tubular epithelial cells [16]. Wild type mice on a limiting zinc diet show beta cell apoptosis and degeneration of pancreatic endocrine function [20]. Importantly, zinc has been implicated in T2D: hypozincemia was observed in patients with diabetes [17,18,39,40]. As serum zinc level decreases in the aging process [19,41], zinc deficiency may contribute to the rise in T2D incidence during aging. Our findings suggest that the lack of granular zinc may be protective, as it increases zinc availability in other cellular compartments. Whether ZnT8 deficiency protects against apoptosis under physiological circumstances in grafted mice as well as in patients remains to be determined. Ma et al. provided an alternative interpretation for the protective effect of ZnT8 in a stem cell model: an inhibitory effect of extracellular zinc on insulin secretion [36]. This zinc would only be released from zinc-containing insulin granules, not from ZnT8 mutant beta cells. However, experimental treatments involved very high concentrations of zinc (100 µM), and 100 µM extracellular zinc has a long-term toxic effect on glucose-stimulated electrical activity of pancreatic beta cells [42]. Lower concentrations of 10 µM have no significant effect [43]. The authors also showed that insulin knockout cells, but not SLC30A8 mutant cells, had an inhibitory effect on insulin secretion when aggregated with wild type beta cells. However, this would require the formation of granules without insulin, but with high zinc levels. Insulin granules form after proinsulin is produced and packed into granular vesicles [44]. Granules form through inclusion of the secreted protein, rather than through inclusion of metal ions alone. Hollow secretory granules without insulin but loaded with high zinc was not demonstrated, and how secreted zinc was measured has not been clarified in the study. Thus, whether the absence of zinc from secreted granules mediates the protective effect of ZnT8 loss of function is not clear. A limitation of this study is that no functional consequences of the heterozygous SLC30A8 mutation are apparent, neither in vitro nor in vivo. The consequences in homozygous mutant cells are surprisingly small, but consistent with a protective effect against T2D. Conclusions based on homozygous mutations assume that the effects of heterozygous mutations have the same directionality as a homozygous mutation, and that homozygous mutant cells can serve as an exaggerated model. It may be that the functional consequences of heterozygous loss of function mutations become apparent only after years of metabolic stress. Of note, the effect of the common T2D-associated SLC30A8 risk variant rs13266634 (R325W) on ZnT8 activity is likely to be a gain-of-function [45]. In addition, currently unknown are the molecular mechanisms for species-specific differences and the variability in phenotypic expression of ZnT8 deletion in mice. Interestingly, the inactivation of ZnT8 in mouse beta cells was reported to result in lowered free cytosolic Zn2+ levels, as measured using the same strategy as used here (the eCALWY GFP-based probe) [8,9], whereas human SLC30A8 KO sc-beta cells displayed the same or elevated cytosolic Zn2+. Loss of ZnT8 function in mice causes impaired glucose tolerance or has no effect [3], whereas in humans, rare loss of function variants offer protection against T2D [4,5]. Differences and potential variability in the effect on cytoplasmic zinc may thus underly these species-specific differences. In summary, our studies show upregulated cytoplasmic free zinc in the SLC30A8 KO sc-beta cells, and a protective effect on beta cell survival. We also demonstrate that ZnT8 is dispensable for both the generation and function of human stem cell-derived beta cells. Inferring from our findings, the protective effect of ZnT8 loss of function for T2D could potentially be substituted through a readily available zinc supplement. Consistent with this view, common SLC30A8 variants influence the protective effect of dietary zinc supplementation on T2D risk in humans [46,47,48]. Adding further relevance to this study, loss of ZnT8 function may facilitate the generation of improved and more apoptosis resistant stem cell derived beta cell grafts for different forms of diabetes, including for T1D where ZnT8 is an important autoantigen [21,49]. ## Note Added in Proof While this study was under consideration for publication, another study reported normal differentiation of ZnT8 loss of function stem cells to beta-like cells [36]. ## 4.1. Human Pluripotent Stem Cell Culture and Gene Editing with CRISPR/Cas9 Human pluripotent stem cells were cultured and maintained on feeder-free plates with StemFlex Medium (Cat. No. A3349401, Thermo Fisher Scientific, Waltham, MA, USA) as described [50]. We established cell lines with mutations by introducing homozygous p.Arg138* mutation and KO mutations into human embryonic stem cell line MEL1 (NIH registry #0139) with CRISPR/Cas9. This cell line carries in insulin-GFP knockin [51]. Guide RNA was designed by following the published protocol and synthesized by Integrated DNA Technologies (IDT) (https://www.idtdna.com/pages (accessed on 1 January 2023). A 120 bp repair ssDNA template with the desired sequence was synthesized by IDT (guide sequences and repair template sequence listed in Table S3). Cas9-GFP plasmid was obtained from Addgene (Cat. No. 44719). Then, 2.5 µg of each component were transfected into 1 million cells with LONZA nucleofector. Normal karyotypes for all cell lines were validated by Cell Line Genetics (Figure S1C). ## 4.2. Insulin-Producing Cell Differentiation from Human Pluripotent Stem Cells Insulin-producing cells were differentiated from human pluripotent stem cell lines at certain passages, ranging from 25–30 passages, using the published protocol [50], with aphidicolin treatment to obtain better differentiation and transplantation results [35]. First, cells were cultured for 4 days using STEMdiff™ Definitive Endoderm Differentiation Kit (Cat. No. 05110, STEMCELL Technologies, Vancouver, BC, Canada) for definitive endoderm induction. Primitive gut tube was induced by RPMI 1640 plus GlutaMAX (Cat. No. 61870-127, Life Technology, Carlsbad, CA, USA) + $1\%$ (v/v) Penicillin–Streptomycin (PS) (Cat. No. 15070-063, Thermo Fisher Scientific, Waltham, MA, USA) + $1\%$ (v/v) B-27 Serum-Free Supplement (50× (Cat. No. 17504044, Life Technology, Carlsbad, CA, USA) + 50 ng/mL FGF7 (Cat. No. 251-KG, R&D System, NE Minneapolis, MN, USA) from day 4 to day 6. Posterior foregut is induced by DMEM plus GlutaMax (DMEM) (Cat. No. 10569-044, Life Technology, Carlsbad, CA, USA) with $1\%$ (v/v) PS + $1\%$ (v/v) B-27 + 0.25 μM KAAD-Cyclopamine (Cat. No. 04-0028, Stemgent, Cambridge, MA, USA) + 2 μM Retinoic acid (Cat. No. 04-0021, Stemgent, Cambridge, MA, USA) + 0.25 μM LDN193189 (Cat. No. 04-0074, Stemgent, Cambridge, MA, USA) from day 6 to day 8. We then changed medium to DMEM + $1\%$ (v/v) PS + $1\%$ (v/v) B-27 + 50 ng/mL EGF (Cat. No. 236-EG, R&D System, NE Minneapolis, MN, USA) + 25 ng/mL FGF7 for 4 days to generate pancreatic progenitor cells. On day 12, cells were dissociated into single cells using TrypLETM Express (Cat. No. 12605036, Life Technologies, Carlsbad, CA, USA) and clustered in AggreWell 400 6-well plates (Cat. No. 34425, STEMCELL Technologies, Vancouver, Canada) or ultra-low attachment 96-well plates (Cat. No. 07-201-680, Thermo Fisher Scientific, Waltham, MA, USA) with DMEM + $1\%$ (v/v) PS + $1\%$ (v/v) B-27 + 1 μM ALK5 inhibitor (Stemgent, cat. No. 04-0015) + 10 μg/mL heparin (Cat. No. H3149, Sigma-Aldrich, Burlington, MA, USA) + 25 ng/mL FGF7 + 10 μM Y-27632, ROCK inhibitor. On day 13, newly formed clusters were transferred into ultra-low attachment 6-well plates (Cat. No. 07-200-601, Thermo Fisher Scientific, Waltham, MA, USA) (or remaining culture in the 96-well ultra-low attachment plates), and medium was changed to RPMI 1640 plus GlutaMAX + $1\%$ (v/v) PS + $1\%$ (v/v) B-27 + 1 μM thyroid hormone (T3) (Cat. No. T6397, Sigma-Aldrich, Burlington, MA, USA) + 10 μM ALK5 inhibitor + 10 μM zinc sulfate (Cat. No. Z4750, Sigma-Aldrich, Burlington, MA, USA) + 10 μg/mL heparin + 100 nM gamma-secretase inhibitor (DBZ) (EMD Millipore, cat. No. 565789) + 10 μM Y-27632, ROCK inhibitor for 7 days. From day 20 to day 27, to induce pancreatic beta cell, medium was changed to RPMI 1640 plus GlutaMAX + $1\%$ (v/v) PS + $1\%$ (v/v) B-27 + $10\%$ (v/v) fetal bovine serum (Cat. No. S11150, Atlanta Biologicals, Flowery Branch, GA, USA). From day 15 to day 27, besides the component mentioned above, 2 μM aphidicolin (Cat. No. A0781-1MG, Sigma-Aldrich, Burlington, MA, USA) was added to the medium for differentiation improvement [35]. Note that the MEL1 cell line only has one insulin allele, but it is sufficient in blood glucose regulation as shown in all of our functional studies in vivo. ## 4.3. Dithizone Staining Dithizone stock solution (1 mg/mL) was prepared by dissolving dithizone (Cat. No. D5130, Sigma-Aldrich) into DMSO and storing at −20 °C. Stem cell derived beta-like cell clusters were stained with dithizone by adding 50 µL of dithizone stock solution into 1 mL of medium to reach a final concentration of 50 µg/mL, and this was incubated for 1 min. Pictures were taken with an OLYMPUS IX73 microscope. ## 4.4. Immunocytochemistry At 27 days of differentiation, clusters were fixed with $4\%$ paraformaldehyde (PFA) at room temperature (RT) for 10 min. Grafts taken from the mice were also fixed with $4\%$ PFA at RT for 1 h. The following steps were performed according to the published method [50]. Primary antibodies are listed in Supplementary Table S1, and secondary antibodies are listed in Supplementary Table S2. Pictures were taken with an OLYMPUS IX73 fluorescent microscope or ZEISS LSM 710 confocal microscope. For each panel, pictures were taken with same exposure time. ## 4.5. Flow Cytometry The beta-like cell clusters were treated with TrypLETM Express (Cat. No. 12605036, Life Technologies) into single cells. Then, they were fixed with $4\%$ PFA for 10 min and permeabilizated with cold methanol at −20 °C for 10 min. Primary antibodies were added at a dilution of indicated ratio in Table S1 in phosphate buffered saline (PBS) containing $0.5\%$ BSA at 4 °C for 1 h. Secondary antibodies were added accordingly (Table S2) at a dilution of 1:500 at room temperature for 1 h. The cells were then filtered with BD Falcon 12 mm × 75 mm tube with a cell strainer cap prior to flow cytometry analysis. ## 4.6. Single-Cell RNA Sequencing and Read Mapping A total of 21,273 cells (WT: 8648; KO: 6397; R138X: 6228) were sequenced. Preparation of the cells was performed according to methods previously described [27]. Briefly, single cells were suspended in PBS + $0.04\%$ BSA and for cell hashing, totalseq-A anti-human hashtag antibodies (BioLegend) were used. Samples from different groups were individually stained with one of the hashtag antibodies and washed three times. Twelve samples for each were pooled at equal concentrations, and the pool was loaded into the 10X Chromium instrument at 32,000 cells per lane. Single-cell RNA-seq libraries were prepared using Chromium Single Cell 3′ Reagent Kits v2 (10X Genomics). Hashtag libraries were generated as described in [52]. Illumina NextSeq500 was used for sequencing. Cell Ranger Single-Cell Software Suite (10X Genomics, v2) was utilized for sequence alignment and quantification of expression. Reads were aligned to the B37.3 Human Genome assembly and UCSC gene model. ## 4.7. Single-Cell Data Analysis Single cell data analysis was performed as previously described [35]. Cell-hashing tags were demultiplexed using HTODemux function (Seurat v3). We excluded empty droplets and doublets. The exclusion criteria were as follows: [1] total UMI bigger or smaller than three folds of median absolute deviation (MAD); [2] detected genes bigger or smaller than 3 folds of MAD; [3] detected genes in the first of the bimodal distribution (classified by mclust); [4] mitochondrial gene ratio bigger than 0.15. Cells from three experiment groups (WT, KO, and R138X) were integrated and clustered to identify cell types and subpopulations (Seurat v3). ## 4.8. Static Glucose Stimulated Insulin Secretion Krebs Ringer buffer (KRB) was prepared by addition of 129 mM NaCl, 4.8 mM KCl, 2.5 mM CaCl2, 1.2 mM MgSO4, 1 mM Na2HPO4, 1.2 mM KH2PO4, 5 mM NaHCO3, 10 mM HEPES and $0.1\%$ BSA in deionized water and was sterilized with a 0.22 μm filter. The 2 mM glucose solutions were prepared in KRB for low glucose challenge of sc-beta cell clusters. Then, 10–20 sc-beta cell clusters (~5 × 105 cells) were collected from WT, KO, and R138X and pre-incubated in 500 μL 2 mM glucose solution for 1 h. Clusters were then washed once with 2 mM glucose solution and subsequently incubated in 200 µL of 2 mM glucose for 1 h. Finally, 130 μL supernatant from each condition was collected. Cell clusters were centrifuged down and resuspended in 50 µL high salt buffer and sonicated for protein content preparation and DNA measurement with Nano Drop Spectrophotometer ND-1000. Protein content and supernatants in 2 mM glucose solution were processed using Mercodia Insulin ELISA kit (Cat. No. 10-1113-01, Mercodia, Uppsala, Sweden). ## 4.9. Transplantation and In Vivo Assay The 8–10 weeks old male immunocompromised mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) from Jackson laboratories, Cat. No. 005557) were used for transplantation. For intra-leg muscle transplantation, ~2 million cells were collected and transferred to a tube with 50 μL Matrigel (Cat. No. 354277, Fisher Scientific). Cluster injections in the leg muscle were performed using a 21G × $\frac{11}{2}$” needle (Cat. No. 305177). The human C-peptide levels in mouse serum were measured every two weeks in the fed state. For analysis of glucose stimulated insulin secretion, at 2 weeks after mouse beta-like cells were ablated with one high dose (150 mg/kg) of streptozotocin (Cat. No. S0130-1G, Sigma-Aldrich), intraperitoneal glucose tolerance test (ipGTT) was performed by fasting overnight (for 16 h) with bedding change and injecting 2 g/kg D-(+)-glucose (Cat. No. G8270, Sigma-Aldrich, Burlington, MA, USA) solution. Blood was obtained by clipping tail tips of the mice and was collected using 75 mm heparinized capillary tubes (Cat. No. 1-000-7500-HC/5, Drummond Scientific, Broomall, PA, USA), then transferred to heparin-coated tubes (Cat. No. 022379208, Eppendorf, Hamburg, Germany) at fed state, at fasting state, and 30 min after glucose injection. Plasma was collected by centrifuging heparin-coated tubes at 3000× g for 15 min at 4 °C. The supernatants were collected for C-peptide and insulin detection with Mercodia Insulin ELISA kit (Cat. No. 10-1113-01, Mercodia, Uppsala, Sweden), Mercodia Ultrasensitive C-peptide ELISA (Cat. No. 10-1141-01, Mercodia, Uppsala, Sweden), and Mercodia Proinsulin ELISA (Cat. No. 10-1118-01, Mercodia, Uppsala, Sweden). Blood glucose levels were measured by a glucometer (FreeStyle Lite, Abbott, Chicago, IL, USA), using a drop of blood obtained from clipping mice tail tips at fed state, fasting state, 15min, and every 30 min after glucose injection for 2 h. All animal protocols were approved by the Institutional Animal Care and Use Committee of Columbia University. ## 4.10. Quantification of Zinc Concentration Using FRET Sensor eCALWY-4 The dual read-out FRET-based assays for intracellular zinc measurement were performed as described in the previous literature [53]. Cell clusters were infected with an adenovirus construct for the genetically encoded zinc sensor eCALWY-4 and let to express for 24 h. Before imaging, cells clusters were dissociated using accutase for 5 min at 37 °C and dispersed by pipetting up and down. Cells were then pelleted, resuspended in normal media, and let to attach for 3 h on glass slides treated with polylysine; then, zinc imaging was performed as previously described [54]. Cells on coverslips were washed twice in Krebs-HEPES-bicarbonate (KHB) buffer (140 mM NaCl, 3.6 mM KCl, 0.5 mM NaH2PO4, 0.2 mM MgSO4, 1.5 mM CaCl2, 10 mM Hepes, 25 mM NaHCO3), which was warmed, bubbled with $95\%$ O2: $5\%$ CO2, set to pH 7.4, and contained 11mM glucose. Cells were then transferred in an imaging chamber and maintained at 37 °C throughout with a heating stage (MC60, LINKAM, Scientific Instruments, Redhill, UK), and KHB buffer was perfused (1.5 to 2 mL/min). Images were captured at 433 nm monochromatic excitation wavelength (Polychrome IV, Till photonics) using an Olympus IX-70 wide-field microscope with a 40×/1.35NA oil immersion objective and a zyla sCMOS camera (Andor Technology, Belfast, UK) controlled by Micromanager software. Acquisition rate was set at 20 images/minute. Emitted light was split and filtered by a Dual-View beam splitter (Photometrics) equipped with a 505dcxn dichroic mirror and two emission filters (Chroma Technology—D$\frac{470}{24}$ for cerulean and D$\frac{535}{30}$ for citrine). ## 4.11. TPEN Treatment and Quantification of TUNEL/C-Peptide Double Positive sc-Beta Cells On day 27 of differentiation, clusters were treated with 5 μM of TPEN for 48 h. Preparations for cryosection were performed following the published method [50] and the above. For TUNEL/immuno-double staining, we followed the protocol described previously [55]. For the quantification of TUNEL/c-peptide double positive cells out of total c-peptide positive cells per cluster section, we performed manual counting using ZEISS Blue Edition. ## 4.12. Image Analysis of Fluorescence Emission Ratio Image analysis process was described previously [54]. Briefly, ImageJ software was used with a designed macro and after subducting the background, the fluorescence emission ratios were derived. Steady-state fluorescence intensity ratio citrine/cerulean was evaluated. Next, maximum and minimum ratios were measured for the calculation of free cytosolic Zn2+ concentration. The value of the concentration was determined using the following formula: [Zn2+] = Kd × (Rmax − R)/(R − Rmin). In the formula, the maximum ratio (Rmax) was achieved when intracellular zinc chelated with 50 μM TPEN. Additionally, the minimum ratio (Rmin) was achieved when adding 100 μM ZnCl2 with Zn2+ ionophore, pyrithione (5 μM) resulted in reaching zinc saturation. ## 4.13. In Vivo/In Vitro Graft Imaging and Graft Content Measurement SLC30A8 mutant and wildtype MEL1 lines possess GAPDHLuciferase/wt and INSGFP/wt double reporter. Before bioluminescence and fluorescence imaging, the NSG mice transplanted with SLC30A8 mutant and wildtype grafts were i.p injected with 150 mg/kg body weight of D-luciferin potassium salt (Gold Biotechnology, luck-2G) in PBS at least 15 min before imaging on a IVIS spectrum optical imaging system (PerkinElmer) (previously described in [35]). Signals were acquired with 1 min exposure and analyzed using the Living image analysis software (Xenogen Corp. version 4.0). Equal sizes of regions of interest (ROI) were drawn for all graft imaging in the same panel. Signal in the left thigh and photons emitted over the time of exposure within the ROI were measured. Luminescence was measured as described for bioluminescence after isolation of the grafts. Background signals were subtracted from a nearby region. For the graft insulin/proinsulin content measurement, we first isolated GFP positive parts from the leg based on in vitro imaging. Individual graft pieces were snap frozen in liquid nitrogen and homogenized into powder using a mortar and pestle. Then, 1mL of complete extraction buffer (100 mM Tris, pH 7.4 + 150 mM NaCl + 1 mM EGTA + 1 mM EDTA + $1\%$ Triton X-100 + $0.5\%$ Sodium deoxycholate + 100× protease and phosphatase inhibitor cocktail (Cat. No. PPC1010, Sigma-Aldrich) + 100× PMSF (Cat. No. P7626, Sigma-Aldrich)) was added to each sample and transferred into a 2 mL tube prefilled with Triple-Pure High Impact Zirconium Beads (Cat. No. D1032-30, Benchmark Scientific). Then, this was homogenized again using an electrical homogenizer. After the second homogenization, samples were centrifuged for 20 min at 13,000 rpm at 4 °C. The supernatants were then collected and with a dilution of 1000×, the insulin and proinsulin content of the grafts were then measured using Mercodia insulin and Mercodia proinsulin kits. For the complete extraction buffer, we followed a recipe from Abcam: https://www.abcam.com/protocols/elisa-sample-preparation-guide-1 (accessed on 1 January 1998). ## 4.14. Electron Microscopy The isolated graft was dissected into approximately 1 mm3 biopsies and placed in fixative ($2.5\%$ glutaraldehyde and $2\%$ paraformaldehyde in 100 mM sodium cacodylate buffer (pH 7.4) (Cat. No. 15960-01, EMSDIASUM) overnight at 4 °C. Samples were then sent to the Electron Microscopy Lab at Nathan S. Kline Institute (http://cdr.rfmh.org/about_facilities.html (accessed on 1 January 2007) for further processing and imaging. The mouse strain was generated by Regeneron for the pancreas sectioning and imaging in Figure 2C. ## 4.15. Statistical Analysis Data were analyzed using an unpaired t-test (for 2-sample comparison) and one-way ANOVA followed by Tukey’s multiple comparison test (for ≥3 samples) (GraphPad Prism 6, GraphPad Software, Inc., La Jolla, CA, USA). All data were plotted as mean ± standard deviation (SD). The differences observed were evaluated as statistically significant when p-value was less than 0.05 and were displayed on figures as follows: * $p \leq 0.05$, $p \leq 0.01$, * $p \leq 0.001$, ** $p \leq 0.0001.$ ## References 1. Davidson H.W., Wenzlau J.M., O’Brien R.M.. Zinc transporter 8 (ZnT8) and beta cell function*. Trends Endocrinol. Metab.* (2014) 25 415-424. DOI: 10.1016/j.tem.2014.03.008 2. Chimienti F., Seve M., Richard S., Mathieu J., Favier A.. Role of cellular zinc in programmed cell death: Temporal relationship between zinc depletion, activation of caspases, and cleavage of Sp family transcription factors. Biochem. Pharmacol. (2001) 62 51-62. DOI: 10.1016/S0006-2952(01)00624-4 3. Rutter G.A., Chimienti F.. SLC30A8 mutations in type 2 diabetes. Diabetologia (2015) 58 31-36. DOI: 10.1007/s00125-014-3405-7 4. Flannick J., Thorleifsson G., Beer N.L., Jacobs S.B., Grarup N., Burtt N.P., Mahajan A., Fuchsberger C., Atzmon G., Benediktsson R.. Loss-of-function mutations in SLC30A8 protect against type 2 diabetes. Nat. Genet. (2014) 46 357-363. DOI: 10.1038/ng.2915 5. Dwivedi O.P., Lehtovirta M., Hastoy B., Chandra V., Krentz N.A.J., Kleiner S., Jain D., Richard A.M., Abaitua F., Beer N.L.. Loss of ZnT8 function protects against diabetes by enhanced insulin secretion. Nat. Genet. (2019) 51 1596-1606. DOI: 10.1038/s41588-019-0513-9 6. Snitsarev V., Budde T., Stricker T.P., Cox J.M., Krupa D.J., Geng L., Kay A.R.. Fluorescent detection of Zn(2+)-rich vesicles with Zinquin: Mechanism of action in lipid environments. Biophys. J. (2001) 80 1538-1546. DOI: 10.1016/S0006-3495(01)76126-7 7. Lemaire K., Ravier M.A., Schraenen A., Creemers J.W., Van de Plas R., Granvik M., Van Lommel L., Waelkens E., Chimienti F., Rutter G.A.. Insulin crystallization depends on zinc transporter ZnT8 expression, but is not required for normal glucose homeostasis in mice. Proc. Natl. Acad. Sci. USA (2009) 106 14872-14877. DOI: 10.1073/pnas.0906587106 8. Gerber P.A., Bellomo E.A., Hodson D.J., Meur G., Solomou A., Mitchell R.K., Hollinshead M., Chimienti F., Bosco D., Hughes S.J.. Hypoxia lowers SLC30A8/ZnT8 expression and free cytosolic Zn2+ in pancreatic beta cells. Diabetologia (2014) 57 1635-1644. DOI: 10.1007/s00125-014-3266-0 9. Mitchell R.K., Hu M., Chabosseau P.L., Cane M.C., Meur G., Bellomo E.A., Carzaniga R., Collinson L.M., Li W.H., Hodson D.J.. Molecular Genetic Regulation of Slc30a8/ZnT8 Reveals a Positive Association with Glucose Tolerance. Mol. Endocrinol. (2016) 30 77-91. DOI: 10.1210/me.2015-1227 10. Nicolson T.J., Bellomo E.A., Wijesekara N., Loder M.K., Baldwin J.M., Gyulkhandanyan A.V., Koshkin V., Tarasov A.I., Carzaniga R., Kronenberger K.. Insulin storage and glucose homeostasis in mice null for the granule zinc transporter ZnT8 and studies of the type 2 diabetes-associated variants. Diabetes (2009) 58 2070-2083. DOI: 10.2337/db09-0551 11. Pound L.D., Sarkar S.A., Benninger R.K., Wang Y., Suwanichkul A., Shadoan M.K., Printz R.L., Oeser J.K., Lee C.E., Piston D.W.. Deletion of the mouse Slc30a8 gene encoding zinc transporter-8 results in impaired insulin secretion. Biochem. J. (2009) 421 371-376. DOI: 10.1042/BJ20090530 12. Pound L.D., Sarkar S.A., Ustione A., Dadi P.K., Shadoan M.K., Lee C.E., Walters J.A., Shiota M., McGuinness O.P., Jacobson D.A.. The physiological effects of deleting the mouse SLC30A8 gene encoding zinc transporter-8 are influenced by gender and genetic background. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0040972 13. Tamaki M., Fujitani Y., Hara A., Uchida T., Tamura Y., Takeno K., Kawaguchi M., Watanabe T., Ogihara T., Fukunaka A.. The diabetes-susceptible gene SLC30A8/ZnT8 regulates hepatic insulin clearance. J. Clin. Invest. (2013) 123 4513-4524. DOI: 10.1172/JCI68807 14. Kleiner S., Gomez D., Megra B., Na E., Bhavsar R., Cavino K., Xin Y., Rojas J., Dominguez-Gutierrez G., Zambrowicz B.. Mice harboring the human SLC30A8 R138X loss-of-function mutation have increased insulin secretory capacity. Proc. Natl. Acad. Sci. USA (2018) 115 E7642-E7649. DOI: 10.1073/pnas.1721418115 15. Scharfmann R., Pechberty S., Hazhouz Y., von Bulow M., Bricout-Neveu E., Grenier-Godard M., Guez F., Rachdi L., Lohmann M., Czernichow P.. Development of a conditionally immortalized human pancreatic beta cell line. J. Clin. Invest. (2014) 124 2087-2098. DOI: 10.1172/JCI72674 16. Zhang X., Zhao Y., Chu Q., Wang Z.Y., Li H., Chi Z.H.. Zinc modulates high glucose-induced apoptosis by suppressing oxidative stress in renal tubular epithelial cells. Biol. Trace. Elem. Res. (2014) 158 259-267. DOI: 10.1007/s12011-014-9922-x 17. Al-Maroof R.A., Al-Sharbatti S.S.. Serum zinc levels in diabetic patients and effect of zinc supplementation on glycemic control of type 2 diabetics. Saudi. Med. J. (2006) 27 344-350. PMID: 16532095 18. Farooq D.M., Alamri A.F., Alwhahabi B.K., Metwally A.M., Kareem K.A.. The status of zinc in type 2 diabetic patients and its association with glycemic control. J. Family Community Med. (2020) 27 29-36. DOI: 10.4103/jfcm.JFCM_113_19 19. Prasad A.S., Beck F.W., Bao B., Fitzgerald J.T., Snell D.C., Steinberg J.D., Cardozo L.J.. Zinc supplementation decreases incidence of infections in the elderly: Effect of zinc on generation of cytokines and oxidative stress. Am. J. Clin. Nutr. (2007) 85 837-844. DOI: 10.1093/ajcn/85.3.837 20. Sisnande T., Lima C.K., da Silva D.C., Beninatto T.M., Alves N.L., Amaral M.J., Miranda-Alves L., Lima L.. Dietary zinc restriction promotes degeneration of the endocrine pancreas in mice. Biochim. Biophys. Acta. Mol. Basis. Dis. (2020) 1866 165675. DOI: 10.1016/j.bbadis.2020.165675 21. Wenzlau J.M., Juhl K., Yu L., Moua O., Sarkar S.A., Gottlieb P., Rewers M., Eisenbarth G.S., Jensen J., Davidson H.W.. The cation efflux transporter ZnT8 (Slc30A8) is a major autoantigen in human type 1 diabetes. Proc. Natl. Acad. Sci. USA (2007) 104 17040-17045. DOI: 10.1073/pnas.0705894104 22. Dodson G., Steiner D.. The role of assembly in insulin’s biosynthesis. Curr. Opin. Struct. Biol. (1998) 8 189-194. DOI: 10.1016/S0959-440X(98)80037-7 23. Williams C.L., Long A.E.. What has zinc transporter 8 autoimmunity taught us about type 1 diabetes?. Diabetologia (2019) 62 1969-1976. DOI: 10.1007/s00125-019-04975-x 24. Murgia C., Devirgiliis C., Mancini E., Donadel G., Zalewski P., Perozzi G.. Diabetes-linked zinc transporter ZnT8 is a homodimeric protein expressed by distinct rodent endocrine cell types in the pancreas and other glands. Nutr. Metab. Cardiovasc. Dis. (2009) 19 431-439. DOI: 10.1016/j.numecd.2008.09.004 25. Maeda M., Abiko N., Sasaki T.. Synthesis of seleno- and thioguanine-platinum(II) complexes and their antitumor activity in mice. J. Med. Chem. (1981) 24 167-169. DOI: 10.1021/jm00134a008 26. Westin G., Schaffner W.. A zinc-responsive factor interacts with a metal-regulated enhancer element (MRE) of the mouse metallothionein-I gene. EMBO J. (1988) 7 3763-3770. DOI: 10.1002/j.1460-2075.1988.tb03260.x 27. Gonzalez B.J., Zhao H., Niu J., Williams D.J., Lee J., Goulbourne C.N., Xing Y., Wang Y., Oberholzer J., Blumenkrantz M.H.. Reduced calcium levels and accumulation of abnormal insulin granules in stem cell models of HNF1A deficiency. Commun. Biol. (2022) 5 779. DOI: 10.1038/s42003-022-03696-z 28. Iadevaia V., Liu R., Proud C.G.. mTORC1 signaling controls multiple steps in ribosome biogenesis. Semin. Cell. Dev. Biol. (2014) 36 113-120. DOI: 10.1016/j.semcdb.2014.08.004 29. Lynch C.J., Patson B.J., Goodman S.A., Trapolsi D., Kimball S.R.. Zinc stimulates the activity of the insulin- and nutrient-regulated protein kinase mTOR. Am. J. Physiol. Endocrinol. Metab. (2001) 281 E25-E34. DOI: 10.1152/ajpendo.2001.281.1.E25 30. Vinkenborg J.L., Nicolson T.J., Bellomo E.A., Koay M.S., Rutter G.A., Merkx M.. Genetically encoded FRET sensors to monitor intracellular Zn2+ homeostasis. Nat. Methods (2009) 6 737-740. DOI: 10.1038/nmeth.1368 31. Sunderman F.W.. The influence of zinc on apoptosis. Ann. Clin. Lab. Sci. (1995) 25 134-142. PMID: 7785963 32. Truong-Tran A.Q., Carter J., Ruffin R.E., Zalewski P.D.. The role of zinc in caspase activation and apoptotic cell death. Biometals (2001) 14 315-330. DOI: 10.1023/A:1012993017026 33. Butler A.E., Janson J., Bonner-Weir S., Ritzel R., Rizza R.A., Butler P.C.. Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes (2003) 52 102-110. DOI: 10.2337/diabetes.52.1.102 34. Jurgens C.A., Toukatly M.N., Fligner C.L., Udayasankar J., Subramanian S.L., Zraika S., Aston-Mourney K., Carr D.B., Westermark P., Westermark G.T.. beta-cell loss and beta-cell apoptosis in human type 2 diabetes are related to islet amyloid deposition. Am. J. Pathol. (2011) 178 2632-2640. DOI: 10.1016/j.ajpath.2011.02.036 35. Sui L., Xin Y., Du Q., Georgieva D., Diedenhofen G., Haataja L., Su Q., Zuccaro M.V., Kim J., Fu J.. Reduced replication fork speed promotes pancreatic endocrine differentiation and controls graft size. JCI Insight (2021) 6 e141553. DOI: 10.1172/jci.insight.141553 36. Ma Q., Xiao Y., Xu W., Wang M., Li S., Yang Z., Xu M., Zhang T., Zhang Z.N., Hu R.. ZnT8 loss-of-function accelerates functional maturation of hESC-derived beta cells and resists metabolic stress in diabetes. Nat. Commun. (2022) 13 4142. DOI: 10.1038/s41467-022-31829-9 37. Augsornworawat P., Maxwell K.G., Velazco-Cruz L., Millman J.R.. Single-Cell Transcriptome Profiling Reveals beta Cell Maturation in Stem Cell-Derived Islets after Transplantation. Cell Rep. (2020) 32 108067. DOI: 10.1016/j.celrep.2020.108067 38. Karsai M., Zuellig R.A., Lehmann R., Cuozzo F., Nasteska D., Luca E., Hantel C., Hodson D.J., Spinas G.A., Rutter G.A.. Lack of ZnT8 protects pancreatic islets from hypoxia- and cytokine-induced cell death. J. Endocrinol. (2022) 253 1-11. DOI: 10.1530/JOE-21-0271 39. Roussel A.M., Kerkeni A., Zouari N., Mahjoub S., Matheau J.M., Anderson R.A.. Antioxidant effects of zinc supplementation in Tunisians with type 2 diabetes mellitus. J. Am. Coll. Nutr. (2003) 22 316-321. DOI: 10.1080/07315724.2003.10719310 40. Garg V.K., Gupta R., Goyal R.K.. Hypozincemia in diabetes mellitus. J. Assoc. Physicians. India. (1994) 42 720-721. PMID: 7883669 41. Haase H., Rink L.. The immune system and the impact of zinc during aging. Immun. Ageing (2009) 6 9. DOI: 10.1186/1742-4933-6-9 42. Ferrer R., Soria B., Dawson C.M., Atwater I., Rojas E.. Effects of Zn2+ on glucose-induced electrical activity and insulin release from mouse pancreatic islets. Am. J. Physiol. (1984) 246 C520-C527. DOI: 10.1152/ajpcell.1984.246.5.C520 43. Ghafghazi T., McDaniel M.L., Lacy P.E.. Zinc-induced inhibition of insulin secretion from isolated rat islets of Langerhans. Diabetes (1981) 30 341-345. DOI: 10.2337/diab.30.4.341 44. Hou J.C., Min L., Pessin J.E.. Insulin granule biogenesis, trafficking and exocytosis. Vitam. Horm. (2009) 80 473-506. DOI: 10.1016/S0083-6729(08)00616-X 45. Merriman C., Huang Q., Rutter G.A., Fu D.. Lipid-tuned Zinc Transport Activity of Human ZnT8 Protein Correlates with Risk for Type-2 Diabetes. J. Biol. Chem. (2016) 291 26950-26957. DOI: 10.1074/jbc.M116.764605 46. Maruthur N.M., Clark J.M., Fu M., Linda Kao W.H., Shuldiner A.R.. Effect of zinc supplementation on insulin secretion: Interaction between zinc and SLC30A8 genotype in Old Order Amish. Diabetologia (2015) 58 295-303. DOI: 10.1007/s00125-014-3419-1 47. Kanoni S., Nettleton J.A., Hivert M.F., Ye Z., van Rooij F.J., Shungin D., Sonestedt E., Ngwa J.S., Wojczynski M.K., Lemaitre R.N.. Total zinc intake may modify the glucose-raising effect of a zinc transporter (SLC30A8) variant: A 14-cohort meta-analysis. Diabetes (2011) 60 2407-2416. DOI: 10.2337/db11-0176 48. Drake I., Hindy G., Ericson U., Orho-Melander M.. A prospective study of dietary and supplemental zinc intake and risk of type 2 diabetes depending on genetic variation in SLC30A8. Genes. Nutr. (2017) 12 30. DOI: 10.1186/s12263-017-0586-y 49. Ogura H., Preston-Hurlburt P., Perdigoto A.L., Amodio M., Krishnaswamy S., Clark P., Yu H., Egli D., Fouts A., Steck A.K.. Identification and Analysis of Islet Antigen-Specific CD8(+) T Cells with T Cell Libraries. J. Immunol. (2018) 201 1662-1670. DOI: 10.4049/jimmunol.1800267 50. Sui L., Leibel R.L., Egli D.. Pancreatic Beta Cell Differentiation From Human Pluripotent Stem Cells. Curr. Protoc. Hum. Genet. (2018) 99 e68. DOI: 10.1002/cphg.68 51. Micallef S.J., Li X., Schiesser J.V., Hirst C.E., Yu Q.C., Lim S.M., Nostro M.C., Elliott D.A., Sarangi F., Harrison L.C.. INS(GFP/w) human embryonic stem cells facilitate isolation of in vitro derived insulin-producing cells. Diabetologia (2012) 55 694-706. DOI: 10.1007/s00125-011-2379-y 52. Stoeckius M., Zheng S., Houck-Loomis B., Hao S., Yeung B.Z., Mauck W.M., Smibert P., Satija R.. Cell Hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics. Genome. Biol. (2018) 19 224. DOI: 10.1186/s13059-018-1603-1 53. Aper S.J., Dierickx P., Merkx M.. Dual Readout BRET/FRET Sensors for Measuring Intracellular Zinc. ACS Chem. Biol. (2016) 11 2854-2864. DOI: 10.1021/acschembio.6b00453 54. Chabosseau P., Tuncay E., Meur G., Bellomo E.A., Hessels A., Hughes S., Johnson P.R., Bugliani M., Marchetti P., Turan B.. Mitochondrial and ER-targeted eCALWY probes reveal high levels of free Zn2+. ACS Chem. Biol. (2014) 9 2111-2120. DOI: 10.1021/cb5004064 55. Ray S.K., Schaecher K.E., Shields D.C., Hogan E.L., Banik N.L.. Combined TUNEL and double immunofluorescent labeling for detection of apoptotic mononuclear phagocytes in autoimmune demyelinating disease. Brain Res. Brain Res. Protoc. (2000) 5 305-311. DOI: 10.1016/S1385-299X(00)00027-1
--- title: AMPK Phosphorylation Impacts Apoptosis in Differentiating Myoblasts Isolated from Atrophied Rat Soleus Muscle authors: - Natalia A. Vilchinskaya - Sergey V. Rozhkov - Olga V. Turtikova - Timur M. Mirzoev - Boris S. Shenkman journal: Cells year: 2023 pmcid: PMC10047078 doi: 10.3390/cells12060920 license: CC BY 4.0 --- # AMPK Phosphorylation Impacts Apoptosis in Differentiating Myoblasts Isolated from Atrophied Rat Soleus Muscle ## Abstract Regrowth of atrophied myofibers depends on muscle satellite cells (SCs) that exist outside the plasma membrane. Muscle atrophy appears to result in reduced number of SCs due to apoptosis. Given reduced AMP-activated protein kinase (AMPK) activity during differentiation of primary myoblasts derived from atrophic muscle, we hypothesized that there may be a potential link between AMPK and susceptibility of differentiating myoblasts to apoptosis. The aim of this study was to estimate the effect of AMPK activation (via AICAR treatment) on apoptosis in differentiating myoblasts derived from atrophied rat soleus muscle. Thirty rats were randomly assigned to the following two groups: control (C, $$n = 10$$) and 7-day hindlimb suspension (HS, $$n = 20$$). Myoblasts derived from the soleus muscles of HS rats were divided into two parts: AICAR-treated cells and non-treated cells. Apoptotic processes were evaluated by using TUNEL assay, RT-PCR and WB. In differentiating myoblasts derived from the atrophied soleus, there was a significant decrease ($p \leq 0.05$) in AMPK and ACC phosphorylation in parallel with increased number of apoptotic nuclei and a significant upregulation of pro-apoptotic markers (caspase-3, -9, BAX, p53) compared to the cells derived from control muscles. AICAR treatment of atrophic muscle-derived myoblasts during differentiation prevented reductions in AMPK and ACC phosphorylation as well as maintained the number of apoptotic nuclei and the expression of pro-apoptotic markers at the control levels. Thus, the maintenance of AMPK activity can suppress enhanced apoptosis in differentiating myoblasts derived from atrophied rat soleus muscle. ## 1. Introduction Skeletal muscle satellite cells (SCs), also called muscle stem cells, are known to play a crucial role in muscle fiber maintenance, regeneration and (re)growth. Under unstressed conditions, SCs, located at the periphery of myofibers under the basal lamina, exist in a quiescent state (i.e., G0 phase of the cell cycle). However, upon stimulation, SCs exit their quiescent state and start to proliferate and differentiate. Differentiated myoblasts (the progeny of SCs) can fuse either into existing myofibers or with each other, forming new muscle fibers [1,2]. Inactivity/mechanical unloading is known to result in a decrease in the number of SCs in postural muscles [3,4,5,6]. Evidence suggests that SC depletion under degenerative conditions (Duchenne muscular dystrophy, chronic muscle denervation, and aging) can be related, at least partially, to apoptosis [7,8]. Available data concerning SC function under unloading/disuse conditions are contradictory. Some authors report a decrease in SC proliferation and differentiation under conditions of mechanical unloading [4,9], while others mark an increase in the activity of muscle SCs [10,11,12]. AMP-activated protein kinase (AMPK) is a well-known cell energy gauge, the activity of which is determined by the AMP:ATP ratio in the cell [13,14]. Under conditions of energy deprivation, AMPK is known to play an important role in the regulation of pathways related to fatty acid and cholesterol metabolism, mitochondrial biogenesis, anabolism, catabolism, autophagy, and coordination of cell survival [15]. In addition, the lack of AMPK in SCs can block normal muscle regeneration after injury [16]. It has been shown that in AMPKα1 knockout mice, skeletal muscle regeneration following injury is significantly weakened compared to wild-type mice. Moreover, AMPKα1 knockout SCs have reduced myogenic capacity when transplanted into wild-type muscles, suggesting that impaired muscle regeneration could be linked to the absence of AMPKα1 in the SCs [17]. Fu et al. [ 2015] have demonstrated that in response to muscle injury, AMPKα1 can serve as a critical mediator linking a non-canonical Sonic hedgehog pathway to Warburg-like glycolysis in SCs, thereby contributing to the activation of muscle stem cells skeletal muscle regeneration [18]. Thus, a certain level of AMPKα1 appears to be required for proper skeletal muscle regeneration following injury [19]. It has also been demonstrated that the loss of AMPK activity is a major reason for impaired muscle regeneration in obese mice [17]. However, hyperactivation of AMPK is able to impair SC proliferation and differentiation [20,21]. Furthermore, AMPK is involved in the regulation of apoptosis that normally accompanies myogenic differentiation of myoblasts [15]. Niesler et al. [ 2007] have shown that some level of AMPK activity is needed to inhibit apoptotic processes in differentiated C2C12 myotubes [22]. Moreover, we have recently found an accelerated differentiation and myotube formation in primary myoblasts derived from rat soleus muscles that were exposed to mechanical unloading prior [23]. We also observed decreased phosphorylation levels of acetyl-CoA carboxylase (ACC), a marker of AMPK activity, in rat soleus-derived myoblasts at later stages (5 days) of differentiation [24]. It is also known that denervation-induced muscle atrophy can increase susceptibility of SCs to apoptosis [8,25]. Since we recently found a decrease in AMPK activity during enhanced differentiation of primary myoblasts derived from atrophic rat soleus muscle [24], we hypothesized that there may be a potential link between AMPK activity and susceptibility of differentiating myoblasts to apoptosis. Hence, using AICAR, a specific AMPK activator, we aimed to estimate the effect of AMPK activation on apoptosis in differentiating myoblasts derived from atrophied rat soleus muscle. ## 2.1. Experimental Design A widely recognized Morey-Holton hindlimb suspension (HS) rodent model was used to induce mechanical unloading [26]. Male Wistar rats (3 months, 180 ± 10) were kept under standard laboratory conditions (room temperature about 21 °C and 12:12 h light/dark cycle) with free access to food and water. The rats were divided into three groups ($$n = 10$$/group): [1] control (C); [2] hindlimb suspension (HS); and [3] hindlimb suspension + AICAR (HS + AICAR). Upon completion of the HS experiment, soleus muscles from both hindlimbs were collected and subsequently used for isolation of muscle stem/progenitor cells. Euthanasia of animals was performed by decapitation under isoflurane anesthesia. Isolation of SCs/myoblasts from the soleus muscle was performed as described in our previous paper [23]. Following isolation, more than $90\%$ of the isolated cells expressed Pax7 (Figure 1b). After obtaining the pure primary myoblast culture, the cells were cultured in growth medium under a humidified atmosphere with $5\%$ CO2 at 37 °C. Two or three days later, when cells reached $80\%$ confluency, myogenic differentiation was induced by changing the media to differentiation media (DMEM medium supplemented with 4.5 g/L D-glucose, L-glutamine, penicillin–streptomycin, and $2\%$ of horse serum). Cells from the HS + AICAR group were incubated with differentiation media containing 1 mM AICAR (cat. ab120358, Abcam, Cambridge, UK) from day 3 to day 5 of differentiation (Figure 1a). All measurements in the study were performed on the 5th day of myogenic differentiation. ## 2.2. In Situ Detection of Apoptotic Cells The detection of DNA double-strand breaks in differentiating myoblasts was performed using the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) technique, as described in Sancilio et al., [ 2022] [27]. Myotubes were fixed for 30 min in $4\%$ paraformaldehyde at RT. They were then rinsed in PBS and incubated in a permeabilizing solution ($0.1\%$ Triton X-100 and $0.1\%$ sodium citrate) for 2 min on ice. DNA strand breaks were determined using the In Situ Cell Death Detection Kit (cat. 11684795910, Roche, Basilea, Switzerland) according to the protocol supplied by the manufacturer. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (cat. D1306, Molecular Probes, Waltham, MA, USA). Fluoromount™ Aqueous Mounting Medium (cat. F4680, Sigma-Aldrich, Saint Louis, MA, USA) was used for mounting coverslips on slides. The Olympus inverted fluorescent microscope (20× magnification) and Cell Sens Dimension Software 3.2 (Build 23706) (Olympus, Tokyo, Japan) were used to acquire and analyse microscopic images. ## 2.3. Immunocytochemistry for Pax7 Detection Detection of Pax7 in rat satellite cells was performed as described in our previous study [23]. ## 2.4. Gene Expression Analysis mRNA expression of target genes was determined by reverse transcription polymerase chain reaction (RT-PCR) as described in our previous study [23]. Primer sequences are provided in Table 1. Gapdh and Ywhaz were used as reference genes. ## 2.5. Western Blot Analysis Western blot analysis was performed as described in our previous studies [28,29]. Primary antibodies used in the study were as follows: p-AMPK (Thr172) (1:500, cat. # Y408289, ABM, Richmond, BC, Canada), t-AMPK (1:1000, cat. # 2523, Cell Signaling Technology, Danvers, MA, USA), p-ACC (S79) (1:1000, cat. # 2535, Cell Signaling Technology, USA), t-ACC (1:1000, cat. # 3662, Cell Signaling Technology, USA), Caspase-3 (1:1000, cat. # 9661, Cell Signaling Technology, USA), p-rpS6 (S$\frac{240}{244}$) (1:2000, cat. # 5364 Cell Signaling Technology, Danvers, MA, USA), rpS6 (1:2000, cat. # 2217, Cell Signaling Technology, Danvers, MA, USA), Bax (1:2000, cat. # ab32503 Abcam, Cambridge, UK) and tubulin (1:3000, cat. # CSB-MA000185 Cusabio Biotech, Wuhan City, China). Horseradish peroxidase-conjugated antibodies to rabbit immunoglobulins (1:60000, cat. # 111-035-003, Jackson Immuno Research, Cambridge, UK) were used as secondary antibodies. Following image capture of phosphorylated proteins, membranes were stripped of the phospho-specific antibodies using RestoreTM Western Blot Stripping Buffer (cat. # 21059, Thermo Fisher Scientific, Waltham, MA, USA). The membranes were then re-probed with primary antibodies for each respective total protein. A total protein staining (Ponceau S) and/or tubulin protein expression were used for normalization of Western blots. ## 2.6. Statistical Analysis Statistical analysis was performed using SigmaPlot 12.5 software. qRT-PCR and Western blot data are shown as mean ± SEM. Two-way ANOVA with post-hoc Tukey test was used to determine the significant differences between group means. Statistical significance was accepted at $p \leq 0.05.$ ## 3.1. Body Weight and Soleus Muscle Weight Seven-day HS induced a slight decrease in rat body weight and a more profound reduction in absolute and normalized soleus muscle weight compared to the control group (Table 2). ## 3.2. Phosphorylation Status of AMPK (Thr172), ACC (Ser79) and rpS6 (Ser 240/244) in Differentiating Myoblasts In the HS myoblasts, there was a significant decrease ($53\%$) in the level of AMPK (Thr172) phosphorylation compared to the control myoblasts (Figure 2a). Treatment of the HS myoblasts with 5-Aminoimidazole-4-carboxamide ribonucleoside (AICAR) (a potent AMPK activator) prevented a decrease in AMPK (Thr172) phosphorylation (Figure 2a). Phosphorylation of ACC on Ser79 was reduced in the HS myoblasts and AICAR prevented this reduced ACC phosphorylation (Figure 2b). Furthermore, phosphorylation status of ribosomal protein S6 (rpS6), a marker of mTORC1 activity, was evaluated. In the HS myoblasts, there was a significant increase ($29\%$) in rpS6 (Ser $\frac{240}{244}$) phosphorylation compared to the control myoblast cultures. Treatment of the HS myoblasts with AICAR abrogated the increased rpS6 (Ser $\frac{240}{244}$) phosphorylation (Figure 2c). ## 3.3. Effect of AICAR Treatment on Morphological Characteristics and Expression of Differentiation Markers in Myoblasts Derived from the Atrophied Rat Soleus Muscle On the 5th day of differentiation, myotubes derived from the atrophied soleus muscle showed an increased fusion index but significantly decreased area, diameter, width and length relative to myotubes derived from the control soleus muscle (Table 3). AICAR treatment of differentiating myoblasts derived from the atrophied muscle reversed changes in fusion index and attenuated or fully prevented morphological alterations in myotubes (Table 3). By the 5th day of myogenic differentiation, RT-PCR analysis also revealed that HS cells exhibit a significant upregulation of genes responsible for myoblast differentiation (myogenin and MyoD) and fusion (Myomaker and Myomixer) compared to the control cells (Table 4). However, treatment of differentiating HS myoblasts with AICAR significantly attenuated the expression of differentiation and fusion markers (Table 4). ## 3.4. The Number of Apoptotic Cells in Myoblast Cultures TUNEL assay revealed that in the HS myoblasts, the number of apoptotic cells was $43\%$ greater than in the control myoblasts (Figure 3). The number of apoptotic cells in the AICAR-treated HS myoblasts did not differ from the control myoblasts (Figure 3). Thus, the maintenance of AMPK activity in differentiating myoblasts derived from the atrophied soleus muscle was able to fully prevent the increased number of TUNEL+ cells. ## 3.5. Expression of Apoptotic Markers in Myoblast Cultures In the HS myoblasts, there was an increase in caspase-9 mRNA expression by $78\%$ (Figure 4a) and p53 mRNA expression by $100\%$ (Figure 4b) compared to the C myoblasts. In the AICAR-treated HS myoblasts, no significant changes in the expression levels of caspase-9 and p53 were found relative to the C myoblasts (Figure 4a,b). In the HS myoblasts, mRNA expression levels of Bcl-2, a protein that inhibits apoptosis, were reduced by $45\%$ compared to the control myoblasts (Figure 4c). The expression levels of Bcl-2 after AICAR treatment of the HS myoblasts did not differ from the control myoblasts (Figure 4c). In the HS cells, we also observed a significant upregulation of pro-apoptotic marker Bax at both mRNA and protein expression levels compared to the control cells (Figure 5). AICAR treatment of HS myoblasts fully prevented this increased Bax mRNA expression and protein abundance (Figure 5). Thus, differentiating myoblasts isolated from the unloaded soleus muscle showed a significant increase in the expression of pro-apoptotic markers (caspase-9, p53, and Bax) and a concomitant decrease in the expression of anti-apoptotic marker (Bcl-2). AICAR-induced prevention of AMPK dephosphorylation in the HS myoblasts resulted in the maintenance of the apoptotic markers’ expression at the control levels. We also evaluated both mRNA expression levels and the content of cleaved caspase-3, a key effector enzyme in apoptosis induction, in differentiating myoblast cultures. As shown in Figure 4a, mRNA expression levels of caspase-3 were significantly upregulated in the HS myoblasts compared to the C cultures. However, AICAR treatment lowered caspase-3 mRNA expression levels below the C values (Figure 6a). The content of cleaved caspase-3 in the HS myoblasts was $53\%$ greater relative to the C myoblasts (Figure 6b). However, in the HS+AICAR myoblasts, the content of cleaved caspase-3 was significantly lower than in the C and HS myoblast cultures (Figure 6b). These data support the findings presented in Figure 3, Figure 4 and Figure 5 about the activation of apoptotic processes in differentiating myoblasts derived from the atrophied rat soleus muscle. ## 4. Discussion Our data demonstrate, for the first time, the direct effects of AICAR treatment (and, hence, the maintenance of AMPK activity) on the apoptosis in differentiating primary myoblasts derived from mechanically unloaded/atrophied rat soleus muscle. Previous studies demonstrated that exposure to mechanical unloading/microgravity can lead to apoptotic processes in skeletal muscles [30]. Radugina et al. [ 2017] have shown that exposure of mice to 30-day microgravity results in the presence of multiple apoptotic nuclei and a smaller number of SCs in quadriceps muscles compared to the control mice [10]. It is known that a certain level of apoptosis normally accompanies myoblast differentiation [31,32]. Enhanced apoptosis during myoblast differentiation can contribute to muscle fiber degeneration, as was revealed in various types of muscular dystrophy and atrophy cases [33,34,35]. In the present study, we found a significant upregulation of apoptosis during differentiation of primary myoblasts isolated from rat soleus after 7-day mechanical unloading. Using TUNEL assay, we identified a greater number of apoptotic cells in the culture of differentiating myoblasts derived from the atrophic soleus muscles compared to the differentiating myoblasts derived from the control soleus muscles. Moreover, the levels of mRNA expression of pro-apoptotic markers (caspase-3 and -9, p53 and Bax) were significantly increased in parallel with reduced mRNA expression of anti-apoptotic Bcl-2. The presence of apoptosis in these HS myoblasts was also confirmed by an increased content of cleaved caspase-3. These results correlate well with some literature data. For instance, Andrianjafiniony et al. [ 2010] have showed a significant increase in the content of caspase-3 and -9 in rat soleus muscle after 14 days of HS [36], which is in agreement with the above mentioned study by Radugina [2017] on the effect of 30-day unloading (microgravity) on murine skeletal muscle [10]. Furthermore, a significant increase in apoptosis was shown in differentiated myotubes derived from skeletal muscles of patients with myotonic dystrophy [37]. We also determined the levels of AMPK activity (by assessing AMPK Thr172 phosphorylation and ACC Ser 79 phosphorylation) since it is known that AMPK can contribute to the regulation of programmed cell death (apoptosis) normally, accompanying myogenesis and muscle regeneration [15,22,38]. Available data on the role of AMPK in the regulation of apoptosis are controversial. While some reports suggest AMPK-dependent stimulation of apoptosis [39,40,41], several studies demonstrate an anti-apoptotic role of AMPK [22,42,43,44]. In the present study, we found a significant decrease in both AMPK (Thr 172) phosphorylation and ACC (Ser 79) phosphorylation in differentiating myoblasts derived from rat soleus muscle after 7-day HS, indicative of a reduction in AMPK kinase activity. Moreover, a decrease in the activity of AMPK signaling was confirmed by a significant increase in Ser $\frac{240}{244}$ phosphorylation of rpS6, a marker of mTORC1 activity, since AMPK is known to be an endogenous inhibitor of mTORC1 and protein synthesis in both skeletal muscles [45] and cultured muscle cells [46,47]. These data are consistent with our previous study showing a decrease in ACC (Ser 79) phosphorylation in primary myoblasts isolated from unloaded skeletal muscle [24]. The results on the activity of AMPK and ACC in differentiating myoblasts derived from soleus muscle after mechanical unloading are in agreement with data previously obtained directly on rat soleus muscle [28,29,48]. However, these unloading-induced changes in AMPK activity were seen in rat soleus muscle at earlier stages of HS (1–3 days) compared to differentiating myoblasts isolated from rat soleus muscle after 7-day HS in the present study. Liu et al. [ 2019] have previously demonstrated, although in non-muscle cells, that AMPK knockdown results in the upregulation of apoptosis [49]. In primary myoblast derived from geriatric skeletal muscle, White and colleagues [2018] have demonstrated that reduced AMPK activity (phosphorylation) is associated with increased apoptosis [44]. Moreover, it has also been shown in differentiating C2C12 myoblasts that an inhibition of AMPK activity contributes to apoptosis upregulation [22]. In the present study, to elucidate the role of AMPK activity in the regulation of apoptosis in differentiating myoblasts derived from the unloaded/atrophied soleus muscle, a specific AMPK activator, AICAR, was used. Incubation of the HS myoblasts with AICAR not only prevented HS-induced reductions in the phosphorylation levels of AMPK and ACC, but also reduced the number of apoptotic cells and maintained the expression of apoptotic markers at the control levels. Thus, our data clearly demonstrate that the maintenance of AMPK activity (Thr 172 phosphorylation) prevents apoptosis development in differentiating myoblasts derived from the rat soleus after 7-day mechanical unloading. It is known that AMPK can participate in the regulation of apoptosis through different mechanisms, including regulation of autophagy [50], Bcl-2-regulated apoptotic pathway [51], mTOR inhibition [52], p53 activation [53] and phosphorylation of cyclin-dependent kinase inhibitor 1B (p27Kip1) [44]. Under conditions of metabolic stress, p27Kip1 is involved in the regulation of cell fate. In particular, p27Kip1 controls cell cycle inhibition, apoptosis and autophagy [38]. It has been shown that p27Kip1 can inhibit the activity of pro-apoptotic protein Bax and prevent apoptosis [54,55]. The regulation of p27Kip1 activity is carried out at the level of transcription, phosphorylation and subcellular localization [44]. It has been demonstrated that nuclear p27Kip1 is able to facilitate quiescence and apoptosis, while cytoplasmic p27Kip1 can promote cell survival and autophagy [44]. Liang and co-workers demonstrated that AMPK-related phosphorylation of p27Kip1 on Thr198 is able to stimulate its sequestration to the cytosol, leading to increased autophagy and decreased apoptosis [38]. In myoblasts derived from aged mice, increased apoptosis was accompanied by decreased AMPK and p27Kip1 phosphorylation [44]. Moreover, upon AMPK activation or p27Kip1 overexpression, apoptosis in these cells was suppressed [44]. We can speculate that in the present study, the development of apoptosis in differentiating myoblasts derived from the unloaded soleus muscle could occur due to reduced AMPK-dependent p27Kip1 phosphorylation, resulting in p27Kip1 nuclear localization and induction of apoptosis. Identification of the precise molecular mechanisms underlying apoptosis development in differentiating myoblasts isolated from atrophied muscles requires further research. In conclusion, our study for the first time revealed the increase in apoptotic processes in differentiating myoblasts derived from the atrophied rat soleus muscle. Further, our data provide the first evidence that the activation of apoptosis in such differentiating myoblasts is associated, at least in part, with reduced AMPK activity. ## References 1. Yin H., Price F., Rudnicki M.A.. Satellite cells and the muscle stem cell niche. Physiol. Rev. (2013) 93 23-67. DOI: 10.1152/physrev.00043.2011 2. Dumont N.A., Bentzinger C.F., Sincennes M.C., Rudnicki M.A.. Satellite Cells and Skeletal Muscle Regeneration. Compr. Physiol. (2015) 5 1027-1059. DOI: 10.1002/cphy.c140068 3. Shenkman B.S., Turtikova O.V., Nemirovskaya T.L., Grigoriev A.I.. Skeletal muscle activity and the fate of myonuclei. Acta Nat. (2010) 2 59-66. DOI: 10.32607/20758251-2010-2-2-59-65 4. Nakanishi R., Hirayama Y., Tanaka M., Maeshige N., Kondo H., Ishihara A., Roy R.R., Fujino H.. Nucleoprotein supplementation enhances the recovery of rat soleus mass with reloading after hindlimb unloading-induced atrophy via myonuclei accretion and increased protein synthesis. Nutr. Res. (2016) 36 1335-1344. DOI: 10.1016/j.nutres.2016.10.007 5. Wang X.D., Kawano F., Matsuoka Y., Fukunaga K., Terada M., Sudoh M., Ishihara A., Ohira Y.. Mechanical load-dependent regulation of satellite cell and fiber size in rat soleus muscle. Am. J. Physiol. Cell Physiol. (2006) 290 C981-C989. DOI: 10.1152/ajpcell.00298.2005 6. Mitchell P.O., Pavlath G.K.. Skeletal muscle atrophy leads to loss and dysfunction of muscle precursor cells. Am. J. Physiol. Cell Physiol. (2004) 287 C1753-C1762. DOI: 10.1152/ajpcell.00292.2004 7. Stewart C.E., Newcomb P.V., Holly J.M.. Multifaceted roles of TNF-alpha in myoblast destruction: A multitude of signal transduction pathways. J. Cell. Physiol. (2004) 198 237-247. DOI: 10.1002/jcp.10387 8. Jejurikar S.S., Kuzon W.M.. Satellite cell depletion in degenerative skeletal muscle. Apoptosis (2003) 8 573-578. DOI: 10.1023/A:1026127307457 9. Matsuba Y., Goto K., Morioka S., Naito T., Akema T., Hashimoto N., Sugiura T., Ohira Y., Beppu M., Yoshioka T.. Gravitational unloading inhibits the regenerative potential of atrophied soleus muscle in mice. Acta Physiol. (2009) 196 329-339. DOI: 10.1111/j.1748-1716.2008.01943.x 10. Radugina E.A., Almeida E.A.C., Blaber E., Poplinskaya V.A., Markitantova Y.V., Grigoryan E.N.. Exposure to microgravity for 30 days onboard Bion M1 caused muscle atrophy and impaired regeneration in murine femoral Quadriceps. Life Sci. Space Res. (2018) 16 18-25. DOI: 10.1016/j.lssr.2017.08.005 11. Ferreira R., Neuparth M.J., Ascensao A., Magalhaes J., Vitorino R., Duarte J.A., Amado F.. Skeletal muscle atrophy increases cell proliferation in mice gastrocnemius during the first week of hindlimb suspension. Eur. J. Appl. Physiol. (2006) 97 340-346. DOI: 10.1007/s00421-006-0197-6 12. Guitart M., Lloreta J., Manas-Garcia L., Barreiro E.. Muscle regeneration potential and satellite cell activation profile during recovery following hindlimb immobilization in mice. J. Cell. Physiol. (2018) 233 4360-4372. DOI: 10.1002/jcp.26282 13. Hardie D.G.. New roles for the LKB1→AMPK pathway. Curr. Opin. Cell Biol. (2005) 17 167-173. DOI: 10.1016/j.ceb.2005.01.006 14. Mounier R., Lantier L., Leclerc J., Sotiropoulos A., Pende M., Daegelen D., Sakamoto K., Foretz M., Viollet B.. Important role for AMPKalpha1 in limiting skeletal muscle cell hypertrophy. FASEB J. Off. Public Fed. Am. Soc. Exp. Biol. (2009) 23 2264-2273. DOI: 10.1096/fj.08-119057 15. Villanueva-Paz M., Cotan D., Garrido-Maraver J., Oropesa-Avila M., de la Mata M., Delgado-Pavon A., de Lavera I., Alcocer-Gomez E., Alvarez-Cordoba M., Sanchez-Alcazar J.A.. AMPK Regulation of Cell Growth, Apoptosis, Autophagy, and Bioenergetics. Exp. Suppl. (2016) 107 45-71. DOI: 10.1007/978-3-319-43589-3_3 16. Thomson D.M.. The Role of AMPK in the Regulation of Skeletal Muscle Size, Hypertrophy, and Regeneration. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19103125 17. Fu X., Zhu M., Zhang S., Foretz M., Viollet B., Du M.. Obesity Impairs Skeletal Muscle Regeneration Through Inhibition of AMPK. Diabetes (2016) 65 188-200. DOI: 10.2337/db15-0647 18. Fu X., Zhu M.J., Dodson M.V., Du M.. AMP-activated protein kinase stimulates Warburg-like glycolysis and activation of satellite cells during muscle regeneration. J. Biol. Chem. (2015) 290 26445-26456. DOI: 10.1074/jbc.M115.665232 19. Fu X., Zhao J.X., Zhu M.J., Foretz M., Viollet B., Dodson M.V., Du M.. AMP-activated protein kinase alpha1 but not alpha2 catalytic subunit potentiates myogenin expression and myogenesis. Mol. Cell. Biol. (2013) 33 4517-4525. DOI: 10.1128/MCB.01078-13 20. Williamson D.L., Butler D.C., Alway S.E.. AMPK inhibits myoblast differentiation through a PGC-1alpha-dependent mechanism. Am. J. Physiol. Endocrinol. Metab. (2009) 297 E304-E314. DOI: 10.1152/ajpendo.91007.2008 21. Fulco M., Cen Y., Zhao P., Hoffman E.P., McBurney M.W., Sauve A.A., Sartorelli V.. Glucose restriction inhibits skeletal myoblast differentiation by activating SIRT1 through AMPK-mediated regulation of Nampt. Dev. Cell (2008) 14 661-673. DOI: 10.1016/j.devcel.2008.02.004 22. Niesler C.U., Myburgh K.H., Moore F.. The changing AMPK expression profile in differentiating mouse skeletal muscle myoblast cells helps confer increasing resistance to apoptosis. Exp. Physiol. (2007) 92 207-217. DOI: 10.1113/expphysiol.2006.034736 23. Komarova M.Y., Rozhkov S.V., Ivanova O.A., Turtikova O.V., Mirzoev T.M., Dmitrieva R.I., Shenkman B.S., Vilchinskaya N.A.. Cultured Myoblasts Derived from Rat Soleus Muscle Show Altered Regulation of Proliferation and Myogenesis during the Course of Mechanical Unloading. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23169150 24. Vilchinskaya N.A., Rozhkov S.V., Komarova M.Y., Dmitrieva R.I., Shenkman B.S.. Effect of simulated gravitational unloading on m. Soleus satellite cells. Aviakosmicheskaya I Ekol. Meditsina (2022) 56 20-29. DOI: 10.21687/0233-528X-2022-56-2-20-29 25. Jejurikar S.S., Marcelo C.L., Kuzon W.M.. Skeletal muscle denervation increases satellite cell susceptibility to apoptosis. Plast. Reconstr. Surg. (2002) 110 160-168. DOI: 10.1097/00006534-200207000-00027 26. Morey-Holton E.R., Globus R.K.. Hindlimb unloading rodent model: Technical aspects. J. Appl. Physiol. (2002) 92 1367-1377. DOI: 10.1152/japplphysiol.00969.2001 27. Sancilio S., Nobilio S., Ruggiero A.G., Filippo E.S.D., Stati G., Fulle S., Bellomo R.G., Saggini R., Pietro R.D.. Effects of Focused Vibrations on Human Satellite Cells. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23116026 28. Vilchinskaya N.A., Mochalova E.P., Nemirovskaya T.L., Mirzoev T.M., Turtikova O.V., Shenkman B.S.. Rapid decline in MyHC I(beta) mRNA expression in rat soleus during hindlimb unloading is associated with AMPK dephosphorylation. J. Physiol. (2017) 595 7123-7134. DOI: 10.1113/JP275184 29. Mirzoev T., Tyganov S., Vilchinskaya N., Lomonosova Y., Shenkman B.. Key Markers of mTORC1-Dependent and mTORC1-Independent Signaling Pathways Regulating Protein Synthesis in Rat Soleus Muscle During Early Stages of Hindlimb Unloading. Cell. Physiol. Biochem. Int. J. Exp. Cell. Physiol. Biochem. Pharmacol. (2016) 39 1011-1020. DOI: 10.1159/000447808 30. Smith H.K., Maxwell L., Martyn J.A., Bass J.J.. Nuclear DNA fragmentation and morphological alterations in adult rabbit skeletal muscle after short-term immobilization. Cell Tissue Res. (2000) 302 235-241. DOI: 10.1007/s004410000280 31. Sandri M., Cantini M., Massimino M.L., Geromel V., Arslan P.. Myoblasts and myotubes in primary cultures deprived of growth factors undergo apoptosis. Basic Appl. Myol. (1996) 6 257-260 32. Wang J., Walsh K.. Resistance to apoptosis conferred by Cdk inhibitors during myocyte differentiation. Science (1996) 273 359-361. DOI: 10.1126/science.273.5273.359 33. Tews D.S., Goebel H.H.. DNA fragmentation and BCL-2 expression in infantile spinal muscular atrophy. Neuromuscul. Disord. (1996) 6 265-273. DOI: 10.1016/0960-8966(96)00018-1 34. Tews D.S., Goebel H.H.. DNA-fragmentation and expression of apoptosis-related proteins in muscular dystrophies. Neuropathol. Appl. Neurobiol. (1997) 23 331-338. DOI: 10.1111/j.1365-2990.1997.tb01304.x 35. Tidball J.G., Albrecht D.E., Lokensgard B.E., Spencer M.J.. Apoptosis precedes necrosis of dystrophin-deficient muscle. J. Cell Sci. (1995) 108 2197-2204. DOI: 10.1242/jcs.108.6.2197 36. Andrianjafiniony T., Dupre-Aucouturier S., Letexier D., Couchoux H., Desplanches D.. Oxidative stress, apoptosis, and proteolysis in skeletal muscle repair after unloading. Am. J. Physiol. Cell Physiol. (2010) 299 C307-C315. DOI: 10.1152/ajpcell.00069.2010 37. Loro E., Rinaldi F., Malena A., Masiero E., Novelli G., Angelini C., Romeo V., Sandri M., Botta A., Vergani L.. Normal myogenesis and increased apoptosis in myotonic dystrophy type-1 muscle cells. Cell Death Differ. (2010) 17 1315-1324. DOI: 10.1038/cdd.2010.33 38. Liang J., Shao S.H., Xu Z.X., Hennessy B., Ding Z., Larrea M., Kondo S., Dumont D.J., Gutterman J.U., Walker C.L.. The energy sensing LKB1-AMPK pathway regulates p27(kip1) phosphorylation mediating the decision to enter autophagy or apoptosis. Nat. Cell Biol. (2007) 9 218-224. DOI: 10.1038/ncb1537 39. Kefas B.A., Heimberg H., Vaulont S., Meisse D., Hue L., Pipeleers D., Van de Casteele M.. AICA-riboside induces apoptosis of pancreatic beta cells through stimulation of AMP-activated protein kinase. Diabetologia (2003) 46 250-254. DOI: 10.1007/s00125-002-1030-3 40. Kilbride S.M., Farrelly A.M., Bonner C., Ward M.W., Nyhan K.C., Concannon C.G., Wollheim C.B., Byrne M.M., Prehn J.H.. AMP-activated protein kinase mediates apoptosis in response to bioenergetic stress through activation of the pro-apoptotic Bcl-2 homology domain-3-only protein BMF. J. Biol. Chem. (2010) 285 36199-36206. DOI: 10.1074/jbc.M110.138107 41. Morishita M., Kawamoto T., Hara H., Onishi Y., Ueha T., Minoda M., Katayama E., Takemori T., Fukase N., Kurosaka M.. AICAR induces mitochondrial apoptosis in human osteosarcoma cells through an AMPK-dependent pathway. Int. J. Oncol. (2017) 50 23-30. DOI: 10.3892/ijo.2016.3775 42. Kewalramani G., Puthanveetil P., Wang F., Kim M.S., Deppe S., Abrahani A., Luciani D.S., Johnson J.D., Rodrigues B.. AMP-activated protein kinase confers protection against TNF-alpha-induced cardiac cell death. Cardiovasc. Res. (2009) 84 42-53. DOI: 10.1093/cvr/cvp166 43. Zhang X.Z., Sun Y., Zhang M.H., Jin Z.. Insulin-like growth factor-1 inhibits apoptosis of rat gastric smooth muscle cells under high glucose condition via adenosine monophosphate-activated protein kinase (AMPK) pathway. Folia Histochem. Cytobiol. (2022) 60 74-88. DOI: 10.5603/FHC.a2022.0005 44. White J.P., Billin A.N., Campbell M.E., Russell A.J., Huffman K.M., Kraus W.E.. The AMPK/p27(Kip1) Axis Regulates Autophagy/Apoptosis Decisions in Aged Skeletal Muscle Stem Cells. Stem Cell Rep. (2018) 11 425-439. DOI: 10.1016/j.stemcr.2018.06.014 45. Bolster D.R., Crozier S.J., Kimball S.R., Jefferson L.S.. AMP-activated protein kinase suppresses protein synthesis in rat skeletal muscle through down-regulated mammalian target of rapamycin (mTOR) signaling. J. Biol. Chem. (2002) 277 23977-23980. DOI: 10.1074/jbc.C200171200 46. Williamson D.L., Bolster D.R., Kimball S.R., Jefferson L.S.. Time course changes in signaling pathways and protein synthesis in C2C12 myotubes following AMPK activation by AICAR. Am. J. Physiol. Endocrinol. Metab. (2006) 291 E80-E89. DOI: 10.1152/ajpendo.00566.2005 47. Nakashima K., Ishida A.. AMP-activated Protein Kinase Activation Suppresses Protein Synthesis and mTORC1 Signaling in Chick Myotube Cultures. J. Poult. Sci. (2022) 59 81-85. DOI: 10.2141/jpsa.0210021 48. Belova S.P., Vilchinskaya N.A., Mochalova E.P., Mirzoev T.M., Nemirovskaya T.L., Shenkman B.S.. Elevated p70S6K phosphorylation in rat soleus muscle during the early stage of unloading: Causes and consequences. Arch. Biochem. Biophys. (2019) 674 108105. DOI: 10.1016/j.abb.2019.108105 49. Liu J., Long S., Wang H., Liu N., Zhang C., Zhang L., Zhang Y.. Blocking AMPK/ULK1-dependent autophagy promoted apoptosis and suppressed colon cancer growth. Cancer Cell Int. (2019) 19 336. DOI: 10.1186/s12935-019-1054-0 50. Kim J., Kundu M., Viollet B., Guan K.L.. AMPK and mTOR regulate autophagy through direct phosphorylation of Ulk1. Nat. Cell Biol. (2011) 13 132-141. DOI: 10.1038/ncb2152 51. Concannon C.G., Tuffy L.P., Weisova P., Bonner H.P., Davila D., Bonner C., Devocelle M.C., Strasser A., Ward M.W., Prehn J.H.. AMP kinase-mediated activation of the BH3-only protein Bim couples energy depletion to stress-induced apoptosis. J. Cell Biol. (2010) 189 83-94. DOI: 10.1083/jcb.200909166 52. Faivre S., Kroemer G., Raymond E.. Current development of mTOR inhibitors as anticancer agents. Nat. Rev. Drug Discov. (2006) 5 671-688. DOI: 10.1038/nrd2062 53. Jones R.G., Plas D.R., Kubek S., Buzzai M., Mu J., Xu Y., Birnbaum M.J., Thompson C.B.. AMP-activated protein kinase induces a p53-dependent metabolic checkpoint. Mol. Cell (2005) 18 283-293. DOI: 10.1016/j.molcel.2005.03.027 54. Gil-Gomez G., Berns A., Brady H.J.. A link between cell cycle and cell death: Bax and Bcl-2 modulate Cdk2 activation during thymocyte apoptosis. EMBO J. (1998) 17 7209-7218. DOI: 10.1093/emboj/17.24.7209 55. Hiromura K., Pippin J.W., Fero M.L., Roberts J.M., Shankland S.J.. Modulation of apoptosis by the cyclin-dependent kinase inhibitor p27(Kip1). J. Clin. Investig. (1999) 103 597-604. DOI: 10.1172/JCI5461
--- title: Prediction of Fatty Liver Disease in a Chinese Population Using Machine-Learning Algorithms authors: - Shuwei Weng - Die Hu - Jin Chen - Yanyi Yang - Daoquan Peng journal: Diagnostics year: 2023 pmcid: PMC10047083 doi: 10.3390/diagnostics13061168 license: CC BY 4.0 --- # Prediction of Fatty Liver Disease in a Chinese Population Using Machine-Learning Algorithms ## Abstract Background: Fatty liver disease (FLD) is an important risk factor for liver cancer and cardiovascular disease and can lead to significant social and economic burden. However, there is currently no nationwide epidemiological survey for FLD in China, making early FLD screening crucial for the Chinese population. Unfortunately, liver biopsy and abdominal ultrasound, the preferred methods for FLD diagnosis, are not practical for primary medical institutions. Therefore, the aim of this study was to develop machine learning (ML) models for screening individuals at high risk of FLD, and to provide a new perspective on early FLD diagnosis. Methods: This study included a total of 30,574 individuals between the ages of 18 and 70 who completed abdominal ultrasound and the related clinical examinations. Among them, 3474 individuals were diagnosed with FLD by abdominal ultrasound. We used 11 indicators to build eight classification models to predict FLD. The model prediction ability was evaluated by the area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, and kappa value. Feature importance analysis was assessed by Shapley value or root mean square error loss after permutations. Results: Among the eight ML models, the prediction accuracy of the extreme gradient boosting (XGBoost) model was highest at $89.77\%$. By feature importance analysis, we found that the body mass index, triglyceride, and alanine aminotransferase play important roles in FLD prediction. Conclusion: XGBoost improves the efficiency and cost of large-scale FLD screening. ## 1. Introduction Fatty liver disease (FLD), a global epidemic disease, is an important risk factor for liver cancer [1,2]. However, the harm is not limited to the liver itself. Some studies have shown that fatty liver can significantly increase the incidence of fatal and non-fatal cardiovascular events [3], and FLD patients are more likely to be associated with obesity [4], hyperlipidemia [5], hypertension, and type 2 diabetes [6] than healthy people. Furthermore, as fatty liver disease progresses, the risk of CKD significantly increases with the degree of liver fibrosis [7]. The fatty liver disease (FLD) guideline for the Asian population [8] highlights that there is no national epidemiological survey on FLD in China, and the reported studies are mostly from economically developed regions, which may lead to some degree of bias in the epidemiological characteristics. Therefore, early FLD screening is not only necessary to reduce the socioeconomic burden of FLD, but also to improve the epidemiological investigation of FLD in China. For FLD, liver biopsy is undoubtedly the “gold standard” for diagnosis [9]. However, as a screening method, its high cost and invasive nature do not make it the first choice for fatty liver screening. Analogously, ultrasound, as an effective diagnostic method, relies on the operation and judgment of the ultrasound doctor. Therefore, an urgent need exists to develop cost-saving and non-invasive methods to screen fatty liver. As a prediction tool, machine learning (ML) represents the latest development of statistics. Unlike the traditional statistical model, which depends on certain assumptions for data and a clear mathematical form, ML does not have any assumptions about the data, and the results eliminate the classical statistical framework based on hypothesis testing. The prediction efficiency of ML models based on algorithms or programs is high, and the results of cross-validation are easy to understand, so ML can be widely used in medical diagnosis trials today. Among them, several ML methods, such as random forest (RF), artificial neural network (ANN), K-nearest neighbor (KNN), and support vector machine (SVM) have played an important role in the prediction of many diseases. Previous studies [10] have extracted the gray-value distribution features of children’s liver ultrasound images in a given region of interest, and then constructed an ML discriminant model of liver lesions by a variety of maximum-likelihood classification methods. The prediction accuracy is better than the traditional liver and kidney index and liver echo intensity attenuation index. Acharya et al. [ 11] extracted abdominal ultrasound features with the curvelet transform method, reduced features through locality-sensitive discriminant analysis, and used a probabilistic neural network classifier based on only six features to distinguish the normal liver, fatty liver, and liver cirrhosis with an accuracy of $97.33\%$, specificity of $100\%$, and sensitivity of $96\%$. Based on in-depth ML approaches to diagnosing FLD, there are many artificial methods that can diagnose fatty liver with high accuracy. However, most of these diagnoses are based on abdominal ultrasound or computed tomography (CT) images, which are costly for fatty liver screening. Compared with the image-based ML screening methods, using physical examination data and blood biochemical indexes as predictive indicators can screen fatty liver in an efficient and economical way. Thus, the main purpose of this study was to build an efficient and robust FLD screening ML model based on these indicators. ## 2.1. Study Data The dataset used in this study was provided by the health management center of the Second Xiangya Hospital of Central South University, Changsha, China, and included the data of 36,527 patients. We enrolled individuals aged 18–70 years from January 2013 to December 2019. During the process of data collection, no privacy information was included. Only 23 indexes, including physical examination data, age, and blood biochemistry indexes were included. Because ML models rely on data integrity, 4908 individuals with missing values of more than $30\%$ were excluded, and the remaining missing values were completed by the multiple interpolation method. In this study, the diagnosis of FLD was based on the results of abdominal ultrasound images. The ultrasound machines used in this study were the Philips Medical Systems model iU22 and model Epiq (Philips Ultrasound, Bothell, WA, USA). All the diagnostic results of abdominal ultrasounds were performed by attending physicians in our hospital’s imaging center and were reviewed by senior physicians. The diagnostic criteria for FLD were based on the guidelines published by the Chinese Medical Association in 2010 [12]. The diagnosis of FLD was confirmed if at least two of the three following findings were present: diffuse echogenicity enhancement of the liver parenchyma in the near field, stronger than that of the kidney; poor visualization of intrahepatic duct structures; and the gradual attenuation of liver echogenicity in the far field. Therefore, we excluded 1045 individuals who had not yet completed the screening process. Finally, of the 30,574 individuals remaining in the study, 3474 of them were diagnosed with FLD by abdominal ultrasound (Figure 1). ## 2.2. Data Processing In machine learning data preprocessing, class imbalance is a common issue where the number of samples in one category is much larger than in the other. This can present challenges for machine learning models, and the severity of the imbalance depends on the proportion of samples in each category. To address this problem, this study used a synthetic minority over-sampling technique nominal continuous (SMOTE-NC) [13] to handle unbalanced class data. SMOTE-NC is an extension of the synthetic minority over-sampling technique (SMOTE) that can handle nominal and continuous features. In an imbalanced dataset, SMOTE-NC generates synthetic samples for the minority class by oversampling the existing samples using interpolation. When generating synthetic samples, SMOTE-NC takes into account both continuous and nominal features and ensures that the synthetic samples are representative of the underlying data distribution. Feature selection is a crucial aspect of classification tasks, as it can significantly impact the performance of the model. The main objective of feature selection is to identify the most relevant subset of features that can improve the accuracy of classification. In the context of the 11 characteristic variables presented in Table 1, the selection of these variables was based on several factors. These included identifying the most commonly used variables for predicting FLD, adding additional variables to increase the variety of features, and using a stepwise backward selection method based on Akaike Information Criterion (AIC) [14]. AIC is a powerful tool for assessing the performance of a model in terms of both its predictive accuracy and its complexity. It is founded on the concept of entropy, which enables it to capture the trade-off between these two competing factors. By comparing the AIC values of different models, researchers can identify the most effective and parsimonious one for their purposes. In order to filter features and avoid multicollinearity, this study utilized the reverse stepwise-regression algorithm based on AIC. This involved introducing all variables into an equation, and then iteratively deleting the variable that maximized the AIC value until the minimum AIC value was reached. The resulting variables and their corresponding AIC values are shown in Table 2. To automate the feature selection process, a R program was used to compare the AIC values of candidate variables and include those that contributed to the model. The variables listed in Table 2 reflect the remaining variables that were found to contribute to the machine learning model, and the corresponding AIC value indicates the change in AIC value after adding each variable. We used the createDataPartition function in the caret package to divide the training set and the test set, in which the test set accounted for $70\%$ of all data. Continuous variables were normalized by subtracting the average value and dividing by the standard deviation. To overcome the imbalance problem in the training set, we used the synthetic minority oversampling technique to randomly generate new individuals of the minority, which have similar features to the original individuals of the minority class. ## 2.3. Establishment of the Model We used the eight most common classifiers to build an FLD screening model. As one of the most commonly used generalized linear regression models for binary data, logistic regression (LR) can not only provide prediction results, but also indicate the weight of each independent variable in the prediction. RF is a classifier that integrates multiple decision trees. All decision trees are independent of each other, and each decision tree splits the maximum information gain, and finally outputs the results of classification after reaching the threshold; the RF results are based on the majority of all decision trees. As a common two-classification model, the SVM maps the feature vector to the space, and finds the separation hyperplane with the largest interval in the feature space. This approach makes the classification results more robust and improves the generalization ability of the model. Linear discriminant analysis (LDA) is a classical supervised learning method based on data dimensionality reduction, which classifies the data by projecting the data from a high-dimensional space to a lower-dimensional space and ensuring that the intra-class variance of each class is small and that the mean difference between classes is large. Quantitative descriptive analysis (QDA) is a variant of LDA that allows the nonlinear separation of data. KNN is regarded as a nonparametric, lazy algorithm model that is based on adjacent samples with the minimum Euclidean distance. This model makes no assumptions about the data, and there is no clear training data process. Because it assigns the same weight to different features, the model is easily affected by noise. Extreme gradient boosting (XGBoost) is an improved boosting algorithm based on the gradient boosted decision tree (GBDT) method. Unlike with classical GBDT, second-order Taylor expansion is used in XGBoost on the error part of the loss function, which improves the accuracy of the loss function definition. Because of the L2 regularization in the cost function, the complexity of the XGBoost model is controlled, which greatly reduces the possibility of overfitting. Because of these characteristics, XGBoost has excellent classification and regression prediction performance. An artificial neural network (ANN) is a black box model constructed by simulating the brain’s neural structure, and is generally composed of an input layer, hidden layer, and output layer. Each layer may contain multiple neurons. The number of neurons in the input layer depends on the input parameters, and the number of neurons in the other layers is adjusted according to the actual situation. In this model, the input parameters are connected to the neuron on the basis of a certain weight. The activation threshold of the neuron is determined by setting the activation function; then, the signal is further transmitted in the network. The neural network can achieve self-learning through forward propagation or back propagation, and gradually optimizing the weights and deviation values in the process until the value of the loss function tends to be stable and reaches the expected value. Finally, the network generates the results through the output layer. In this study, all the parameters in the models were adjusted by cyclic traversal, and the highest area under the curve (AUC) value was regarded as the selection standard of the model parameters (Figure 2). A 10-fold cross validation was carried out to estimate the performance of each model. ## 2.4. Model Performance Assessment We verified the predictive ability of the ML model by calculating the area under the curve (AUC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and kappa value. In this section, we provide an overview of the various metrics used to evaluate the performance of machine learning models. Accuracy is a crucial evaluation metric in machine learning and represents the proportion of correctly predicted samples in the overall sample. A higher accuracy indicates better classification performance. The formula used for calculating accuracy is as follows:Accuracy = (TP + TN)/(TP + FP + TN + FN) × $100\%$ where TP, TN, FP, and FN represent true positive, true negative, false positive, and false negative, respectively. Sensitivity, as known by the true positive rate, is a critical performance metric for machine learning models, as it quantifies the model’s ability to accurately identify patients who test positive. It measures the proportion of actual positive cases that the model correctly identifies as positive, providing insight into the model’s ability to detect true positives. Sensitivity = TP/(TP + FN) × $100\%$ Specificity refers to the proportion of negative cases identified out of all negative cases. It is a measure of the ability of a machine learning model to correctly identify negative cases. The higher the specificity, the lower the false positive rate, and the more accurate the model’s negative predictions. Specificity = TN/(TN + FP) × $100\%$ Positive predictive value (PPV) is a performance metric in machine learning that measures the proportion of true positive predictions made by the model among all positive predictions. In other words, PPV represents the probability that a positive prediction is actually correct. PPV = TP/(TP + FP) × $100\%$ Negative predictive value (NPV) is another performance metric in machine learning that measures the proportion of true negative predictions made by the model among all negative predictions. NPV represents the probability that a negative prediction is actually correct. NPV = TN/(TN + FN) × $100\%$ Kappa is a statistical measure of agreement that takes values between −1 and 1. In the context of the classification problem under study, it indicates the degree of agreement between the model’s predicted results and the actual classification results. As a rule of thumb, a higher kappa value is often regarded as indicative of stronger agreement between the classifier and the actual results. Kappa = (Po − Pe)/(1 − Pe) where Po is the observed proportion of agreement between the two classifiers, and Pe is the expected proportion of agreement due to chance. Receiver operating characteristic (ROC) curves are a graphical representation of the performance of a binary classifier system. The ROC curve is created by plotting the true positive rate (TPR) against the false positive rate (FPR) at various threshold settings. A perfect classifier has an ROC curve that passes through the top left corner of the plot, indicating a high TPR and low FPR. AUC is a measure of the classifier’s ability to distinguish between positive and negative classes, with an AUC of 1.0 indicating perfect classification and an AUC of 0.5 indicating random guessing. The Shapley value is a concept from cooperative game theory that measures the marginal contribution of each player to a cooperative game. For machine learning, the Shapley value represents the contribution of a feature to a prediction by considering all possible combinations of features that could have been used in the model. It measures the average change in the model’s output when a feature is added, compared to when the feature is not included. The Shapley value can help identify the most important features for a particular prediction and to understand how the model makes decisions. While the SHAP library’s kernel Explainer is capable of computing Shapley values for any machine learning model, it is computationally inefficient for KNN models. Although the k-means clustering algorithm can be used to summarize the data and improve computational efficiency, this comes at the expense of the model’s accuracy. To address this, we evaluated the feature importance of the KNN model using root mean square error loss after permutation in this study. ## 3.1. Features of Participants A total of 30,574 participants were finally included in this study, including 14,250 males and 16,324 females. Among the participants, 3474 participants were diagnosed with fatty liver; their average age was 45.7 ± 10.9 years, whereas the age of the individuals without FLD was 41.7 ± 12.0 years. All variables we included are shown as mean (SD), and the two-sample t-test showed a significant difference between the two groups (Table 3). The density distribution curves of all included characteristic variables are shown in Figure 3. ## 3.2. Model Performance The ROC curves for all models are shown in Figure 3. After comparing the selected ML models, we found the following: [1] The prediction accuracy of XGBoost for FLD was $89.7\%$, and it had high AUC, sensitivity, and specificity. [ 2] The prediction ability of SVM was closest to that of XGBoost, and its sensitivity was better than that of XGBoost, suggesting that SVM is also one of the best prediction models. [ 3] The kappa values of XGBoost and SVM were both higher than $70\%$, which indicates their good repeatability. [ 4] Although the accuracy of RF and KNN was close to $75\%$, their positive predictive values and kappa values were low, suggesting that the models based on these two algorithms have low positive prediction efficiency and poor repeatability (Table 4). According to the results of feature importance analysis, we found that BMI, ALT, and Tg play important roles in all models (Figure 4). ## 4. Discussion This study included the clinical data of 30,572 subjects and 8 ML models, which makes it by far the largest machine learning study based on physical examination and blood biochemical indicators to predict fatty liver in the Chinese population. According to the results, it is not difficult to find that ML can efficiently predict the occurrence of FLD, and the XGBoost model is the best predictor among all the analyzed models. This is likely due to the XGBoost model’s ability to adaptively adjust the depth of trees and weights of leaf nodes to minimize the loss function, as well as mitigate overfitting issues by incorporating regularization terms. The XGBoost model can handle datasets with a large number of features and samples, and identify key factors through feature importance evaluation, thereby improving the model’s interpretability and reliability. These advantages make the XGBoost model highly accurate in binary classification predictions and perform exceptionally well in the diagnosis of many clinical conditions [15,16,17]. Patients with fatty liver disease frequently exhibit metabolic syndrome characteristics, such as overweight, insulin resistance, and atherogenic dyslipidemia that are characterized by elevated plasma triglyceride concentrations. Research indicates that the prevalence of NAFLD among obese patients is as high as $80\%$, compared to $16\%$ in individuals with a normal body mass index and no metabolic risk factors [18,19]. Dyslipidemia in patients with FLD is primarily characterized by hypertriglyceridemia due to large very-low-density lipoprotein particles, increased levels of small and dense low-density lipoprotein particles, and reduced high-density cholesterol levels [20,21]. This alteration in lipid profile can be attributed to heightened cholesteryl ester transfer protein activity [22]. During the natural progression of fatty liver disease, liver enzyme levels also fluctuate, and around $20\%$ of patients with NAFLD have substantial changes in liver enzyme levels, with aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels remaining within the normal range or being modestly elevated (1.5–2 times the upper limit of normal) [23]. ALT is considered an essential indicator of liver inflammation and a significant marker of disease amelioration. A recent study corroborated that serum ALT levels are an effective indicator of histological changes and can be utilized as an efficient treatment indicator [24]. These observations support the significance of BMI, ALT, and triglycerides in most models in this study. Moreover, on the basis of the decision tree model, we constructed a simplified screening model (Figure 5) for physicians to evaluate FLD in the absence of imaging and pathological evidence, which is helpful for the preliminary screening of patients with fatty liver. The traditional diagnosis of fatty liver mainly depends on imaging results or invasive biopsies, which have high medical and human resource requirements. However, the predictive index of FLD screening based on machine learning is easier to obtain, and the results do not rely on the subjective judgments of doctors. The dataset used for predicting fatty liver disease is mainly unbalanced and categorical, with a much lower number of patients with fatty liver disease than those without. This study utilized the SMOTE-NC method to preprocess the unbalanced data and applied the AIC backward propagation technique for feature engineering. As a result, our model achieved an accuracy comparable to that of a previous study while using fewer feature variables [25,26], including demographic indicators, blood glucose, liver function test, and blood lipid profiles. This narrower range of features reduced the dimensionality issue caused by having too many features, making it easier to collect data from primary medical institutions. Lipid deposition and fibrosis commonly coexist in the liver during the course of FLD. Among them, fibrosis becomes more representative when liver dysfunction reaches the end stage [27]. Samir Hassoun et al. [ 28] developed a machine learning model based on the general population in the United States that can effectively screen for severe liver fibrosis features. This complements our research well and provides a more comprehensive model coverage for screening for FLD. Therefore, ML model-based screening can be carried out in most basic medical institutions and provides new insight into doctors’ diagnoses. Because of the high robustness and prediction accuracy of the model, there is no doubt that ML is feasible for large-scale FLD screening. This study had some limitations. First, although more than 30,000 samples were included in this study, these samples were all from the Second Xiangya Hospital of Central South University, and the population representation was less comprehensive than that of multi-center clinical studies. The generalization ability of the model in different ethnic groups is open to question. Second, the response variables used in the machine learning model constructed in this study are solely based on the diagnostic results of abdominal ultrasound, which have a lower level of evidence compared to liver biopsy and magnetic resonance imaging (MRI). This may potentially affect the accuracy of the predictions. Thirdly, tumor, hepatitis, and other metabolic diseases (such as diabetes, hyperthyroidism, etc.) were not excluded from the population included in this study. As a result, the potential impact of these factors on the predictive model could not be fully assessed. Lastly, it should be noted that we did not gather data on alcohol consumption and medication history among the study population. Therefore, we were unable to rule out the potential interference caused by alcohol and drugs. Previous studies have indicated that both factors can influence the development of fatty liver [29,30,31]. To improve our model, future studies should gather more detailed information on alcohol intake and medication history. Even so, the ML model based on the XGBoost algorithm still had an accuracy of nearly $90\%$ and its AUC value reached $96\%$; thus, it can play an important role in large-scale FLD screening. ## 5. Conclusions In this study, we found that ML models, especially XGBoost, can predict FLD through demographic indicators, blood glucose, liver function test, and blood lipid profile with high accuracy and good repeatability. Among 11 predictive indicators, BMI, Alt, and Tg are vitally important for most models. With the aid of these machine learning methods, physicians can now evaluate a patient’s fatty liver condition in its incipient stages, even in the absence of liver biopsy or imaging evidence. This early assessment can facilitate timely diagnosis and provide a novel avenue for the early screening of fatty liver in primary healthcare facilities. In future research, we will collect more data from various populations to further modify the model and give it better generalization ability. ## References 1. Powell E.E., Wong V.W., Rinella M.. Non-alcoholic fatty liver disease. Lancet (2021) 397 2212-2224. DOI: 10.1016/S0140-6736(20)32511-3 2. Younossi Z., Anstee Q.M., Marietti M., Hardy T., Henry L., Eslam M., George J., Bugianesi E.. Global burden of NAFLD and NASH: Trends, predictions, risk factors and prevention. Nat. Rev. Gastroenterol. Hepatol. (2018) 15 11-20. DOI: 10.1038/nrgastro.2017.109 3. Kasper P., Martin A., Lang S., Kütting F., Goeser T., Demir M., Steffen H.M.. NAFLD and cardiovascular diseases: A clinical review. Clin. Res. Cardiol. (2021) 110 921-937. DOI: 10.1007/s00392-020-01709-7 4. Polyzos S.A., Kountouras J., Mantzoros C.S.. Obesity and nonalcoholic fatty liver disease: From pathophysiology to therapeutics. Metabolism (2019) 92 82-97. DOI: 10.1016/j.metabol.2018.11.014 5. Alqahtani S.A., Schattenberg J.M.. NAFLD in the Elderly. Clin. Interv. Aging (2021) 16 1633-1649. DOI: 10.2147/CIA.S295524 6. Younossi Z.M., Golabi P., de Avila L., Paik J.M., Srishord M., Fukui N., Qiu Y., Burns L., Afendy A., Nader F.. The global epidemiology of NAFLD and NASH in patients with type 2 diabetes: A systematic review and meta-analysis. J. Hepatol. (2019) 71 793-801. DOI: 10.1016/j.jhep.2019.06.021 7. Ciardullo S., Ballabeni C., Trevisan R., Perseghin G.. Liver Stiffness, Albuminuria and Chronic Kidney Disease in Patients with NAFLD: A Systematic Review and Meta-Analysis. Biomolecules (2022) 12. DOI: 10.3390/biom12010105 8. Eslam M., Sarin S.K., Wong V.W., Fan J.G., Kawaguchi T., Ahn S.H., Zheng M.H., Shiha G., Yilmaz Y., Gani R.. The Asian Pacific Association for the Study of the Liver clinical practice guidelines for the diagnosis and management of metabolic associated fatty liver disease. Hepatol. Int. (2020) 14 889-919. DOI: 10.1007/s12072-020-10094-2 9. Chalasani N., Younossi Z., Lavine J.E., Charlton M., Cusi K., Rinella M., Harrison S.A., Brunt E.M., Sanyal A.J.. The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases. Hepatology (2018) 67 328-357. DOI: 10.1002/hep.29367 10. Das A., Connell M., Khetarpal S.. Digital image analysis of ultrasound images using machine learning to diagnose pediatric nonalcoholic fatty liver disease. Clin. Imaging (2021) 77 62-68. DOI: 10.1016/j.clinimag.2021.02.038 11. Acharya U.R., Raghavendra U., Fujita H., Hagiwara Y., Koh J.E., Jen Hong T., Sudarshan V.K., Vijayananthan A., Yeong C.H., Gudigar A.. Automated characterization of fatty liver disease and cirrhosis using curvelet transform and entropy features extracted from ultrasound images. Comput. Biol. Med. (2016) 79 250-258. DOI: 10.1016/j.compbiomed.2016.10.022 12. Fan J.G., Jia J.D., Li Y.M., Wang B.Y., Lu L.G., Shi J.P., Chan L.Y.. Guidelines for the diagnosis and management of nonalcoholic fatty liver disease: Update 2010: (published in Chinese on Chinese Journal of Hepatology. J. Dig. Dis. (2011) 12 38-44. DOI: 10.1111/j.1751-2980.2010.00476.x 13. Chawla N.V., Bowyer K.W., Hall L.O., Kegelmeyer W.P.. SMOTE: Synthetic minority over-sampling technique. J. Artif. Intell. Res. (2002) 16 321-357. DOI: 10.1613/jair.953 14. Sakamoto Y., Ishiguro M., Kitagawa G.. Akaike Information Criterion Statistics (1986) Volume 81 26853 15. Nwanosike E.M., Conway B.R., Merchant H.A., Hasan S.S.. Potential applications and performance of machine learning techniques and algorithms in clinical practice: A systematic review. Int. J. Med. Inform. (2022) 159 104679. DOI: 10.1016/j.ijmedinf.2021.104679 16. Turki T., Wang J.T.L.. Clinical intelligence: New machine learning techniques for predicting clinical drug response. Comput. Biol. Med. (2019) 107 302-322. DOI: 10.1016/j.compbiomed.2018.12.017 17. Mancini A., Vito L., Marcelli E., Piangerelli M., De Leone R., Pucciarelli S., Merelli E.. Machine learning models predicting multidrug resistant urinary tract infections using “DsaaS”. BMC Bioinform. (2020) 21. DOI: 10.1186/s12859-020-03566-7 18. Williams C.D., Stengel J., Asike M.I., Torres D.M., Shaw J., Contreras M., Landt C.L., Harrison S.A.. Prevalence of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis among a largely middle-aged population utilizing ultrasound and liver biopsy: A prospective study. Gastroenterology (2011) 140 124-131. DOI: 10.1053/j.gastro.2010.09.038 19. Bellentani S., Saccoccio G., Masutti F., Crocè L.S., Brandi G., Sasso F., Cristanini G., Tiribelli C.. Prevalence of and risk factors for hepatic steatosis in Northern Italy. Ann. Intern. Med. (2000) 132 112-117. DOI: 10.7326/0003-4819-132-2-200001180-00004 20. Männistö V.T., Simonen M., Soininen P., Tiainen M., Kangas A.J., Kaminska D., Venesmaa S., Käkelä P., Kärjä V., Gylling H.. Lipoprotein subclass metabolism in nonalcoholic steatohepatitis. J. Lipid Res. (2014) 55 2676-2684. DOI: 10.1194/jlr.P054387 21. DeFilippis A.P., Blaha M.J., Martin S.S., Reed R.M., Jones S.R., Nasir K., Blumenthal R.S., Budoff M.J.. Nonalcoholic fatty liver disease and serum lipoproteins: The Multi-Ethnic Study of Atherosclerosis. Atherosclerosis (2013) 227 429-436. DOI: 10.1016/j.atherosclerosis.2013.01.022 22. Heeren J., Scheja L.. Metabolic-associated fatty liver disease and lipoprotein metabolism. Mol. Metab. (2021) 50 101238. DOI: 10.1016/j.molmet.2021.101238 23. Gao X., Fan J.G.. Diagnosis and management of non-alcoholic fatty liver disease and related metabolic disorders: Consensus statement from the Study Group of Liver and Metabolism, Chinese Society of Endocrinology. J. Diabetes (2013) 5 406-415. DOI: 10.1111/1753-0407.12056 24. Seko Y., Sumida Y., Tanaka S., Mori K., Taketani H., Ishiba H., Hara T., Okajima A., Yamaguchi K., Moriguchi M.. Serum alanine aminotransferase predicts the histological course of non-alcoholic steatohepatitis in Japanese patients. Hepatol. Res. (2015) 45 E53-E61. DOI: 10.1111/hepr.12456 25. Chen Y.Y., Lin C.Y., Yen H.H., Su P.Y., Zeng Y.H., Huang S.P., Liu I.L.. Machine-Learning Algorithm for Predicting Fatty Liver Disease in a Taiwanese Population. J. Pers. Med. (2022) 12. DOI: 10.3390/jpm12071026 26. Pei X., Deng Q., Liu Z., Yan X., Sun W.. Machine Learning Algorithms for Predicting Fatty Liver Disease. Ann. Nutr. Metab. (2021) 77 38-45. DOI: 10.1159/000513654 27. Younossi Z.M., Loomba R., Anstee Q.M., Rinella M.E., Bugianesi E., Marchesini G., Neuschwander-Tetri B.A., Serfaty L., Negro F., Caldwell S.H.. Diagnostic modalities for nonalcoholic fatty liver disease, nonalcoholic steatohepatitis, and associated fibrosis. Hepatology (2018) 68 349-360. DOI: 10.1002/hep.29721 28. Hassoun S., Bruckmann C., Ciardullo S., Perseghin G., Di Gaudio F., Broccolo F.. Setting up of a machine learning algorithm for the identification of severe liver fibrosis profile in the general US population cohort. Int. J. Med. Inform. (2023) 170 104932. DOI: 10.1016/j.ijmedinf.2022.104932 29. Lemmer P., Manka P., Best J., Kahraman A., Kälsch J., Vilchez-Vargas R., Link A., Chiang H., Gerken G., Canbay A.. Effects of Moderate Alcohol Consumption in Non-Alcoholic Fatty Liver Disease. J. Clin. Med. (2022) 11. DOI: 10.3390/jcm11030890 30. Jarvis H., O’Keefe H., Craig D., Stow D., Hanratty B., Anstee Q.M.. Does moderate alcohol consumption accelerate the progression of liver disease in NAFLD? A systematic review and narrative synthesis. BMJ Open (2022) 12 e049767. DOI: 10.1136/bmjopen-2021-049767 31. Andrade R.J., Aithal G.P., Björnsson E.S., Kaplowitz N., Kullak-Ublick G.A., Larrey D., Karlsen T.H.. EASL Clinical Practice Guidelines: Drug-induced liver injury. J. Hepatol. (2019) 70 1222-1261. DOI: 10.1016/j.jhep.2019.02.014
--- title: Conditioned Media from Head and Neck Cancer Cell Lines and Serum Samples from Head and Neck Cancer Patients Drive Catabolic Pathways in Cultured Muscle Cells authors: - Nicolas Saroul - Nicolas Tardif - Bruno Pereira - Alexis Dissard - Laura Montrieul - Phelipe Sanchez - Jérôme Salles - Jens Erik Petersen - Towe Jakobson - Laurent Gilain - Thierry Mom - Yves Boirie - Olav Rooyakers - Stéphane Walrand journal: Cancers year: 2023 pmcid: PMC10047086 doi: 10.3390/cancers15061843 license: CC BY 4.0 --- # Conditioned Media from Head and Neck Cancer Cell Lines and Serum Samples from Head and Neck Cancer Patients Drive Catabolic Pathways in Cultured Muscle Cells ## Abstract ### Simple Summary Cancer cachexia in head and neck cancer (HNC) is mainly due to a decrease in food intake, but other causal mechanisms could also be involved. The role of secreted factors from the tumor cells in driving cancer cachexia and especially muscle loss is unknown. In this way, we wanted to study both the action of secreted factors from HNC cell lines and circulating factors in HNC patients on skeletal muscle protein catabolism. We used a conditioned media model and mix of sera from cancer patients to analyze the in vitro metabolic response with primary myotubes. The same metabolic response was obtained with tumor-conditioned media and mix of sera from cancer patients. Patient plasma compounds produced specifically by the tumor seemed to have this effect. Our results indicated that the atrophy observed in HNC patients cannot be solely explained by a deficit in food intake. ### Abstract Background: The role of secreted factors from the tumor cells in driving cancer cachexia and especially muscle loss is unknown. We wanted to study both the action of secreted factors from head and neck cancer (HNC) cell lines and circulating factors in HNC patients on skeletal muscle protein catabolism. Methods: Conditioned media (CM) made from head and neck cancer cell lines and mix of sera from head and neck cancer (HNC) patients were incubated for 48 h with human myotubes. The atrophy and the catabolic pathway were monitored in myotubes. The patients were classified regarding their skeletal muscle loss observed at the outset of management. Results: Tumor CM (TCM) was able to produce atrophy on myotubes as compared with control CM (CCM). However, a mix of sera from HNC patients was not able to produce atrophy in myotubes. Despite this discrepancy on atrophy, we observed a similar regulation of the catabolic pathways by the tumor-conditioned media and mix of sera from cancer patients. The catabolic response after incubation with the mix of sera seemed to depend on the muscle loss seen in patients. Conclusion: This study found evidence that the atrophy observed in HNC patients cannot be solely explained by a deficit in food intake. ## 1. Introduction Head and neck cancers (HNC) account for half a million new cancer cases in the world each year [1]. The prevalence of HNC due to tobacco and alcohol use is decreasing, while the prevalence of HNC due to human papillomavirus (HPV) infection is constantly increasing [2,3,4]. Undernutrition is very common in patients with HNC [5,6]. The main predisposing factors for malnutrition in HNC patients are poor dietary habits (the patient population are typically tobacco and alcohol users), severity of tumor stage (high tumor volume) at the time of treatment due to late diagnosis, and swallowing disorders or even mechanical obstruction caused by the tumor itself. The decrease in food intake seemed to be the main issue for nutritional imbalance and, therefore, etiology of malnutrition and cachexia in HNC [7]. Cancer cachexia is characterized by a loss of lean body mass, particularly skeletal muscle mass. One in three cancer patients are thought to die due to cancer cachexia, although the rate varies with cancer type. Cancer cachexia increases the risk of surgical complications during cancer surgery, impairs patient quality of life, and decreases patient survival. Cachexia-driven loss of skeletal muscle mass is multifactorial, and cannot be reversed by conventional nutritional support only [8]. Skeletal muscle primarily has a contractile function but it also plays a key role in body metabolic homeostasis, as it is the main body reservoir of amino acids and a primary site of insulin-stimulated glucose transport and utilization. Muscle protein homeostasis is an interplay between two key processes: anabolism that promotes muscle protein synthesis and inhibits muscle protein degradation, and catabolism that inhibits muscle protein synthesis and promotes muscle protein breakdown. If the protein anabolism–protein catabolism balance is disrupted, it results in a net loss of muscle proteins, which can lead to cachexia. A number of catabolic or anabolic drivers were identified in both animal models and in humans. Low levels of anabolic hormones (testosterone, growth hormone, insulin-like growth factor (IGF-1)), altered insulin production and tissue sensitivity to insulin, high levels of myostatin and high levels of inflammatory cytokines (TNFα, IL-6) are considered the main endocrine-system hypo-anabolic factors likely to drive cancer cachexia [9]. Among the catabolic factors, significant activation of autophagy and ubiquitin proteasome pathways are thought to drive increased muscle catabolism in cancer cachexia [9]. Some cancers have a high prevalence of cachexia while others have a lower one [10]. This could be due to a decrease in food intake and/or nutrient absorption, especially in digestive tract cancer. However, these food-related mechanisms cannot entirely explain the high prevalence of cachexia in other cancers, such as lung cancer or HNC [11]. One under-researched hypothesis is that some tumor cells have the ability to produce and secrete compounds that have pro-catabolic and anti-anabolic effects on skeletal muscle. These secreted molecules could, thus, act from a distance of the tumor site and dysregulate protein metabolism, thus contributing to cachexia. One model used to test this hypothesis is to treat muscle cells with conditioned media (CM) made from tumor cells [12,13,14,15,16]. Data produced using this type of model argued that tumor cell-derived factors are able to induce muscle atrophy [17,18,19]. However, these results were challenged by a recent study showing that tumor secretions had no effect on muscle atrophy [20]. Furthermore, there is no data available for HNC tumors, despite the fact that almost $50\%$ of HNC patients are in cachexia at the outset of management [5]. Here, we aimed to evaluate the effect of factors secreted from HNC cell lines or circulating in HNC patients on muscle metabolism. We also aimed to determine whether this change was the same with sera from HNC patients experiencing severe vs. mild muscle loss. For that purpose, we chose two well-known HNC cell lines (UT-SCC-60A, UT-SCC-5) [21,22,23,24,25] and carried out a clinical study to collect sera from cancer and non cancer patients. We hypothesized that some compounds in the secretome produced by HNC tumors or HNC cell lines would deregulate key drivers of protein metabolism within the skeletal muscle cell. ## 2. Materials and Methods This paper includes results from in vitro experiments with cancer cell lines and from a clinical trial named MYOMEC. The MYOMEC trial was approved by the local institutional review board (‘Comité de Protection des Personnes Sud-Est’) and the French drug safety agency (ANSM—Agence Nationale de Sécurité des Médicaments) and was registered at ClinicalTrails.gov under number NCT03111771. All patients included in MYOMEC provided written consent to participate, and the study was performed in full compliance with the Declaration of Helsinki guidelines on medical research involving human participants. ## 2.1. Study Design and Population For the clinical study, patients aged 18 to 75 years old were enrolled at the Department of Otolaryngology–Head and Neck Surgery, Clermont-Ferrand University Hospital. Patients included were assigned into two groups: group 1 was formed by HNC patients and group 2 was a control group. To be included in group 1, a diagnosis of HNC had to be clearly established. Only squamous cell carcinoma was included. Skin cancer and nasopharyngeal, nasal cavity, or salivary gland carcinoma were not included in this study. Group 2 was formed of noncancer patients operated at the otolaryngology unit for benign tumors (such as parotidectomy or thyroidectomy). Exclusion criteria for group 2 were presence of cachexia or a diagnosis of cancer on anatomopathological analysis post-surgery. Exclusion criteria for both groups were: heart failure, respiratory failure (requiring long-term oxygen therapy), chronic renal failure (modification of diet in renal disease (MDRD) clearance < 60 mL/min), moderate or severe chronic obstructive pulmonary disease, and insulin-dependent diabetes mellitus. Participants included in the cancer group were recruited at panendoscopy, which is the cornerstone procedure for HNC assessment. Panendoscopy was performed at the beginning of management, before any nutritional support was engaged. By definition, at that time, the future treatment path chosen for the patient following tumor assessment was not yet known. The patients were classified on the basis of their muscle mass defined by CT scan. Two equal groups were formed: a first group with the lowest L3 muscle-mass index (L3MMI), i.e., severe sarcopenia (SS group), and a second group with no or only mild sarcopenia (MS group). The cut-off in our series was L3MMI = 46 cm2/m2, i.e., the median L3MMI in our cancer cohort. General patient characteristics were collected: age, gender, medical records, tobacco and alcohol use (classified as current user, former user, or never user), usual treatment, tumor characteristics, i.e., site, divided into classes: oral, oropharyngeal, laryngeal, hypopharyngeal cancer, or carcinoma of unknown primary (CUP) of the neck, classified by the tumor-node-metastasis (TNM) staging (TNM 7th edition). Two 5 mL blood samples were harvested just before the panendoscopy. Serum preparations were carried out immediately. The blood was allowed to coagulate for 60 min, and then, sera was obtained by centrifugation at 4000× g at 4 °C for 10 min. Blood samples were snap-frozen in liquid nitrogen and stored at −80 °C until analysis. ## 2.2. Clinical and Biological Nutritional Assessment The nutritional assessment included the measurements of usual weight (weight 3 months beforehand, in kg), current weight (kg), and height (cm). Skeletal muscle function was assessed by the short physical performance battery test (SPPB). The SPPB test comprises 3 tasks (balance, 4 m walk, and 5 chair stands) with a maximum score of 12. Impaired muscle function was defined as a score ≤8 [26,27]. Muscle strength was assessed via a handgrip strength test (mean of 3 measures on the dominant arm with the forearm bent at 90°, result in kg). Blood albumin (g/L), transthyretin (g/L), and C-reactive protein (CRP) were measured on the day of surgery. These measures were used to calculate the following scores:Nutritional risk index (NRI) = 1.519 × blood albumin + (current weight/usual weight) × 41.7;Percent weight loss over the previous 3 months = [(usual weight-current weight)/current weight] × 100;Body mass index (BMI) = current weight/height2 (kg/m2). ## 2.3. Skeletal Muscle Mass Assessment Body composition was assessed by bioelectrical impedance analysis [28,29] including impedance, resistance, and reactance. Impedance values were measured at 5, 50, 100, and 200 kHz. Resistance and reactance values were measured at 50 kHz. These measures were used to calculate the lean mass index using Kyle’s equation and muscle mass index was calculated using Janssen’s equation. The cut-off values for low muscularity were ≤17 kg/m2 in men or ≤15 kg/m2 in women for the Kyle lean mass index [30], and ≤10.76 kg/m2 in men or ≤6.76 kg/m2 in women for the Janssen muscle mass index [31]. Skeletal muscle mass was also assessed by CT scan of the abdomen at the third lumbar vertebrae (L3 level) [32,33,34,35]. This CT scan was performed during the disease staging procedure prior to surgery. The method employed, which is described elsewhere [32], serves to determine L3MMI (cm2/m2). The malnutrition thresholds were set at 52.4 cm2/m2 for men and 38.5 cm2/m2 for women [32]. ## 2.4. Serum Amino Acids Concentration Amino acid concentrations were analyzed using a HPLC method described previously [36]. Briefly, serum samples were deproteinized in $3\%$ 5-sulfosalisylic acid dihydrate (SSA) containing 200 µM norvaline as internal standard. Amino acids from the serum were analyzed using precolumn derivatization with ortho-phthaldialdehyde/3-mercaptoproprionic acid on an HPLC system (Waters 2690 Alliance system with Waters 474 fluorescence detectors; Waters®, Stockholm, Sweden). Serum amino acid concentrations were measured for glutamic acid, asparagine, serine, glutamine, histidine, glycine, threonine, 3-methylhistidine, citrulline, arginine, alanine, taurine, tyrosine, valine, methionine, tryptophan, phenylalanine, isoleucine, ornithine, leucine, lysine, and used to calculate serum essential amino acids (EAA), serum branched-chain amino acids (BCAA), serum total amino acids (TAA). BCAA was the sum of valine + leucine + isoleucine. EAA was the sum of tryptophan + threonine + phenylalaline + lysine + histidine + methionine + valine + leucine + isoleucine. TAA was the sum of all 21 amino acids measured. EAA-to-TAA ratio was calculated. ## 2.5. Plasma Elisa Test Plasma levels of IL6, IL8, IGF1, GDF8, activin-A, follistatin, FGF-21, and GDF-15 were measured by enzyme-linked immunosorbent assay (ELISA) using a Raybiotech® kit under conditions set by the supplier. Plasma levels of proteins were measured on the day of panendoscopy on patients who fasted for at least 6 h. ## 2.6. Cell Culture Human primary myoblasts were obtained from Gibco®. The cell lines used were A11440 and A12555 (lot number 19F, 603, 597). The myoblasts were sub-cultured in T175 flask in growth media (Dulbecco’s Modified Eagle Medium (DMEM)/F12 GlutaMAXTM, Gibco®, Grand Island, NY, USA) supplemented with $10\%$ fetal calf serum (FCS) certified heat-inactivated (Gibco®) and $1\%$ antibiotic–antimycotic (ABAM, Gibco®) at 37 °C and $5\%$ CO2. For experiments, myoblasts were detached at $70\%$ confluence with TrypLETM (Gibco®) and transferred to 6-well plates coated with an attachment factor (Gibco®). Seeding density for the experiments was 85,000 cells per cm2. At $90\%$ confluence, cell differentiation was induced in the myotubes using differentiation medium (DMEM/F12 Glutamax supplemented with $2\%$ horse serum (Gibco®), $1\%$ antibiotic/antimycotic (ABAM, Gibco®)). The medium was replaced every two days. Myotube differentiation was monitored using light microscopy. Myotubes used for experiments were at day 7 to day 10 of differentiation. Cells were washed once with phosphate-buffered saline (PBS) and were then incubated in serum-free media (DMEM F12 GlutaMAXTM) supplemented with $10\%$ of patient serum (for patient serum experiments) or with $4\%$ horse serum media supplemented with a $50\%$ conditioned media from cancer cell lines (final concentration of $2\%$ horse serum in the experimental media). Experiments with conditioned media were carried out in triplicate ($$n = 3$$) and experiments with patient serum were carried out in duplicate ($$n = 2$$). Mouse C2C12 myoblast cells were purchased from the ATCC (American Type Culture Collection, Manassas, VA, USA). Myoblasts were cultured in a growth medium composed of DMEM containing 4.5 g/L glucose, 2.4 g/L sodium bicarbonate, $10\%$ fetal bovine serum, 100 IU/mL penicillin, and 0.1 mg/mL streptomycin, and incubated at 37 °C in humidified air with $5\%$ CO2. The medium was changed every other day to ensure growth until $90\%$ confluence. Myotube formation was induced by changing the growth medium to a differentiation medium consisting of DMEM supplemented with $2\%$ horse serum, 100 IU/mL penicillin, and 0.1 mg/mL streptomycin for 5 days before cell treatment. Passages between 4 and 10 were used for experiments. Tumor cell lines were gifted as a courtesy of Pr. Dalianis (Department of Oncology and Pathology, Karolinska Institutet, Stockholm, Sweden) with permission from Pr. Grenman (Otolaryngology, University of Turku, Finland). The cell lines used were UT-SCC-60A and UT-SCC-5. These cell lines were cultured in a T75 flask in growth media (DMEM/F12 GltuaMAXTM, Gibco®) supplemented with $10\%$ FCS-certified heat-inactivated (Gibco®) and $1\%$ ABAM (Gibco®) at 37 °C and $5\%$ CO2. The cells were used for experiments at confluence. Differentiated human myotubes were incubated for 48 h with conditioned media or patient serum. ## 2.7. Myotube Morphology Analysis Myotubes were photographed directly in the culture plates without fixation, using an AxioCam ERc5s digital camera coupled with an AxioVert. A1 microscope and ZEN 2.3 software (Zeiss, Germany). Myotube diameter was measured from three independent experiments on myotubes in each condition. For each myotube, three random measurements were performed along the length of the myotube ($$n = 3$$ measurements/myotube) using the ZEN 2.3 software, and the average of these three measurements was considered as a single value. ## 2.8. Generation of Tumor Cell-Conditioned Medium (TCM) Tumor cell lines were cultured until confluence in T75 flasks with growth media. At confluence, the cells were washed three times with PBS and were then incubated with 7 mL of DMEM/F12 media supplemented with $1\%$ ABAM without serum. Tumor-conditioned medium (TCM) was collected after 24 h, centrifuged at 3000 g for 5 min at room temperature to remove cell debris, sterilized by filtration (on a 22 µm filter), and stored at −80 °C until use. Control-conditioned medium (CCM) was made by incubating 7 mL of DMEM/F12 medium supplemented with $1\%$ ABAM without cells and without serum for 24 h in a T75 flask, centrifuged at 3000× g for 5 min at room temperature, sterilized by filtration (on a 22 µm filter), and stored at −80 °C until use. ## 2.9. RNA Isolation, Reverse Transcription, and Quantitative Polymerase Chain Reaction (RT-qPCR) Total RNA from the myotubes was isolated by the TRIzol method (Thermo Fischer Scientific®, Waltham, MA, USA) following the manufacturer’s protocol. RNA purity and quality were assessed by spectrophotometry (Nanophotometer, Implen®, München, Germany) and agarose gel electrophoresis. Complementary DNA (cDNA) synthesis was performed using iScript™ Reverse Transcription Supermix (BioRad®, Hercules, CA, USA). qPCR was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad®). The gene expression assays were as follows: 3 µL cDNA, 1 µL specific primers, 6 µL nuclease-free water, and 10 µL SYBR Green Supermix. The gene expressions studied were: P62, LC3B, MurF1, TRAF6, Fox O, MafBx, DDIT3, Perilipin-3 (PLIN3), and IL6. To study the autophagy pathway, we measured the expression of LC3 and p62. The proteasome pathway was monitored by observing expression levels of the E3 ligases, MURF1, and atrogin1/MafBx. We also observed TRAF6 and FOXO3, two transcription factors involved in the regulation of autophagy and proteasome pathways. We measured the expression of DDIT, also known as CHOP, as a marker of endoplasmic reticulum stress. The IL-6 gene expression was used as a marker of inflammation and the expression of perilipin-3 (plin3) as a marker of intramuscular lipolysis. GAPDH expression was used as a reference control. Reactions were performed using a thermal cycler (CFX 96-Well Real-Time Cycler, BioRad®) using thermal cycles of 95 °C for 2 min and then, 40 cycles of 5 s at 95 °C, followed by 30 s at 60 °C. Cycle threshold (Ct) was normalized against GAPDH reference gene to give a relative normalized expression (Cq). These Cq values were compared between exposure to TCM and CCM. The primer pairs used for RT-qPCR were designed using Primer-Blast via PubMed and then, were purchased from Invitrogen (Waltham, MA, USA). These custom primer pairs are reported in Appendix A. Primers not listed in Appendix A were purchased directly from Invitrogen. ## 2.10. Protein Isolation and Western Blotting Myotubes were treated either with CCM or TCM from the two tumor cell lines (UT-SCC-60A and UT-SCC-5) for different durations. The cells at 7 days of differentiation were washed three times with PBS. The myotubes were then incubated in a 6-well plate with 1 mL of DMEM/F12 Glutamax supplemented with $4\%$ horse serum (Gibco®) and $1\%$ ABAM, added at 1 mL per well of TCM or CCM for 48 h. The cells were then scraped on ice with Laemmli 2X lysis buffer (Sigma-Aldrich®, St. Louis, MO, USA). Denatured proteins were then separated by SDS-page on a polyacrylamide precast gradient gel (mini-Protean TGX-Gel®, BioRad). After UV activation for further quantification, proteins were blotted on a polyvinylidene membrane (immobillon-FL; Millipore®, Burlington, MA, USA), and the membrane was UV-scanned for total protein quantification (BioRad®). Immunoblots were blocked 1 h with a blocking buffer (Odyssey blocking buffer®, LI-COR Bioscience, Lincoln, NE, USA) and then, were probed with the following primary antibodies: anti-myosin ($\frac{1}{1000}$, MF20, DSHB®), anti-MuRF1 ($\frac{1}{1000}$, no. NBP1-54939; Novus Biologicals®, Centennial, CO, USA), anti-atrogin-1 ($\frac{1}{1000}$; no. AP2041; ECM Biosciences®, Versailles, KY, USA), anti-LC3 ($\frac{1}{1000}$; no. 0231–100; Nanotools, München, Germany), anti-phospho-AKT (Ser 473) ($\frac{1}{1000}$; #9271; Cell Signalling®, Danvers, MA, USA), and anti-p62 ($\frac{1}{2000}$, no. H00008878-M01; Abnova®, Taipei, Taiwan). After several washes in PBS plus $0.1\%$ Tween 20, the immunoblots were incubated with IRDye 800 CW or 680 LT (LI-COR Biosciences®). The membranes were then analyzed with an odyssey scan (LI-COR Bioscience®) or with ImageJ software 1.52 (NIH) for total protein quantification. ## 2.11. Quantification of Autophagy Flux Myotubes were cultured in 6-well plates (Greiner Bio-One®) and incubated for 48 h at 37 °C and $5\%$ CO2 with either CCM or TCM for the CM experiments and with a $10\%$ mixed serum from cancer patients (with two different mixes: severe (SS group) or mild (MS group) sarcopenia (see below)) or control patients. At 6 h before the end of the 48 h incubation, half of the wells were incubated with 50 µM chloroquine (CQ), an autophagy inhibitor. Cells were then harvested as described in the protein isolation/immunoblotting method. Autophagy flux was calculated by the following calculation: autophagy flux = (LC3B2 expression with CQ − LC3B2 expression without CQ) × 100. ## 2.12. Proteasome Activity Measurements After experimental myotube incubation, cells were lysed on ice with a homogenization buffer (50 mmol Tris-HCl/L, pH = 7.5, 1 mmol EDTA/L, 5 mmol MgCl2/L, 0.1 mmol dithiothreitol/L, $10\%$ glycerol). After centrifugation and before measurement of protease activities, protein content was measured by spectrophotometry (Nanophotometer NP 80, Implen®, Germany). Chymotrypsin-like activity of the proteasome fraction was measured using the fluorogenic substrate SUC-LLVY-AMC [succinyl-Leu-Val-Tyr-7-amido-4-methylcoumarin; AMC)] (Sigma-Aldrich®) [37]. Then, 10 µL of the supernatant fluid (∼10 μg protein) was incubated in 100 μL of buffer (50 mmol Tris-HCl/L, pH 7.5, 1 mmol ATP/L, 5 mmol MgCl2/L, and 150 μmol LLVY/L) in microplates. Standard curves were prepared using the AMC. Fluorescence was measured continuously over 1 h at 37 °C on a SpectraMax® i3X system (Molecular devices®) at λex = 380 nm and λem = 460 nm. Proteolytic activity was calculated from the increment of the curves from samples and standards and were expressed as pmol of AMC released/μg protein per minute. ## 2.13. Statistical Analysis All statistical analyses were performed using Stata software (version 13, StataCorp®, College Station, TX, USA). Significance was set at a two-sided type I error of $5\%$. Patient characteristics were presented as mean ± standard deviation (SD) or median [interquartile range] for continuous data (assumption of normality assessed using the Shapiro–Wilk test) and as number of patients and associated percentages for categorical variables. Quantitative variables were compared between independent groups (control vs. cancer) by a Student t-test or a Mann–Whitney test if the conditions of the t-test did not hold ((i) normality and (ii) homoscedasticity studied via a Fisher–Snedecor test). Between-group comparisons were performed using a chi-square test or, when appropriate, Fischer’s exact test for categorical variables. Relationships between quantitative outcomes were studied using correlation coefficients (Pearson or Spearman according to statistical distribution). Šidák correction was applied to address the problem of multiple comparisons. Finally, in paired contexts, we applied the usual appropriate statistical tests: paired Student test or a Wilcoxon test according to the assumptions of the t-test. ## 3.1. Epidemiological Characteristics The MYOMEC study included 34 patients. The key epidemiological characteristics of the patients are summarized in Table 1 and Table 2. The cancer group was mainly composed of male patients, aged mostly between 50 and 70 years old, with an average BMI of 22.0 ± 3.2 kg/m2. Cancer-group patients had lost on average $6\%$ of their usual weight in the last 3 months, and were mostly in a state of malnutrition, with an average L3MMI of 43.5 ± 7.8 cm2/m2 and an average NRI of 93.9 ± 10.9. All the sub-localizations of HNC were represented in our population (Table 2). The control group was mostly composed of younger female participants, with a stable weight and a tendency to be overweight (BMI = 30.4 ± 7.0). The difference in the recruitment, especially concerning gender, was due to our department’s surgical activity, as the control group was mainly operated for a benign tumor of the thyroid (Table 3). ## 3.2. Tumor CM Drove Atrophy on Differentiated Myotubes The first step was to test the effect of the secretions of HNC cell lines on C2C12 myotubes after 48 h of incubation with conditioned media. The tumor-conditioned media (TCM) clearly induced C2C12 atrophy compared to control-conditioned media (CCM) treatment. The average myotube diameter was 23.1 ± 3.4 µm after incubation with TCM ($$n = 324$$) vs. 27.0 ± 3.8 µm after incubation with CCM ($$n = 231$$; $p \leq 0.0001$; Figure 1A–C). This result was mainly due to the presence of a higher number of myotubes with a small average diameter after incubation with TCM (Figure 1D). We investigated whether the change in myotube average diameter (assessed by microscopy) corresponded to a change in myotube myosin content (assessed by Western blot analysis). The myosin expression was $46\%$ lower in the myotubes incubated with TCM compared to CCM (137 vs. 74 A.U.; CCM vs. TCM; $$p \leq 0.01$$; Figure 1E). In response to these results and our primary aim, we explored the effect of TCM in human-differentiated myotubes. We found a $55\%$ decrease in myosin content in human-differentiated myotubes after 48 h incubation with TCM (5.9 ± 1.4 vs. 2.7 ± 0.7 A.U; CCM vs. TCM; $p \leq 0.001$; Figure 2). Given the more physiological aspect of the human myotubes, we decided to continue the experiments with human myotubes. ## 3.2.1. Tumor CM Disrupted the Catabolic Pathway and Its Regulation Compared to controls, human myotubes treated with TCM had lower p62 gene expression (0.92 ± 0.06 vs. 0.68 ± 0.20 A.U.; CCM vs. TCM; $$p \leq 0.04$$) but no difference in LC3 gene expression (0.98 ± 0.01 vs. 1.04 ± 0.25 A.U. for CCM vs. TCM; $$p \leq 0.73$$; Figure 3A). We found an overall decrease in proteasome gene expression in muscle cells cultured with TCM (Murf1 = 1.1 ± 0.07 for CCM vs. 0.3 ± 0.2 for TCM; $p \leq 0.001$; Fox $O = 1.1$ ± 0.07 vs. 0.6 ± 0.3, $$p \leq 0.002$$; TRAF6 = 1.0 ± 0.1 vs. 0.7 ± 0.2, $$p \leq 0.04$$; MafBx = 0.8 ± 0.2 vs. 0.9 ± 0.2, $$p \leq 0.4$$) (Figure 3B). However, compared to incubation with CCM, incubation with TCM led to no change in DDIT3 expression (0.88 ± 0.16 vs. 1.20± 0.42, $$p \leq 0.25$$), a decrease in PLIN3 expression (1.22 ± 0.31 vs. 0.82 ± 0.15, $p \leq 0.001$), and an increase in IL6 expression (0.02 ± 0.003 vs. 0.14 ± 0.06, $p \leq 0.001$) (Figure 3C). As already shown [38], the decreased proteasome gene expression seen in muscle cells exposed to TCM could be due to feedback control of gene expression after high activation of the proteasome during the first hours of incubation. To test this idea, we ran an enzymatic time–course analysis of proteasome activity. As expected, we found that TCM was able to activate the proteasome after a short period of exposure, i.e., 6 h, although this change did not reach statistical significance (+$37.5\%$, $$p \leq 0.16$$). Interestingly, after a sharp increase in proteasome markers at 6 h of incubation, the level of proteasome gene expression fell rapidly up to 20 h of incubation with TCM (Figure 3D). ## 3.2.2. TCM Induced the Autophagy/Lysosome Pathway There was significant activation of the autophagy/lysosome pathway after 48 h incubation with the TCM compared with the CCM (Figure 4). Taken together, the early proteasome activation followed by a later increase in autophagic activity likely explain the decrease in myotube diameter. ## 3.3. Sera from Cancer Patients Were Not Able to Drive Atrophy in Differentiated Myotubes but Changed Myotube Metabolism in the Same Way as Conditioned Media As the hypothesis was that the muscle loss seen in cancer patients is partially due to tumor secretions, we ran the same experiments but using patient sera from the MYOMEC study. We did not observe any significant difference in myosin content after incubation with either a mix of sera from cancer patients or a mix of sera from control patients. The mean myosin level was 16.70 ± 0.20 A.U. after incubation with the control-group mix and 21.42 ± 6.00 after incubation with the cancer-group mix ($$p \leq 0.23$$, Figure 5). We investigated whether a mix of sera from patients with the highest muscle loss was able to induce atrophy, but we did not find any difference between the mix of sera from cancer patients with the highest muscle loss (SS group) vs. patients with less muscle loss (MS group) (Figure 5). ## 3.3.1. Autophagy Genes The incubation of differentiated myotubes with sera from cancer patients led to a decrease in both LC3 expression and p62 expression (LC3 = 1.54 ± 0.76 for the control group vs. 0.94 ± 0.14 for the cancer group, $$p \leq 0.01$$; p62 = 0.90 ± 0.26 for the control group vs. 0.54 ± 0.23 for the cancer group, $$p \leq 0.008$$). The decrease in gene expression seemed to be dependent on the magnitude of muscle loss, with lower gene expression in the case of severe muscle loss (p62 = 0.93 ± 0.45 for the MS group vs. 0.6 ± 0.34 for the SS group, $$p \leq 0.005$$; LC3 = 0.98 ± 0.18 for the MS group vs. 0.91 ± 0.10 for the SS group, $$p \leq 0.43$$) (Figure 6A). Taken as a whole, the results on autophagy gene expression in differentiated myotubes obtained with patient sera were similar to the results obtained with TCM. ## 3.3.2. Proteasome Genes As already observed with TCM, incubation of differentiated myotubes with sera from cancer patients decreased proteasome gene expression in the differentiated myotubes. All gene expressions were repressed or tended to be lower (Figure 6B): Murf1 = 0.87 ± 0.64 for the control group vs. 0.59 ± 0.43 for the cancer group, $$p \leq 0.14$$; MafBx = 1.41 ± 0.47 for the control group vs. 0.75 ± 0.11 for the cancer group, $$p \leq 0.002$$; TRAF6 = 1.01 ± 0.29 for the control group vs. 0.98 ± 0.24 for the cancer group, $$p \leq 0.85$$; Fox $O = 1.25$ ± 0.31 for the control group vs. 0.98 ± 0.30 for the cancer group, $$p \leq 0.08.$$ In addition, the decrease in proteasome gene expression was linked to the magnitude of muscle loss in cancer patients. When we reviewed the effect of muscle mass, we observed that proteasomal gene expression was reduced in myotubes incubated with sera from severely sarcopenic HNC patients, as found for TCM (Murf 1 = 0.31 ± 0.26 for SS vs. 0.80 ± 0.43 for MS, $p \leq 0.001$; MafBx = 0.81 ± 0.14 for SS vs. 0.68 ± 0.03 for MS, $$p \leq 0.02$$; TRAF6 = 0.85 ± 0.19 for SS vs. 1.12 ± 0.24 for MS, $$p \leq 0.01$$; Fox $O = 0.79$ ± 0.17 for SS vs. 1.18 ± 0.25 for MS, $p \leq 0.001$ (Figure 6)). ## 3.3.3. Endoplasmic Reticulum Stress/Lipid Metabolism/Inflammation Genes PLIN3 gene expression increased after incubation with a mix of sera from cancer patients (PLIN3 = 0.72 ± 0.08 vs. 0.89 ± 0.08 ($p \leq 0.001$) for the control group and cancer group, respectively). There was a decrease in DDIT3 expression (DDIT3 = 1.56 ± 1.0 vs. 0.89 ± 0.25 for the control group and the cancer group, respectively, $$p \leq 0.01$$) and no change in IL-6 expression between the control group and cancer group sera mixtures (IL6= 0.51 ± 0.19 vs. 0.55 ± 0.20, $$p \leq 0.7$$). However, we found results that diverged from our previous findings with TCM, concerning the effect of sera selected according to the extent of patient muscle mass loss. With patient sera, IL-6 expression increased according to muscle loss (IL-6 = 1.07 ± 0.16 for the SS group vs. 0.47 ± 0.15 for the MS group, $p \leq 0.001$). PLIN3 gene expression increased with increased muscle loss (PLIN3 = 0.93 ± 0.09 for SS vs. 0.85 ± 0.04 for MS, $$p \leq 0.001$$) whereas DDIT3 showed no change (DDIT3 = 0.84 ± 0.17 for SS vs. 0.94 ± 0.40 for MS, $$p \leq 0.35$$) (Figure 6C). ## 3.3.4. Proteasome Activity There was no change in proteasome activity after muscle cell treatment with sera from control vs. cancer patients (11.64 ± 2.01 pmol/min/µg for the control group vs. 12.69 ± 3.26 pmol/min/µg for the cancer group, $$p \leq 0.7$$; Figure 7A). However, proteasome activity significantly increased in patients with low muscle loss (14.97 ± 1.06 pmol/min/µg) but decreased in patients with severe muscle loss (10.41 ± 3.1 pmol/min/µg) ($p \leq 0.001$; Figure 7A). ## 3.3.5. Autophagy/Lysosomal Pathway The autophagy/lysosomal pathway was activated in differentiated human myotubes after incubation with serum from the SS group (mean value 0.9 ± 0.4), but there was no difference between MS sera and control sera (0.5 ± 0.2 for MS vs. 0.6 ± 0.1 for control; Figure 7B). ## 3.3.6. Synthesis of the Results Finally, we produced a synthesis table of the effect seen with the tumor-conditioned media and the serum-conditioned media on human-differentiated myotubes. The results are shown in Table 4. ## 3.4. Blood Parameters That Could Explain the Serum Effect A recent study established a link between high serum essential amino acids and skeletal muscle depletion in gastrointestinal cancer cachexia [39]. Moreover, much evidence suggests that inflammatory cytokines are key factors of muscle catabolism [40]. We, therefore, investigated the levels of these factors in our patient sera. We found no statistically significant change in blood levels of C-reactive protein, total protein, total cholesterol, triglycerides, HDL, TSH, GDF-15, FGF-21, Testosterone, IL-6, IL-8, Follistatin, and IGF-1, including between SS-group and MS-group sera. The results are reported in Supplementary Tables S1 and S2. We found no correlations between amino acid concentrations and muscle mass. ## 4. Discussion Muscle loss is very frequent at the outset of HNC management, with almost $60\%$ of patients having muscle wasting, i.e., cachexia [5]. Cachexia in HNC patients is mainly due to malnutrition, as HNC causes pain, odynophagia, anorexia, and other nutritional impact symptoms (NIS) that decrease food intake [7]. The main NIS is reported to be dysphagia, as HNC patients with dysphagia experienced $5\%$ body weight loss 45 days earlier than those without difficulty swallowing [7], but anorexia, pain, and mouth sores are also important explanatory factors. Other factors, such as tumor secretions, could also contribute to weight loss and especially muscle loss in HNC patients. However, to the best of our knowledge, there is no data on the potential effect of HNC tumor secretions on regulation of protein metabolism in skeletal muscle. Therefore, the main objective of the present work was to assess in vitro whether tumor secretions from HNC can contribute to muscle atrophy. We chose to include only patients at the outset of management in order to decrease the potential bias of treatment (surgery/radiotherapy/chemotherapy) effects on nutritional status. The patients were classified into subgroups according to degree of muscle loss based on L3MMI, i.e., a group with severe muscle loss (SS group) and a group with no or mild muscle loss (MS group). We found a higher proportion of cachectic patients with tumor stage T3-4 compared to non-cachectic patients ($73\%$ vs. $55\%$, not significant). This observation supports the fact that the pathogenesis of cancer cachexia in HNC could be a nutritional issue, because the presence of a larger tumor in T3-4 patients could contribute to dysphagia and upper digestive tract obstruction, leading to decreased food intake and weight loss. However, in the population studied in this study, food intake, dysphagia, and tumor stages were not different between the SS group and the MS group (Saroul N. CHU Clermont-Ferrand, INRAE, UNH, Université Clermont Auvergne 63000 Cler-mont-Ferrand, France. PhD thesis. Université Clermont Auvergne. 2021). Taken together, these data argue that skeletal muscle loss in HNC cannot be explained only by a decrease in food intake and subsequent low nutrient absorption. Another hypothesis is that a larger tumor is able to secrete more active factors involved in weight loss, contributing to greater muscle loss in patients suffering from a T3-4-stage HNC. Many studies focused on potential biomarkers of cancer cachexia [41,42,43,44,45,46,47,48,49,50]; however, only one of these studies included HNC patients [45]. The main blood biomarkers of cachexia found in these studies were pro-inflammatory proteins (CRP, IL-6, IL-8, IL-10, IFNγ, TNFα), members of the TGFβ family (myostatin, activin), tumor-derived factors (ZAG, midkine), markers of lipolysis (leptin, adiponectin), and other proteins such as IGF-1, albumin, and angiotensin II. None of these studies defined the cachexia state on the basis of skeletal muscle loss but considered only weight loss. Here, none of the biological markers measured were clearly associated with either weight loss or muscle loss (Saroul N. CHU Clermont-Ferrand, INRAE, UNH, Université Clermont Auvergne, 63000 Cler-mont-Ferrand, France. PhD thesis. Université Clermont Auvergne. 2021.). We only found trends towards increased blood CRP, white blood cells, GDF-15, IL-8, and FGF-21, and decreased in testosteronemia in SS patients compared to MS patients. This could be due to the small sample size of our population. Note, however, that blood IL6 concentrations in SS (37 pg/mL) and MS patients (42 pg/mL) were higher than those measured in similar studies regarding HNC, with the mean blood level in the study population of IL6 varying from 1.35 to 9.7 pg/mL [51,52,53,54]. Il-6 is known to be increased in cancer cachexia and to correlate with weight loss in pancreatic cancer patients [55], and anti-IL-6 drugs are able to diminish muscle loss in preclinical models [56]. It was previously shown that HNC tumors are able to secrete IL-6 [57]. The very low level of IGF-1 in our study compared to normal values was another interesting result [58]. It was previously demonstrated that blood IGF-1 was reduced in oral cancer, cancer cachexia, and fasting [59,60,61]. Low IGF-1 is probably due to degraded nutritional status. IGF-1 is a key driver of protein metabolism. This hormone was shown to stimulate protein synthesis and inhibit protein degradation, especially in skeletal muscle, and decreased circulating IGF-1levels were associated with cancer cachexia [62]. Therefore, a reduction in plasma IGF-1 concentrations may have contributed to the effects of conditioned media from HNC plasma on muscle cells. Because the effect of tumor secretion on muscle atrophy is still under debate, we set out to evaluate the effect of tumor secretion on a simple easily-cultivable model of mouse-differentiated myotubes (C2C12 cell line). C2C12 cell lines is a commonly used model to study muscle cell metabolism in vivo, despite the fact that this is a non-human cell line [63]. With this model, we observed that tumor secretions from HNC cell lines are able to induce atrophy in skeletal muscle cells. This observation was then confirmed in a more physiological model of human primary myotubes, where we observed a $50\%$ decrease in myosin content after incubation with TCM. The next step was to find out the muscle cell pathway involved in the ability of cancer cell line secretions to induce atrophy. Our results support the idea that tumor secretions likely activate the proteasome and autophagy/lysosome pathways in a time-dependent manner, probably with early activation of the proteasome pathway and longer-term activation of the autophagy/lysosome pathway. This long-term activation of the autophagy process in our TCM model was already reported in vitro, where it was linked to IL-6 trans-signaling [64]. In this in vitro model [64], autophagy flux continued to increase after 3 days of myotubes incubation with a TCM. We obtained the same result after incubation of differentiated myotubes with both cancer-cell-conditioned media and cancer-patient sera. Another important result was the repression of key genes of the ubiquitin/proteasome system in both cancer-cell-conditioned media and cancer-patient sera. To the best of our knowledge, there are no other data available on the decrease in ubiquitin/proteasome gene expression after long-term incubation with TCM. However, in most publications, ubiquitin/proteasome-system gene expression increased during the first hours of incubation and decreased thereafter [38,65,66,67,68]. For instance, Zhang et al. reported that, at 24 h, mRNA levels of MafBx were lower in cells treated with cancer-patient CM than cells treated with control-subject CM [38]. Taken together, these data suggest that ubiquitin/proteasome gene expression is activated in a time-dependent manner in muscle cells treated with media containing tumor secretions. One of the mechanisms that may explain the reduction in proteasome activity could be the energy demand induced by the proteolytic system. The proteasome system demands much energy in the form of ATP. We observed a reduction in expression levels of key mitochondrial genes in muscle cells treated with serum from HNC patients (Saroul N. CHU Clermont-Ferrand, INRAE, UNH, Université Clermont Auvergne 63000 Cler-mont-Ferrand, France. PhD thesis. Université Clermont Auvergne. 2021.). One of the compounds present in the plasma of cancer patients could impair the production or consumption of ATP by muscle cells, thus limiting the activity of the proteolytic system [14]. Note that both the mix of sera from severely sarcopenic HNC patients and the CM from tumor cell lines were able to induce this decrease in proteasome-system gene expression. Our results also showed that the genes from the ubiquitin/proteasome system were not always involved in the atrophic program in the same way. This differential activation was described previously, with some studies finding a significant increase in atrogin1/MAFbx gene expression without change in Murf1 gene expression, whereas other studies found an activation of Murf 1 without changes in atrogin1/MAFbx gene expression [69,70,71]. Another explanation for muscle atrophy in HNC patients is a decrease in muscle protein synthesis, as seen in CM-treated myotubes. Decreased muscle protein synthesis is a hallmark of cancer cachexia and was observed in both human and mice models [72,73]. Measuring the phosphorylation state of the different proteins of the Akt/mTOR pathway can be challenging in terms of the timing of sampling after the anabolic stimuli and necessitates more complex experimental preparation with the need for an insulin and amino acid restriction period before stimulation. This is why the actual measure of protein synthesis with the use of a stable isotope or via the sunset method is the best method to monitor protein synthesis. However, in our study, we did not measure the protein synthesis rate and only measured the phosphorylation state of AKT at the serine 473, as AKT/mTOR is the main anabolic pathway regulating muscle protein synthesis [40,74]. We found that the TCM was able to decrease the phosphorylation of AKT, but we did not observe any difference between the control and HNC patient groups (both SS and MS; Saroul N. CHU Clermont-Ferrand, INRAE, UNH, Université Clermont Auvergne 63000 Cler-mont-Ferrand, France. PhD thesis. Université Clermont Auvergne. 2021). Therefore, some compounds secreted by tumor cells or the low IGF1 level measured could have downregulated the Akt/mTOR pathway, leading to a decrease in protein synthesis. However, more data are needed to understand the role of tumor-secreted factors on muscle protein synthesis. Our study had some limitations that must be addressed. In our conditioned media experiments, we used a no-cell-control approach, a better control would have been to expose the myotubes to conditioned media made after incubation for 24 h hour, with a non-tumorigenic cell line corresponding to squamous cell carcinoma. With our approach, it was difficult to confirm that the effect of the TCM was only due to tumor secretions and not by the depletion of the media component by tumor cells during the 24 h incubation. However, the TCM was produced from the tumor CM diluted with fresh media and supplemented with $2\%$ serum, as we used when we differentiated the myotubes with a media change every 48 h. The second limitation of this study was that our control group did not match with the patient group in terms of age, sex, and body composition. Our results comparing control and cancer group could, therefore, have been an effect of the difference in hormone and metabolic profile between the two groups. Due to this reason, we chose to investigate, within our cancer group, the impact of patients’ muscle mass status on the in vitro effect of their serum on human myotubes. Furthermore, we observed that our most intriguing results were between the mild sarcopenic and severely sarcopenic HNC patients, where no difference in age, sex, and body composition could explain the differences observed. The third limitation of our study was the comparison between experiments using tumor-conditioned media and patients’ sera. For the patients, we used a concentration of $10\%$ serum, a concentration much higher than the concentration of fetal bovine serum ($2\%$) used during the tumor CM experiments. This might have been a confounding factor, which would explain the discrepancy observed between the impact of TCM on myotubes atrophy that was not observed with patients’ sera. ## 5. Conclusions This study showed, for the first time, that cancer cell lines from HNC were able to induce atrophy in differentiated myotubes. Our results indicated that the atrophy observed in HNC patients cannot be solely explained by a deficit in food intake. It is now necessary to dissect the secretome of tumors in HNC patients, in order to find new biomarkers of deregulated muscle metabolism in cancer patients and to define new therapeutic approaches and targets that can help reduce cancer cachexia in HNC patients. ## References 1. Bray F., Ferlay J., Soerjomataram I., Siegel R.L., Torre L.A., Jemal A.. Global Cancer Statistics 2018: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. (2018.0) 68 394-424. DOI: 10.3322/caac.21492 2. Haeggblom L., Attoff T., Yu J., Holzhauser S., Vlastos A., Mirzae L., Ährlund-Richter A., Munck-Wikland E., Marklund L., Hammarstedt-Nordenvall L.. Changes in Incidence and Prevalence of Human Papillomavirus in Tonsillar and Base of Tongue Cancer during 2000-2016 in the Stockholm Region and Sweden. Head Neck (2019.0) 41 1583-1590. DOI: 10.1002/hed.25585 3. Gillison M.L., Chaturvedi A.K., Anderson W.F., Fakhry C.. Epidemiology of Human Papillomavirus-Positive Head and Neck Squamous Cell Carcinoma. J. Clin. Oncol. (2015.0) 33 3235-3242. DOI: 10.1200/JCO.2015.61.6995 4. Mirghani H., Bellera C., Delaye J., Dolivet G., Fakhry N., Bozec A., Garrel R., Malard O., Jegoux F., Maingon P.. Prevalence and Characteristics of HPV-Driven Oropharyngeal Cancer in France. Cancer Epidemiol. (2019.0) 61 89-94. DOI: 10.1016/j.canep.2019.05.007 5. Saroul N., Pastourel R., Mulliez A., Farigon N., Dupuch V., Mom T., Boirie Y., Gilain L.. Which Assessment Method of Malnutrition in Head and Neck Cancer?. Otolaryngol. Head Neck Surg. (2018.0) 158 1065-1071. DOI: 10.1177/0194599818755995 6. Fearon K.C.H.. Cancer Cachexia and Fat-Muscle Physiology. N. Engl. J. Med. (2011.0) 365 565-567. DOI: 10.1056/NEJMcibr1106880 7. Kubrak C., Olson K., Jha N., Jensen L., McCargar L., Seikaly H., Harris J., Scrimger R., Parliament M., Baracos V.E.. Nutrition Impact Symptoms: Key Determinants of Reduced Dietary Intake, Weight Loss, and Reduced Functional Capacity of Patients with Head and Neck Cancer before Treatment. Head Neck (2010.0) 32 290-300. DOI: 10.1002/hed.21174 8. Fearon K., Strasser F., Anker S.D., Bosaeus I., Bruera E., Fainsinger R.L., Jatoi A., Loprinzi C., MacDonald N., Mantovani G.. Definition and Classification of Cancer Cachexia: An International Consensus. Lancet Oncol. (2011.0) 12 489-495. DOI: 10.1016/S1470-2045(10)70218-7 9. Webster J.M., Kempen L.J.A.P., Hardy R.S., Langen R.C.J.. Inflammation and Skeletal Muscle Wasting During Cachexia. Front. Physiol. (2020.0) 11 597675. DOI: 10.3389/fphys.2020.597675 10. Sun L., Quan X.-Q., Yu S.. An Epidemiological Survey of Cachexia in Advanced Cancer Patients and Analysis on Its Diagnostic and Treatment Status. Nutr. Cancer (2015.0) 67 1056-1062. DOI: 10.1080/01635581.2015.1073753 11. Bossi P., Delrio P., Mascheroni A., Zanetti M.. The Spectrum of Malnutrition/Cachexia/Sarcopenia in Oncology According to Different Cancer Types and Settings: A Narrative Review. Nutrients (2021.0) 13. DOI: 10.3390/nu13061980 12. Brown J.L., Lee D.E., Rosa-Caldwell M.E., Brown L.A., Perry R.A., Haynie W.S., Huseman K., Sataranatarajan K., Van Remmen H., Washington T.A.. Protein Imbalance in the Development of Skeletal Muscle Wasting in Tumour-Bearing Mice. J. Cachexia Sarcopenia Muscle (2018.0) 9 987-1002. DOI: 10.1002/jcsm.12354 13. Gao S., Carson J.A.. Lewis Lung Carcinoma Regulation of Mechanical Stretch-Induced Protein Synthesis in Cultured Myotubes. Am. J. Physiol. Cell Physiol. (2016.0) 310 C66-C79. DOI: 10.1152/ajpcell.00052.2015 14. McLean J.B., Moylan J.S., Andrade F.H.. Mitochondria Dysfunction in Lung Cancer-Induced Muscle Wasting in C2C12 Myotubes. Front. Physiol. (2014.0) 5 503. DOI: 10.3389/fphys.2014.00503 15. Sun R., Zhang S., Hu W., Lu X., Lou N., Yang Z., Chen S., Zhang X., Yang H.. Valproic Acid Attenuates Skeletal Muscle Wasting by Inhibiting C/EBPβ-Regulated Atrogin1 Expression in Cancer Cachexia. Am. J. Physiol. Cell Physiol. (2016.0) 311 C101-C115. DOI: 10.1152/ajpcell.00344.2015 16. Sustova H., De Feudis M., Reano S., Alves Teixeira M., Valle I., Zaggia I., Agosti E., Prodam F., Filigheddu N.. Opposing Effects of 25-Hydroxy- and 1α,25-Dihydroxy-Vitamin D3 on pro-Cachectic Cytokine-and Cancer Conditioned Medium-Induced Atrophy in C2C12 Myotubes. Acta Physiol. (2019.0) 226 e13269. DOI: 10.1111/apha.13269 17. Tisdale M.J.. Are Tumoral Factors Responsible for Host Tissue Wasting in Cancer Cachexia?. Future Oncol. (2010.0) 6 503-513. DOI: 10.2217/fon.10.20 18. Yano C.L., Ventrucci G., Field W.N., Tisdale M.J., Gomes-Marcondes M.C.C.. Metabolic and Morphological Alterations Induced by Proteolysis-Inducing Factor from Walker Tumour-Bearing Rats in C2C12 Myotubes. BMC Cancer (2008.0) 8. DOI: 10.1186/1471-2407-8-24 19. Zhang G., Liu Z., Ding H., Miao H., Garcia J.M., Li Y.-P.. Toll-like Receptor 4 Mediates Lewis Lung Carcinoma-Induced Muscle Wasting via Coordinate Activation of Protein Degradation Pathways. Sci. Rep. (2017.0) 7 2273. DOI: 10.1038/s41598-017-02347-2 20. Guigni B.A., van der Velden J., Kinsey C.M., Carson J.A., Toth M.J.. Effects of Conditioned Media from Murine Lung Cancer Cells and Human Tumor Cells on Cultured Myotubes. Am. J. Physiol. Endocrinol. Metab. (2020.0) 318 E22-E32. DOI: 10.1152/ajpendo.00310.2019 21. Yaromina A., Quennet V., Zips D., Meyer S., Shakirin G., Walenta S., Mueller-Klieser W., Baumann M.. Co-Localisation of Hypoxia and Perfusion Markers with Parameters of Glucose Metabolism in Human Squamous Cell Carcinoma (HSCC) Xenografts. Int. J. Radiat. Biol. (2009.0) 85 972-980. DOI: 10.3109/09553000903232868 22. Meinzer A., Yaromina A., Hoinkis C., Zips D., Thames H.D., Petersen C., Baumann M.. Tumor Microenvironment during Fractionated Irradiation: First Results in UT-SCC-5 and UT-SCC-15 Tumors 23. Koi L., Bergmann R., Brüchner K., Pietzsch J., Pietzsch H.-J., Krause M., Steinbach J., Zips D., Baumann M.. Radiolabeled Anti-EGFR-Antibody Improves Local Tumor Control after External Beam Radiotherapy and Offers Theragnostic Potential. Radiother. Oncol. (2014.0) 110 362-369. DOI: 10.1016/j.radonc.2013.12.001 24. Lange M.J., Lasiter J.C., Misfeldt M.L.. Toll-like Receptors in Tonsillar Epithelial Cells. Int. J. Pediatr. Otorhinolaryngol. (2009.0) 73 613-621. DOI: 10.1016/j.ijporl.2008.12.013 25. Göttgens E.-L., Ansems M., Leenders W.P.J., Bussink J., Span P.N.. Genotyping and Characterization of HPV Status, Hypoxia, and Radiosensitivity in 22 Head and Neck Cancer Cell Lines. Cancers (2021.0) 13. DOI: 10.3390/cancers13051069 26. Cruz-Jentoft A.J., Landi F., Schneider S.M., Zúñiga C., Arai H., Boirie Y., Chen L.-K., Fielding R.A., Martin F.C., Michel J.-P.. Prevalence of and Interventions for Sarcopenia in Ageing Adults: A Systematic Review. Report of the International Sarcopenia Initiative (EWGSOP and IWGS). Age Ageing (2014.0) 43 748-759. DOI: 10.1093/ageing/afu115 27. Freire A.N., Guerra R.O., Alvarado B., Guralnik J.M., Zunzunegui M.V.. Validity and Reliability of the Short Physical Performance Battery in Two Diverse Older Adult Populations in Quebec and Brazil. J. Aging Health (2012.0) 24 863-878. DOI: 10.1177/0898264312438551 28. Pichard C., Kyle U.G.. Body Composition Measurements during Wasting Diseases. Curr. Opin. Clin. Nutr. Metab. Care (1998.0) 1 357-361. DOI: 10.1097/00075197-199807000-00008 29. Jackson W., Alexander N., Schipper M., Fig L., Feng F., Jolly S.. Characterization of Changes in Total Body Composition for Patients with Head and Neck Cancer Undergoing Chemoradiotherapy Using Dual-Energy x-Ray Absorptiometry. Head Neck (2014.0) 36 1356-1362. DOI: 10.1002/hed.23461 30. Kyle U.G., Genton L., Karsegard L., Slosman D.O., Pichard C.. Single Prediction Equation for Bioelectrical Impedance Analysis in Adults Aged 20–94 Years. Nutrition (2001.0) 17 248-253. DOI: 10.1016/S0899-9007(00)00553-0 31. Janssen I., Heymsfield S.B., Baumgartner R.N., Ross R.. Estimation of Skeletal Muscle Mass by Bioelectrical Impedance Analysis. J. Appl. Physiol. (2000.0) 89 465-471. DOI: 10.1152/jappl.2000.89.2.465 32. Baracos V.E., Reiman T., Mourtzakis M., Gioulbasanis I., Antoun S.. Body Composition in Patients with Non-Small Cell Lung Cancer: A Contemporary View of Cancer Cachexia with the Use of Computed Tomography Image Analysis. Am. J. Clin. Nutr. (2010.0) 91 1133S-1137S. DOI: 10.3945/ajcn.2010.28608C 33. Mourtzakis M., Prado C.M.M., Lieffers J.R., Reiman T., McCargar L.J., Baracos V.E.. A Practical and Precise Approach to Quantification of Body Composition in Cancer Patients Using Computed Tomography Images Acquired during Routine Care. Appl. Physiol. Nutr. Metab. (2008.0) 33 997-1006. DOI: 10.1139/H08-075 34. Kobayashi J., Tadokoro N., Watanabe M., Shinomiya M.. A Novel Method of Measuring Intra-Abdominal Fat Volume Using Helical Computed Tomography. Int. J. Obes. Relat. Metab. Disord. (2002.0) 26 398-402. DOI: 10.1038/sj.ijo.0801921 35. Mitsiopoulos N., Baumgartner R.N., Heymsfield S.B., Lyons W., Gallagher D., Ross R.. Cadaver Validation of Skeletal Muscle Measurement by Magnetic Resonance Imaging and Computerized Tomography. J. Appl. Physiol. (1998.0) 85 115-122. DOI: 10.1152/jappl.1998.85.1.115 36. Vesali R.F., Klaude M., Rooyackers O.E., TJäder I., Barle H., Wernerman J.. Longitudinal Pattern of Glutamine/Glutamate Balance across the Leg in Long-Stay Intensive Care Unit Patients. Clin. Nutr. (2002.0) 21 505-514. DOI: 10.1054/clnu.2002.0583 37. Klaude M., Fredriksson K., Tjäder I., Hammarqvist F., Ahlman B., Rooyackers O., Wernerman J.. Proteasome Proteolytic Activity in Skeletal Muscle Is Increased in Patients with Sepsis. Clin. Sci. (2007.0) 112 499-506. DOI: 10.1042/CS20060265 38. Zhang G., Jin B., Li Y.-P.. C/EBPβ Mediates Tumour-Induced Ubiquitin Ligase Atrogin1/MAFbx Upregulation and Muscle Wasting. EMBO J. (2011.0) 30 4323-4335. DOI: 10.1038/emboj.2011.292 39. Kitagawa M., Haji S., Amagai T.. High Serum Essential Amino Acids as a Predictor of Skeletal Muscle Depletion in Patients With Cachexia and Advanced Gastrointestinal Cancers. Nutr. Clin. Pract. (2017.0) 32 645-651. DOI: 10.1177/0884533617724742 40. Baracos V.E., Martin L., Korc M., Guttridge D.C., Fearon K.C.H.. Cancer-Associated Cachexia. Nat. Rev. Dis. Primers. (2018.0) 4 17105. DOI: 10.1038/nrdp.2017.105 41. Aversa Z., Pin F., Lucia S., Penna F., Verzaro R., Fazi M., Colasante G., Tirone A., Rossi Fanelli F., Ramaccini C.. Autophagy Is Induced in the Skeletal Muscle of Cachectic Cancer Patients. Sci. Rep. (2016.0) 6 30340. DOI: 10.1038/srep30340 42. Agustsson T., Wikrantz P., Rydén M., Brismar T., Isaksson B.. Adipose Tissue Volume Is Decreased in Recently Diagnosed Cancer Patients with Cachexia. Nutrition (2012.0) 28 851-855. DOI: 10.1016/j.nut.2011.11.026 43. Kim H.J., Kim H.J., Yun J., Kim K.H., Kim S.H., Lee S.-C., Bae S.B., Kim C.K., Lee N.S., Lee K.T.. Pathophysiological Role of Hormones and Cytokines in Cancer Cachexia. J. Korean Med. Sci. (2012.0) 27 128-134. DOI: 10.3346/jkms.2012.27.2.128 44. Lerner L., Tao J., Liu Q., Nicoletti R., Feng B., Krieger B., Mazsa E., Siddiquee Z., Wang R., Huang L.. MAP3K11/GDF15 Axis Is a Critical Driver of Cancer Cachexia. J. Cachexia Sarcopenia Muscle (2016.0) 7 467-482. DOI: 10.1002/jcsm.12077 45. Penafuerte C.A., Gagnon B., Sirois J., Murphy J., MacDonald N., Tremblay M.L.. Identification of Neutrophil-Derived Proteases and Angiotensin II as Biomarkers of Cancer Cachexia. Br. J. Cancer (2016.0) 114 680-687. DOI: 10.1038/bjc.2016.3 46. Loumaye A., de Barsy M., Nachit M., Lause P., Frateur L., van Maanen A., Trefois P., Gruson D., Thissen J.-P.. Role of Activin A and Myostatin in Human Cancer Cachexia. J. Clin. Endocrinol. Metab. (2015.0) 100 2030-2038. DOI: 10.1210/jc.2014-4318 47. Batista M.L., Olivan M., Alcantara P.S.M., Sandoval R., Peres S.B., Neves R.X., Silverio R., Maximiano L.F., Otoch J.P., Seelaender M.. Adipose Tissue-Derived Factors as Potential Biomarkers in Cachectic Cancer Patients. Cytokine (2013.0) 61 532-539. DOI: 10.1016/j.cyto.2012.10.023 48. Burney B.O., Hayes T.G., Smiechowska J., Cardwell G., Papusha V., Bhargava P., Konda B., Auchus R.J., Garcia J.M.. Low Testosterone Levels and Increased Inflammatory Markers in Patients with Cancer and Relationship with Cachexia. J. Clin. Endocrinol. Metab. (2012.0) 97 E700-E709. DOI: 10.1210/jc.2011-2387 49. Felix K., Fakelman F., Hartmann D., Giese N.A., Gaida M.M., Schnölzer M., Flad T., Büchler M.W., Werner J.. Identification of Serum Proteins Involved in Pancreatic Cancer Cachexia. Life Sci. (2011.0) 88 218-225. DOI: 10.1016/j.lfs.2010.11.011 50. Krzystek-Korpacka M., Matusiewicz M., Diakowska D., Grabowski K., Blachut K., Kustrzeba-Wojcicka I., Banas T.. Impact of Weight Loss on Circulating IL-1, IL-6, IL-8, TNF-Alpha, VEGF-A, VEGF-C and Midkine in Gastroesophageal Cancer Patients. Clin. Biochem. (2007.0) 40 1353-1360. DOI: 10.1016/j.clinbiochem.2007.07.013 51. Uz U., Eskiizmir G.. Association Between Interleukin-6 and Head and Neck Squamous Cell Carcinoma: A Systematic Review. Clin. Exp. Otorhinolaryngol. (2021.0) 14 50-60. DOI: 10.21053/ceo.2019.00906 52. Chang K.-P., Kao H.-K., Wu C.-C., Fang K.-H., Chang Y.-L., Huang Y.-C., Liu S.-C., Cheng M.-H.. Pretreatment Interleukin-6 Serum Levels Are Associated with Patient Survival for Oral Cavity Squamous Cell Carcinoma. Otolaryngol. Head Neck Surg. (2013.0) 148 786-791. DOI: 10.1177/0194599813478573 53. Hao W., Zhu Y., Zhou H.. Prognostic Value of Interleukin-6 and Interleukin-8 in Laryngeal Squamous Cell Cancer. Med. Oncol. (2013.0) 30 333. DOI: 10.1007/s12032-012-0333-6 54. De Schutter H., Landuyt W., Verbeken E., Goethals L., Hermans R., Nuyts S.. The Prognostic Value of the Hypoxia Markers CA IX and GLUT 1 and the Cytokines VEGF and IL 6 in Head and Neck Squamous Cell Carcinoma Treated by Radiotherapy +/− Chemotherapy. BMC Cancer (2005.0) 5. DOI: 10.1186/1471-2407-5-42 55. Moses A.G.W., Maingay J., Sangster K., Fearon K.C.H., Ross J.A.. Pro-Inflammatory Cytokine Release by Peripheral Blood Mononuclear Cells from Patients with Advanced Pancreatic Cancer: Relationship to Acute Phase Response and Survival. Oncol. Rep. (2009.0) 21 1091-1095. DOI: 10.3892/or_00000328 56. Ando K., Takahashi F., Kato M., Kaneko N., Doi T., Ohe Y., Koizumi F., Nishio K., Takahashi K.. Tocilizumab, a Proposed Therapy for the Cachexia of Interleukin6-Expressing Lung Cancer. PLoS ONE (2014.0) 9. DOI: 10.1371/journal.pone.0102436 57. Richey L.M., George J.R., Couch M.E., Kanapkey B.K., Yin X., Cannon T., Stewart P.W., Weissler M.C., Shores C.G.. Defining Cancer Cachexia in Head and Neck Squamous Cell Carcinoma. Clin. Cancer Res. (2007.0) 13 6561-6567. DOI: 10.1158/1078-0432.CCR-07-0116 58. Astradsson T., Sellberg F., Berglund D., Ehrsson Y.T., Laurell G.F.E.. Systemic Inflammatory Reaction in Patients With Head and Neck Cancer—An Explorative Study. Front. Oncol. (2019.0) 9 1177. DOI: 10.3389/fonc.2019.01177 59. Brady G., O’Regan E., Miller I., Ogungbowale A., Kapas S., Crean S.J.. Serum Levels of Insulin-like Growth Factors (IGFs) and Their Binding Proteins (IGFBPs), -1, -2, -3, in Oral Cancer. Int. J. Oral. Maxillofac. Surg. (2007.0) 36 259-262. DOI: 10.1016/j.ijom.2006.09.009 60. Wu X., Zhao H., Do K.-A., Johnson M.M., Dong Q., Hong W.K., Spitz M.R.. Serum Levels of Insulin Growth Factor (IGF-I) and IGF-Binding Protein Predict Risk of Second Primary Tumors in Patients with Head and Neck Cancer. Clin. Cancer Res. (2004.0) 10 3988-3995. DOI: 10.1158/1078-0432.CCR-03-0762 61. Dunn S.E., Kari F.W., French J., Leininger J.R., Travlos G., Wilson R., Barrett J.C.. Dietary Restriction Reduces Insulin-like Growth Factor I Levels, Which Modulates Apoptosis, Cell Proliferation, and Tumor Progression in P53-Deficient Mice. Cancer Res. (1997.0) 57 4667-4672. PMID: 9354418 62. Martín A.I., Priego T., Moreno-Ruperez Á., González-Hedström D., Granado M., López-Calderón A.. IGF-1 and IGFBP-3 in Inflammatory Cachexia. Int. J. Mol. Sci. (2021.0) 22. DOI: 10.3390/ijms22179469 63. Abdelmoez A.M., Sardón Puig L., Smith J.A.B., Gabriel B.M., Savikj M., Dollet L., Chibalin A.V., Krook A., Zierath J.R., Pillon N.J.. Comparative Profiling of Skeletal Muscle Models Reveals Heterogeneity of Transcriptome and Metabolism. Am. J. Physiol.-Cell Physiol. (2020.0) 318 C615-C626. DOI: 10.1152/ajpcell.00540.2019 64. Pettersen K., Andersen S., Degen S., Tadini V., Grosjean J., Hatakeyama S., Tesfahun A.N., Moestue S., Kim J., Nonstad U.. Cancer Cachexia Associates with a Systemic Autophagy-Inducing Activity Mimicked by Cancer Cell-Derived IL-6 Trans-Signaling. Sci. Rep. (2017.0) 7 2046. DOI: 10.1038/s41598-017-02088-2 65. Bohnert K.R., Gallot Y.S., Sato S., Xiong G., Hindi S.M., Kumar A.. Inhibition of ER Stress and Unfolding Protein Response Pathways Causes Skeletal Muscle Wasting during Cancer Cachexia. FASEB J. (2016.0) 30 3053-3068. DOI: 10.1096/fj.201600250RR 66. Puppa M.J., Gao S., Narsale A.A., Carson J.A.. Skeletal Muscle Glycoprotein 130’s Role in Lewis Lung Carcinoma-Induced Cachexia. FASEB J. (2014.0) 28 998-1009. DOI: 10.1096/fj.13-240580 67. Seto D.N., Kandarian S.C., Jackman R.W.. A Key Role for Leukemia Inhibitory Factor in C26 Cancer Cachexia. J. Biol. Chem. (2015.0) 290 19976-19986. DOI: 10.1074/jbc.M115.638411 68. Silva K.A.S., Dong J., Dong Y., Dong Y., Schor N., Tweardy D.J., Zhang L., Mitch W.E.. Inhibition of Stat3 Activation Suppresses Caspase-3 and the Ubiquitin-Proteasome System, Leading to Preservation of Muscle Mass in Cancer Cachexia. J. Biol. Chem. (2015.0) 290 11177-11187. DOI: 10.1074/jbc.M115.641514 69. Ottenheijm C.A.C., Heunks L.M.A., Li Y.-P., Jin B., Minnaard R., van Hees H.W.H., Dekhuijzen P.N.R.. Activation of the Ubiquitin-Proteasome Pathway in the Diaphragm in Chronic Obstructive Pulmonary Disease. Am. J. Respir. Crit. Care Med. (2006.0) 174 997-1002. DOI: 10.1164/rccm.200605-721OC 70. Léger B., Vergani L., Sorarù G., Hespel P., Derave W., Gobelet C., D’Ascenzio C., Angelini C., Russell A.P.. Human Skeletal Muscle Atrophy in Amyotrophic Lateral Sclerosis Reveals a Reduction in Akt and an Increase in Atrogin-1. FASEB J. (2006.0) 20 583-585. DOI: 10.1096/fj.05-5249fje 71. Cohen S., Brault J.J., Gygi S.P., Glass D.J., Valenzuela D.M., Gartner C., Latres E., Goldberg A.L.. During Muscle Atrophy, Thick, but Not Thin, Filament Components Are Degraded by MuRF1-Dependent Ubiquitylation. J. Cell Biol. (2009.0) 185 1083-1095. DOI: 10.1083/jcb.200901052 72. Dworzak F., Ferrari P., Gavazzi C., Maiorana C., Bozzetti F.. Effects of Cachexia Due to Cancer on Whole Body and Skeletal Muscle Protein Turnover. Cancer (1998.0) 82 42-48. DOI: 10.1002/(SICI)1097-0142(19980101)82:1<42::AID-CNCR5>3.0.CO;2-M 73. White J.P., Baynes J.W., Welle S.L., Kostek M.C., Matesic L.E., Sato S., Carson J.A.. The Regulation of Skeletal Muscle Protein Turnover during the Progression of Cancer Cachexia in the Apc(Min/+) Mouse. PLoS ONE (2011.0) 6. DOI: 10.1371/journal.pone.0024650 74. Goodman C.A., Frey J.W., Mabrey D.M., Jacobs B.L., Lincoln H.C., You J.-S., Hornberger T.A.. The Role of Skeletal Muscle MTOR in the Regulation of Mechanical Load-Induced Growth. J. Physiol. (2011.0) 589 5485-5501. DOI: 10.1113/jphysiol.2011.218255
--- title: Performance Evaluation of Quantum-Based Machine Learning Algorithms for Cardiac Arrhythmia Classification authors: - Zeynep Ozpolat - Murat Karabatak journal: Diagnostics year: 2023 pmcid: PMC10047100 doi: 10.3390/diagnostics13061099 license: CC BY 4.0 --- # Performance Evaluation of Quantum-Based Machine Learning Algorithms for Cardiac Arrhythmia Classification ## Abstract The electrocardiogram (ECG) is the most common technique used to diagnose heart diseases. The electrical signals produced by the heart are recorded by chest electrodes and by the extremity electrodes placed on the limbs. Many diseases, such as arrhythmia, cardiomyopathy, coronary heart disease, and heart failure, can be diagnosed by examining ECG signals. The interpretation of these signals by experts may take a long time, and there may be differences between expert interpretations. Since technological developments are intertwined with the medical sciences, computer-assisted diagnostic methods have recently come forward. In computer science, machine learning techniques are often preferred for automatic detection. Quantum-based structures have emerged to increase the machine learning algorithm’s speed and classification performance. In this study, a quantum-based machine learning algorithm is applied to classify heart rhythms. The ECG properties were converted to qubit structure using principal component analysis (PCA). The resulting qubits are classified using the quantum support vector machine (QSVM) algorithm. Quantum computer simulation over Qiskit was used for classification studies. Within the scope of experimental studies, comparisons between classical SVM and QSVM were made using different data amounts and qubit numbers. In the results of the analysis, classical SVM achieved $86.96\%$ accuracy, and QSVM achieved $84.64\%$ accuracy. Despite the fact that the entire dataset was not used due to various limitations, these successful performances were achieved. Classification of medical data such as that from ECG has shown that quantum-based machine learning frameworks perform well despite current resource constraints. In this respect, the study includes essential contributions to the use of quantum-based machine learning methods on signal data in medicine. ## 1. Introduction Early diagnosis of cardiovascular diseases is critical in determining treatment and preventing important risks, such as mortality. Rhythm disorders (arrhythmias) occur in the heart due to cardiovascular diseases. Electrocardiography (ECG) is the general diagnostic method used to diagnose arrhythmias. Electrical signals of the heart are recorded with an ECG and interpreted by experts through observation. One of the main reasons that many diseases cannot be treated is the lack of timely intervention. Due to factors such as a rise in patient numbers, the inadequate quality of medical equipment, and a shortage of doctors, early diagnosis may be delayed. In addition to these parameters, the results of the analyses take a long time [1]. Since the heart is one of the vital organs in the body, there is a great deal of research in computer science regarding heart disease. Li et al. [ 2] have developed a custom networking structure called a Beat-aligned Transformer (BaT) to take advantage of the repetitive features of ECG data. The concept of “deep learning” [3,4], in which both classification and feature inference have coexisted recently as a result of improvements in machine learning, has rapidly become widespread. In ECG analysis, deep learning architectures have also provided good performances. Baloglu et al. [ 5] applied a convolutional neural network (CNN) for the diagnosis of myocardial infarction (MI) by processing the 12-leads ECG signal. As a result of their analysis, high performance results were achieved. In another study for MI, Yıldırım et al. [ 6] developed a deep neural network (DNN) trained with surface ECG to detect clinical MI disease in people. The superiority of the developed model was demonstrated by comparing it with Q-wave analysis. Similarly, it has produced positive results in the treatment of atrial fibrillation (AF) caused by cardiovascular conditions [7,8]. Computational power in the advancement of machine learning methods is mostly based on hardware, and parallel software improvements produce significant impacts. The purpose of machine learning is to increase speed and performance. An outcome can be obtained more quickly if all potential solutions are calculated. One of the main reasons for entering the quantum world is that a resolution can be made quickly by following a different path for each possibility. Quantum physics, one of the important subjects of physics, is a field that contains theories about the entire subatomic microscopic particle system. It has been researched from a very different perspective in recent years as a result of its contributions to the world of informatics [9]. Unlike the bits used in classical computers, qubits are used in quantum computers. A qubit can take the values 1 or 0, or it can be both 1 and 0 simultaneously. As a result, quantum computers can compute multiple probabilities at the same time [10]. In recent years, quantum-based machine learning algorithms have rapidly become popular in the literature. Maheshwari et al. [ 11] evaluated analysis results by applying both classical and machine learning algorithms to diabetic patient data. Gupta et al. [ 12] compared deep learning (DL) and quantum machine learning (QML) algorithms in another study on diabetes. Zhang and Ni [13] have suggested in their research that some of the supervised and unsupervised machine learning algorithms based on the quantum circuit model focus on the quantum base. As a result of their studies, they determined that quantum algorithms show a speed-up in results compared to their classical versions. Blance and Spannowsky [14] aimed to increase performance in solving classification problems by combining quantum computing methods with classical neural network techniques. In this study, the use of quantum-based machine learning algorithms in ECG analysis, which is one of the important problems in the medical field, is proposed. For this purpose, an ECG dataset [15] containing four different rhythms was analyzed using both the classical and quantum-based support vector machine (SVM) methods. For quantum SVM, qubits were created with principal component analysis (PCA), a size reduction algorithm, in parallel with the existing hardware resources. Comparisons between the performances of classical SVM and quantum support vector machine (QSVM) were examined, with both the number of data points and qubit numbers increasing at different rates. The organizational structure of this study is as follows. In Section 2, the materials and methods are introduced. In Section 3, details about the experimental studies are given. Section 4 and Section 5 conclude our study; this includes the discussion and results, respectively. ## 2. Materials and Methods In this article, analyses were performed with a quantum computer simulator using a dataset of ECG signals. The dataset was converted from bit form to qubit form [16]. Following the data preparation for analysis, a classification procedure was carried out using the QSVM method from open-source Qiskit codes [17]. The classical SVM algorithm was applied to the dataset in qubit form, which was labeled by reducing its size. The purpose of this was to compare the performances of the QSVM and SVM algorithms on the same data. A block representation of the materials and methods used in the study is given in Figure 1. ## 2.1. Arrhythmia Dataset In this study, the dataset created by Zheng et al., which contains ECG data from 10,588 patients, was used [15]. This dataset was created from the Chapman University and Shaoxing and Ningbo People’s Hospital (Chapman) database. The dataset includes raw ECG signals from 12 leads and 11 clinically obtained ECG features. These features are: ventricular rate (VR), atrial rate (AR), QRS duration (QRSD), Q interval, QT corrected, R axis, T axis, QRS count, Q onset, Q offset, and T offset. The noise-free ECG dataset consists of 12-lead ECG signals sampled at 500 Hz and categorized by 11 rhythm classes. These are atrial flutter (AF), atrial fibrillation (AFIB), atrial tachycardia (AT), atrioventricular node reentrant tachycardia (AVNRT), atrioventricular reentrant tachycardia (AVRT), sinus irregularity (SI), sinus atrium to atrial wandering rhythm (SAAWR), sinus bradycardia (SB), sinus rhythm (SR), sinus tachycardia (SINT), and supraventricular tachycardia (SVT). Murat et al. [ 18], using this data in their study, created 4 different class labels by converting classes with a small number of patients into groups that are related to each other. Information about these four rhythm classes combined is given in Table 1. ## 2.2. Proposed Method This study employs the recently popular quantum-based machine learning approaches in classifying heart arrhythmias. For this purpose, first, size reduction was performed on a determined dataset using the PCA technique. The features reduced by the dimension reduction technique were converted to qubit format and used in the classification stage of the QSVM algorithm. Apart from the QSVM algorithm, there are quantum classification algorithms such as quantum neural network (QNN) [19,20], quantum K-nearest neighbors (Q-KNN) [21,22], and quantum means (Q-Means) [23]. The QSVM algorithm is preferred in this study, because SVM is mainly used in classical algorithms in ECG classification. Since it is aimed at comparing classical and quantum-based ML, QSVM has been determined as the most suitable algorithm. A block representation of the proposed method within the scope of the study is given in Figure 2. ## 2.2.1. Principal Component Analysis (PCA) One of the earliest statistical tools, PCA is a method for converting oversized data into lower-dimensional data to reduce cost and speed. PCA maintains changes in data, allowing data to be expressed, at least at a loss, with fewer components than in its original state. In doing so, it aims to determine the best transformation and ensure that all the resulting components are independent of each other. In this direction, the variance of the data, eigenvalues, and eigenvectors are used while making calculations [24]. As a result of applied mathematical operations, it is ensured that the original dataset is expressed with different axes. Thus, more efficient analyses can be made, as the data is provided from a different perspective. ## 2.2.2. Quantum Support Vector Machine (QSVM) The orientation of computer science toward the quantum world has paved the way for the use of quantum-based programming in the classification stages. To run the SVM algorithm on a quantum computer, the algorithm must be rescheduled according to quantum rules [25]. The QSVM algorithm, a quantum adaptation of SVM, performs the computations for the basic SVM using the laws of quantum mechanics. Whereas classical SVM requires a graphics processing unit (GPU) or central process unit (CPU) to increase performance, QSVM uses the power of quantum software. When performing QML operations, classical data are converted into quantum data (qubits) to be used in quantum computers. Then, the processing steps required by the QML algorithm are applied. The result obtained is returned in the classical form [26]. ## 2.2.3. Experimental Setups The ECG dataset used consists of 10,588 pieces of data. Since there is no access to quantum computers in the real environment, the dataset was run using the Qiskit framework in the existing computer architecture through the Anaconda package program. Since a real quantum computer cannot be used, the large amount of data creates a disadvantage in execution time. Data with a reduced original number of data are called a data case. The number of data is reduced using 7 different test sizes for data cases. The quantum computer system is still in development. Therefore, it can serve with limited qubits. PCA, one of the conversion methods, was used to avoid exceeding the qubit limit while preserving the structure of the features. SVM and QSVM performances were compared for 5 data cases: 3, 5, 7, 9, and 11. The attribute numbers given here as dim also indicate the qubit numbers simultaneously. The dataset, which initially had 11 dimensions, was reduced to 4 dimensions by applying PCA. Quantum simulation is provided in the classical computer with ZZFeatureMap [27], which is used for the qubits to enter the entanglement state. The circuit model of ZZFeatureMap is given in Figure 3a. A quantum circuit model for the case in which the number of qubits is determined to be 3 is shown in Figure 3b. In this study, the Qiskit Library was used for the QSVM implementation [26]. Qiskit is an open-source, quantum-computing environment developed by IBM. Thanks to the Qiskit library, quantum experiments can be run on classical computers with quantum simulations. The Qiskit environment is used with the Python programming language. Various libraries must be added to Qiskit to perform quantum computations [28]. These libraries include Qiskit Terra, Qiskit Aer, Qiskit Ignis, Qiskit Nature, Qiskit Machine Learning, Qiskit Finance, and Qiskit Optimization. Many of these libraries were used during this study. Computer features used in the experimental studies are as follows: an Intel(R) Xeon(R) W-2245 with CPU@ 3.90GHz 3.91 processor, 32 GB RAM, and a NVIDIA Quadro RTX 4000 video card. ## 3. Experimental Results This section presents the performance results of the QSVM method on heart rate data. QSVM and SVM algorithms are compared with two different scenarios, data states, and qubit numbers. In the first scenario planned, the performances of the QSVM algorithm were analyzed using different qubit numbers. Performance comparisons were made with the performances of the SVM algorithm for the same data cases. In the second scenario, the ways in which the change in the data space affects the performance of the QSVM algorithm are observed. The results obtained here are compared with the SVM algorithm as in the first scenario. ## 3.1. Scenario 1: Different Number of Qubits PCA has been applied to 11 features (dimensions) of the ECG signals in the dataset used in this study and has been made available for the QSVM algorithm. The “feature_and_label_transform” plugin in Qiskit was used to re-label qubits after the PCA application. As a result of the applications, the dataset was reduced to dim 3, dim 5, dim 7, dim 9, and dim 11. The dimensions achieved after these reductions now constitute qubits. Figure 4 below gives the analysis visuals for the SVM and QSVM algorithms in different data states for four different qubit values: Q [3], Q [5], Q [7], and Q [9]. When Table 2 is examined, it is determined that the QSVM performance is lower than or nearly equal to the general SVM. Among the results obtained for five qubits, it is seen that the QSVM algorithm is superior to SVM in the case of 3133 pieces of data. Even though there is only a very slight difference in this instance, it is projected that the QSVM algorithm will perform better under the right circumstances. Table 2 was obtained with average values based on 10 different cases. When the results in the Q [5] analyses for the 3133 pieces of data are examined, the QSVM achieved a performance of $80.90\%$ accuracy, whereas the SVM showed a result of $78.84\%$ accuracy under the same conditions. The confusion matrix of this situation is shown in Figure 5. In Table 3, the precision, sensitivity, specificity, and F1 score performance metrics for the provided confusion matrices are given. When examining the tables and figures provided, performance increases as the amount of data increases. SVM achieved $83.17\%$ accuracy, and QSVM achieved $82.73\%$ accuracy as the highest performance (see Table 2) for Qubit = 9. The confusion matrix obtained from the QSVM algorithm is given in Figure 6a. The lowest performance is observed when the number of data cases and qubits is the least. In the case of data number 209 for Qubit = 3, the worst results were obtained, with $65.09\%$ accuracy with the SVM and $59.51\%$ accuracy with the QSVM (See Table 2). The confusion matrix of these values is given in Figure 6b. The information, including the confusion matrices’ performance metrics, is shown in Table 4. When analyzing the results, the reduction in qubits significantly affects the performance rate. As the number of qubits and size increase, the data attributes become clearer. This has increased the performance. At the same time, the increase in the sample used for the analysis also positively affects the performance. ## 3.2. Scenario 2: Different Amount of Data The arrhythmia dataset used in this study consists of 10588 pieces of patient data. The analysis takes a very long time due to the quantity of data. For this reason, data cases have been created using seven different test sizes while designing the data to obtain faster results by reducing the amount of data. Information about these data cases is given in Table 2. These results were obtained by calculating the mean and standard deviation of 10 different random state values. Figure 7 shows the performance variation in the qubits for each data case. When Table 2 and Figure 7 above are examined, the highest performance for Data Case 7 was observed as 83.17 ± 1.38 with the SVM and 82.73 ± 1.13 with the QSVM in Q [9]. Table 2 shows the performance of Q [9] in bold in all data cases. When the number of qubits is at its maximum and the number of data increases, the performance gradually increases, as the sample space to be used in the classification increases, as shown in Table 2. The main reason that the analysis could not be conducted with the total size of the dataset is that the studies take longer as the amount of data increases. For this reason, the amount of data was increased gradually. The experiment was carried out with a maximum of 3133 pieces of data. Further data analysis was not possible due to hardware restrictions. ## 4. Discussion In this study, quantum-based machine learning algorithms are used to recognize heart rhythm classes automatically. Since quantum technology is a newly developing field, it has not been possible to develop a new algorithm due to structural deficiencies. The main purpose of the article is to observe the effects of parameters such as the qubit and data number on the performance using existing techniques. The results of various studies on similar datasets and the proposed method are compared in Table 5. Some of the studies in Table 5 are given performance comparison purposes, because they use the same dataset. In a study by Aziz et al. [ 29], in the SVM classification made for the SPNH database, the highest performance of $84.2\%$ accuracy was obtained in the PR + RT + Age + Sex classes compared to other combinations. In MLP, on the other hand, $90.7\%$ success was achieved under the same conditions. In an article by Sepahvand et al. [ 30], the model proposed by the authors is a CNN model, which is the teacher and student model. As a result of their studies, they obtained $98.96\%$ accuracy in the teacher model and $98.13\%$ accuracy in the student mode for seven rhythm classes. Faust et al. [ 31] achieved $99.98\%$ success in the SPNH dataset with the ResNet deep learning algorithm used by the authors in their study. Dhananjay et al. [ 32] compared classical classification algorithms as well as their proposed model, the CatBoost model, in their study. Whereas $71\%$ success was achieved with SVM, the success rate was $99\%$ with the method suggested by the authors. Murat et al. [ 18] used deep learning algorithms to reduce property sizes using PCA with the SPNH dataset. They achieved a success rate of $84.06\%$ in the SVM algorithm. Baygin et al. [ 33] presented a new classification model for the classification of ECG data, also affected by the homomorphically irreducible tree (HIT) problem with the SPNH dataset. The model they installed consisted of HIT model creation, maximum absolute pooling (MAP), Chi2 selective, and the SVM algorithm in classification. Their success rate was $97.18\%$. According to Table 5, it is observed that the QSVM algorithm performs poorly compared to other studies. The entire Chapman database could not be run in the QSVM algorithm, as the analysis took too long due to hardware deficiencies. Although only about $30\%$ of the dataset was used in the study, the SVM algorithm achieved $86.96\%$ success (for Data Case 2—Q [11]), and the QSVM algorithm achieved $84.64\%$ success (for Data Case 7—Q [9]). Though the runtime takes hours for the case in which QSVM achieves $84.64\%$ success, the SVM algorithm gives results in about 2 s under the same conditions (Data Case 7—Q [9]). These results show that the simulation environment creates a disadvantage in terms of time in quantum-based algorithms. However, it is clear that the QSVM algorithm competes with the classical SVM in terms of accuracy. To the best of the authors’ knowledge, this study is the first to classify an ECG dataset using the QSVM algorithm. It aims to form a basis for future studies in the field of ECG. In addition to adding to the limited studies using QML in the literature, this study presents comparisons with classical SVM using the quantum-based SVM algorithm, which has not been used before in the classification of ECG data. The research shows that the QSVM method offers a comparable performance to the traditional SVM technique. It is predicted that the increase in the number of qubits, known as the size, and the increase in the amount of data in the classical environment will positively affect the algorithm. It is thought that this performance will become more competitive when the existing deficiencies are eliminated, and the entire dataset is run. For these reasons, the QSVM algorithm gives promising results in ECG diagnosis. The main limitations of this study include the following. An IBM Q computer could not be used in this study due to the preprocessing processes applied. Instead, analyses were performed on a classical computer by creating a quantum environment. PCA was used because it has a qubit limitation. Due to the structural features of PCA, there may be a loss of features while reducing the size. This can negatively affect performance. The purpose of using quantum computers is basically to provide acceleration. However, the desired level could not be reached at the time of calculation, since the adaptation process with today’s computers has not yet been realized. Though the increase in the number of data causes an increase in performance, the results of the analyses take longer than with the classical algorithms. Since a quantum-based algorithm was used in this study, the entire dataset could not be used in the analysis due to hardware deficiencies. The primary purpose of this study was to demonstrate the usability of quantum machine learning algorithms that are under development for use with medical data such as ECG data. The analysis results given here were obtained using the Qiskit simulation environment. Analyses could not be performed in the real circuit due to issues with the use of the Noisy Intermediate-Scale Quantum (NISQ). Although a low performance was achieved according to state-of-the-arts studies, it is thought that this performance will improve when various limitations are overcome. If the same analyses were performed on the actual circuit, faster execution would probably be possible. The limitations will be eliminated in future studies, and analyses will be applied in the NISQ environment. Comparisons are limited due to the fact that we do not have the source codes and parameter values of the studies conducted on the same dataset [18,29,30,31,32,33]. For example, the effects of the amount of data on other methods have not been clearly demonstrated. In future studies, it is estimated that if the deficiencies of hardware are eliminated, and the entire dataset is used, both algorithms will provide better performances close to those of the classical studies existing. For performance comparisons, QML algorithms other than QSVM should also be used. It is known that the IBM Q computer is advantageous in terms of time in different datasets. After the Chapman dataset is made suitable, we aim to carry out analyses in an IBM Q real computer environment. After these studies, a detailed comparison with the results of the simulation environment will be presented. ## 5. Conclusions In this study, quantum machine learning algorithms were used to classify arrhythmias. For a quantum-based machine learning algorithm to be applied, the dataset must first be converted to qubit format, known as quantum bits. A dimension reduction method, principal component analysis (PCA), has been applied. Reduced size classes were converted to qubit form with a converter. ECG features for 10,588 pieces of patient data were used in a quantum simulator on a classical computer. Different rates of dimensionality reduction were applied for the bits converted to qubit form. Classical SVM and QSVM algorithms were applied to this new qubit-format dataset, and the performances were compared. The QSVM algorithm’s performance seemed comparable to that of the traditional SVM. This performance of the QVSM on a limited number of ECG data is an important step in QML algorithms. Future research should aim to compare outcomes when various QML algorithms are applied to ECG signals. ## References 1. Ogur N.B., Ceken C.. Real Time Data Analytics Architecture for ECG. Proceedings of the 2018 3rd International Conference on Computer Science and Engineering (UBMK) 286-291 2. Li X., Li C., Wei Y., Sun Y., Wei J., Li X., Qian B.. BaT: Beat-aligned Transformer for Electrocardiogram Classification. Proceedings of the 2021 IEEE International Conference on Data Mining (ICDM) 320-329 3. Goodfellow I., Bengio Y., Courville A.. Deep Learning (2016.0) 4. LeCun Y., Bengio Y., Hinton G.. Deep learning. Nature (2015.0) 521 436-444. DOI: 10.1038/nature14539 5. Baloglu U.B., Talo M., Yildirim O., San Tan R., Acharya U.R.. Classification of myocardial infarction with multi-lead ECG signals and deep CNN. Pattern Recognit. Lett. (2019.0) 122 23-30. DOI: 10.1016/j.patrec.2019.02.016 6. Yildirim O., Baloglu U.B., Talo M., Ganesan P., Tung J.S., Kang G., Rogers A.J.. Deep Neural Network Trained on Surface ECG Improves Diagnostic Accuracy of Prior Myocardial Infarction over Q Wave Analysis. Proceedings of the 2021 Computing in Cardiology (CinC) Volume 48 1-4 7. Murat F., Sadak F., Yildirim O., Talo M., Murat E., Karabatak M., Acharya U.R.. Review of deep learning-based atrial fibrillation detection studies. Int. J. Environ. Res. Public Health (2021.0) 18. DOI: 10.3390/ijerph182111302 8. Hu Y., Feng T., Wang M., Liu C., Tang H.. Detection of Paroxysmal Atrial Fibrillation from Dynamic ECG Recordings Based on a Deep Learning Model. 2022, SSRN 4098696 9. Steane A.. Quantum computing. Rep. Prog. Phys. (1998.0) 61 117. DOI: 10.1088/0034-4885/61/2/002 10. Paparo G.D., Dunjko V., Makmal A., Martin-Delgado M.A., Briegel H.J.. Quantum speedup for active learning agents. Phys. Rev. X (2014.0) 4 031002. DOI: 10.1103/PhysRevX.4.031002 11. Maheshwari D., Garcia-Zapirain B., Sierra-Soso D.. Machine learning applied to diabetes dataset using Quantum versus Classical computation. Proceedings of the 2020 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT) 1-6 12. Gupta H., Varshney H., Sharma T.K., Pachauri N., Verma O.P.. Comparative performance analysis of quantum machine learning with deep learning for diabetes prediction. Complex Intell. Syst. (2021.0) 8 3073-3087. DOI: 10.1007/s40747-021-00398-7 13. Zhang Y., Ni Q.. Recent advances in quantum machine learning. Quantum Eng. (2020.0) 2 e34. DOI: 10.1002/que2.34 14. Blance A., Spannowsky M.. Quantum machine learning for particle physics using a variational quantum classifier. J. High Energy Phys. (2021.0) 2 1-20. DOI: 10.1007/JHEP02(2021)212 15. Zheng J., Zhang J., Danioko S., Yao H., Guo H., Rakovski C.. A 12-lead electrocardiogram database for arrhythmia research covering more than 10,000 patients. Sci. Data (2020.0) 7 1-8. DOI: 10.1038/s41597-020-0386-x 16. Source Code for Qiskit_Machine_Learning.Datasets.Iris 17. Aleksandrowicz G., Alexander T., Barkoutsos P., Bello L., Ben-Haim Y., Bucher D., Marques M.. Qiskit: An open-sourceframework for quantum computing. Qiskit (2019.0) 55-63 18. Murat F., Yildirim O., Talo M., Demir Y., Tan R.S., Ciaccio E.J., Acharya U.R.. Exploring deep features and ECG attributes to detect cardiac rhythm classes. Knowl.-Based Syst. (2021.0) 232 107473. DOI: 10.1016/j.knosys.2021.107473 19. Wan K.H., Dahlsten O., Kristjánsson H., Gardner R., Kim M.S.. Quantum generalisation of feedforward neural networks. NPJ Quantum Inf. (2017.0) 3 1-8. DOI: 10.1038/s41534-017-0032-4 20. Zhao J., Zhang Y.H., Shao C.P., Wu Y.C., Guo G.C., Guo G.P.. Building quantum neural networks based on a swap test. Phys. Rev. A (2019.0) 100 012334. DOI: 10.1103/PhysRevA.100.012334 21. Wiebe N., Kapoor A., Svore K.. Quantum algorithms for nearest-neighbor methods for supervised and unsupervised learning. arXiv (2019.0). DOI: 10.26421/QIC15.3-4-7 22. Ruan Y., Xue X., Liu H., Tan J., Li X.. Quantum algorithm for k-nearest neighbors classification based on the metric of hamming distance. Int. J. Theor. Phys. (2017.0) 56 3496-3507. DOI: 10.1007/s10773-017-3514-4 23. Kerenidis I., Landman J., Luongo A., Prakash A.. q-means: A quantum algorithm for unsupervised machine learning. Adv. Neural Inf. Process. Syst. (2019.0) 32 4136-4146 24. Karamizadeh S., Abdullah S.M., Manaf A.A., Zamani M., Hooman A.. An overview of principal component analysis. J. Signal Inf. Process. (2020.0) 4 173. DOI: 10.4236/jsip.2013.43B031 25. Yang J., Awan A.J., Vall-Llosera G.. Support vector machines on noisy intermediate scale quantum computers. arXiv (2019.0) 26. Rebentrost P., Mohseni M., Lloyd S.. Quantum support vector machine for big data classification. Phys. Rev. Lett. (2014.0) 113 130503. DOI: 10.1103/PhysRevLett.113.130503 27. ZZFeatureMap 28. Qiskit 0.42.0 Documentation 29. Aziz S., Ahmed S., Alouini M.S.. ECG-based machine-learning algorithms for heartbeat classification. Sci. Rep. (2021.0) 11 18738. DOI: 10.1038/s41598-021-97118-5 30. Sepahvand M., Abdali-Mohammadi F.. A novel method for reducing arrhythmia classification from 12-lead ECG signals to single-lead ECG with minimal loss of accuracy through teacher-student knowledge distillation. Inf. Sci. (2022.0) 593 64-77. DOI: 10.1016/j.ins.2022.01.030 31. Faust O., Kareem M., Ali A., Ciaccio E.J., Acharya U.R.. Automated arrhythmia detection based on RR intervals. Diagnostics (2021.0) 11. DOI: 10.3390/diagnostics11081446 32. Dhananjay B., Sivaraman J.. Analysis and classification of heart rate using CatBoost feature ranking model. Biomed. Signal Process. Control (2021.0) 68 102610. DOI: 10.1016/j.bspc.2021.102610 33. Baygin M., Tuncer T., Dogan S., Tan R.-S., Acharya U.R.. Automated arrhythmia detection with homeomorphically irre-ducible tree technique using more than 10,000 individual subject ECG records. Inf. Sci. (2021.0) 575 323-337. DOI: 10.1016/j.ins.2021.06.022
--- title: Transcriptomic Establishment of Pig Macrophage Polarization Signatures authors: - Jing Li - Teng Yuan - Anjing Zhang - Peidong Yang - Li He - Keren Long - Chuang Tang - Li Chen - Mingzhou Li - Lu Lu journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10047103 doi: 10.3390/cimb45030151 license: CC BY 4.0 --- # Transcriptomic Establishment of Pig Macrophage Polarization Signatures ## Abstract Macrophages are the foremost controllers of innate and acquired immunity, playing important roles in tissue homeostasis, vasculogenesis, and congenital metabolism. In vitro macrophages are crucial models for understanding the regulatory mechanism of immune responses and the diagnosis or treatment of a variety of diseases. Pigs are the most important agricultural animals and valuable animal models for preclinical studies, but there is no unified method for porcine macrophage isolation and differentiation at present; no systematic study has compared porcine macrophages obtained by different methods. In the current study, we obtained two M1 macrophages (M1_IFNγ + LPS, and M1_GM-CSF) and two M2 macrophages (M2_IL4 + IL10, and M2_M-CSF), and compared the transcriptomic profiles between and within macrophage phenotypes. We observed the transcriptional differences either between or within phenotypes. Porcine M1 and M2 macrophages have consistent gene signatures with human and mouse macrophage phenotypes, respectively. Moreover, we performed GSEA analysis to attribute the prognostic value of our macrophage signatures in discriminating various pathogen infections. Our study provided a framework to guide the interrogation of macrophage phenotypes in the context of health and disease. The approach described here could be used to propose new biomarkers for diagnosis in diverse clinical settings including porcine reproductive and respiratory syndrome virus (PRRSV), African swine fever virus (ASFV), Toxoplasma gondii (T. gondii), porcine circovirus type 2 (PCV2), Haemophilus parasuis serovar 4 (HPS4), Mycoplasma hyopneumoniae (Mhp), Streptococcus suis serotype 2 (SS2), and LPS from Salmonella enterica serotype minnesota Re 595. ## 1. Introduction Macrophages are important effectors in specific and non-specific immunity, functioning in the generation and defense of many diseases. Macrophages are derived from macrophage precursor cells and with high plasticity, including two main subtypes, M1 and M2 [1]. Macrophages exist in a variety of physiological and pathological processes, and the proportion of M1 and M2 dynamically changes [2,3,4]. M1 macrophages feature in the following aspects, producing pro-inflammatory cytokines, mediating resistance to pathogens, exhibiting strong microbicidal properties, and also contributing to tissue destruction [5,6,7,8]. Classically, M1 macrophages can be activated when cells receive stimuli such as IFNγ, LPS, as well as GM-CSF [9]. Phenotypically, M1 macrophages express high levels of MHC-II, CD68, CD80, and CD86, they are also characterized by an elevated ability to secrete cytokines such as IL1β, TNF, IL12, and IL18 [6,10,11]. In contrast, M2 macrophages are activated through a pathway opposite to the classical pathway, which responds to stimuli factors such as CSF-1, IL4, IL10, TGF-β, and IL13. M2 macrophages play a central role in responses to parasites, tissue remodeling, angiogenesis, and allergic diseases [4,12,13]. The polarization of M2 can hydrolyze L-ornithine to arginine 1 (ARG1), which is the basic amino acid that makes up proline and hydroxyproline [14]. Proline and hydroxyproline are essential amino acids of collagen, which is an important protein in tissue repair, which helps to form an external matrix related to tissue repair [15,16,17]. Owing to their crucial role in host immunity, in vitro macrophage models have been widely applied in basic research studies. Porcine macrophages are similar to human macrophages in that they have a wide range of pattern recognition receptors that detect pathogen-associated molecular patterns (PAMP) on pathogens [18]. The porcine biomedical model is ideal for many studies on human infection, inflammation, energy metabolism, and obesity [12,16,19,20,21,22]. In specific aspects, porcine is closer to human, compared to mice to human for example, the overlap degree of immune parameters of porcine and human is greater than $80\%$ [21,23,24,25,26,27]. However, there is no standard in vitro porcine macrophage model and the differences between macrophages induced by different methods are not clear. Here, we questioned whether cultured macrophages help define specific functional phenotypes encountered in disease and the reliability of isolation, differentiation, and culture of porcine macrophages. With this in mind, a reliable method for describing the isolation, differentiation, and culture of porcine bone marrow-derived macrophages (BMDM) can be regarded as a valuable tool for classifying and studying the defined subset of macrophages found in specific environments. Since different macrophage phenotypes are profoundly involved in the development and outcome of many microbial infected diseases, and are key cells in controlling normal physiological processes, we question whether a restricted set of gene signatures could be applied to define a particular functional phenotype encountered in the context of microbial infectious diseases. One of the most useful animal models for preclinical research is the pig. Thus, we applied RNA-seq to compare the transcriptome differences of porcine macrophages induced by different methods. By analyzing the correlation between the transcriptome of macrophages induced by different methods and the transcriptome of macrophages infected by different pathogens and at different stages of infection, a theoretical reference value was provided for the diagnosis and molecular mechanisms of swine disease, such as PRRSV, ASFV, T. gondii, and so on. ## 2.1. Animals Seven day old Duroc × (Landrace × Yorkshire) hybrid pigs (DLY) used in this study were obtained from the experimental farm of Sichuan Agricultural University (Ya’an Campus). The animal experiment was approved by the Experimental Animal Ethics Committee of Sichuan Agricultural University under permit number 20210167 and was performed following the guidelines for the management and use of laboratory animals. According to the IACUC guidelines, pigs were killed by bloodletting, and then the femurs were collected to separate BMDM. The femur was separated and used for further bone marrow cell isolation. ## 2.2. Cell Culture In this study, bone marrow cells were obtained by puncture and passed through a 40 μM cell strainer (FALCON, New York, NY, USA, 352340). After erythrocytes were removed by an ACK lysate kit (Gibco, Grand Island, Now York, NY, USA, A1049201), the mononuclear cells were resuspended and cultured with DMEM/F12 (Gibco, Grand Island, Now York, USA, 11330-0320) supplemented with $10\%$ heat-inactivated fetal bovine serum (FBS) (Gibco, Grand Island, Now York, NY, USA, 10099141C), 100 U/mL penicillin, and 100 mg/mL streptomycin (Gibco, Temecula, CA, USA, 030311B) (DMEM/F12 $10\%$ FBS) at 37 °C in $5\%$ CO2 humidified air. After 4 h, the unattached cells were enriched and seeded in a new flask for macrophage differentiation by the different stimuli detailed (M1_IFNγ + LPS: 100 ng/mL IFNγ and 20 ng/mL LPS, M1_GM-CSF: 20 ng/mL GM-CSF, M2_IL4 + IL10: 10 ng/mL IL10 and 10 ng/mL IL4, and M2_M-CSF: 20 ng/mL M-CSF) (Table 1). The experiments were performed in triplicate. ## 2.3. RNA-Seq After the induction, macrophage cells were collected and used for total RNA extraction using Trizol (Invitrogen, San Francisco, CA, USA, 15596026), and the RNA-seq libraries were constructed using the NEBNext® UltraTM RNA Library Prep Kit (NEB, Ipswich, MA, USA, 7530). The paired-end RNA-seq sequencing libraries were further sequenced by the Illumina Novaseq6000 platform (PE150), yielding 151Gb raw data and an average of 804 million 150-bp paired-end raw reads (Novogene Co. Ltd., Tianjing, China). Sequenced reads were aligned to the pig reference genome (Sus_scrofa.Sscrofa11.1.104.gtf and Sus_scrofa.Sscrofa11.1.cdna.all.fa.gz). Gene expression was quantified by using Kallistov0.44.0 [30] and obtaining read counts for each transcript. We standardize read counts to TPM (per million transcript readings) and differential gene expression analysis was performed using the Edge R package. KEGG and GO annotation were performed using the online tool Metascape and hub genes were identified by using Metascape (https://metascape.org/gp/index.html, accessed on 4 November 2022), KEGG pathway analysis, and GO biological process analysis was performed for different genes. In order to compare the protein functions of macrophages between different methods, the protein–protein interaction network (PPI) was constructed by String (https://string-db.org, accessed on 17 November 2022) and the key Hub genes were identified. Raw and processed RNA-seq data were deposited in the NCBI GSE202115. ## 2.4. Gene Set Enrichment Analysis (GSEA) and Network Construction To assay, whether our gene signatures (M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF) can discriminate macrophages infected with various pathogens (Table 2), we applied GSEA to explore the correlation between our macrophage signatures with pathogen-infected macrophages obtained from the publicly available Gene Expression Omnibus (GEO) NCBI database (http://www.ncbi.nlm.nih.gov/geo/, accessed on 30 November 2022) [31]. Since the GSEA method was originally developed for analyzing microarray data, we normalized the raw count for standard GSEA by TPM and transformed TPM into a GCT format gene expression data set. Genes were ranked based on the correlation between their expression and class distinction, by evaluating if an a priori-defined set of genes (M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF) were randomly distributed or were primarily associated with a tested class. ## 3.1. Comparing M1 with M2 to Generate Porcine Macrophage Gene Signatures To fully characterize the specificity of two macrophage phenotypes polarized by different methods, we applied RNA-seq and compared the transcriptomic commonalities and differences across phenotypes and methods within phenotype. The principal components analysis (PCA) plot indicated that the macrophages clearly separated between phenotypes and within phenotype (Figure 1A). DEGs were evaluated by GO and KEGG after being transformed into gene IDs. The BP (biological process) components of the GO annotations of DEGs were used to examine the functional enrichment of DEGs. To ascertain the connection between DEGs and signaling pathways, KEGG analysis was performed. The DEGs of M1 (M1 IFN + LPS, and M1 GM-CSF) compared to M2 (M2 IL4 + IL10, and M2 M-CSF) were found to be enriched in biological processes related to immune response, such as Cell activation, Regulation of cell activation, Inflammatory response, Innate immune response, and pathways such as Hematopoietic cell lineage, Cytokine–cytokine receptor interaction (Figure 1B, Supplementary Figure S2 and Additional file S1). Moreover, we observed that the upregulated DEGs mainly enriched into phagosome pathways, pathways in cancer, and disease-related pathways, while the down-regulated genes enriched into tissue development pathways. To verify the reliability of polarized macrophages, 49 classical macrophage marker genes were retrieved from the previous literature, containing 29 and 20 macrophage markers for M1 and M2, respectively (Supplementary Table S2). Consistently, all 29 classical M1 and 20 M2 markers are highly expressed in M1 and M2, respectively (Figure 1C, Additional file S1). Using a p value < 0.01 and absolute log2FC value > 1 as criteria, 730 differential expressed genes (DEGs) were identified. Comparing M1 to M2 620 and 110 genes were significantly upregulated and downregulated, respectively (Additional file S1). To examine more closely the correlations between the genes of the core response to M1 macrophages (M1_IFNγ + LPS, and M1_GM-CSF) and M2 (M2_IL4 + IL10, and M2_M-CSF) macrophages, we ran a protein–protein interaction analysis using String to identify the hub genes. Ranked by connectivity and betweenness centrality, MPO, S100A12, CTSG, CCR2, CAMP, S100A9, KIT, CXCR4, and CYBB were identified as the major hub genes (Figure 1D, and Additional file S1). ## 3.2. Revealing Transcriptomic Differences of Macrophages within Phenotypes We also observed that the specific macrophage phenotypes were separately clustered by induction methods, M1_IFNγ + LPS versus M1_GM-CSF and M2_IL4 + IL10 versus M2_M-CSF (Figure 2). Comparing M1_IFNγ + LPS to M1_GM-CSF, 207 and 391 genes were significantly upregulated and downregulated, respectively (Additional file S2). DEGs’ annotation indicates that they are enriched in biological processes related to immune response, such as Regulation of cell activation, Inflammatory response, Response to the bacterium, and pathways such as Staphylococcus aureus infection, Cytokine–cytokine receptor interaction, Pathways in cancer (Figure 2C, Supplementary Figure S2, and Additional file S2). To examine more closely the correlations between the genes of the core response to M1_IFNγ + LPS and M1_GM-CSF, we incorporated a network-based protein–protein interaction analysis approach by String. Ranked by connectivity and betweenness centrality, CD68, MRC1, CD28, HLA-DRA, PPARG, A2M, S100A9, SELP, RETN, and CAMP were identified as the primary hub and bottleneck genes of macrophages (Figure 2E, and Additional file S2). Comparing M2_IL4 + IL10 to M2_M-CSF, 391 and 211 genes were significantly upregulated and downregulated, respectively (Additional file S3). Running GO and KEGG annotations, the DEGs enriched in tissue development-related biological processes such as Tube morphogenesis, Reproductive structure development, Regulation of vasculature development, and pathways such as Osteoclast differentiation, Hematopoietic cell lineage, Cell adhesion molecules (Figure 2D, Supplementary Figure S3 and Additional file S3). IL6, CD4, IGF1, IL10, PTGS2, FOS, PPARG, NTRK1, ADIPOQ, FLT1, and HMOX1 were identified as the primary hub and bottleneck genes of macrophages induced by colony factors and cytokines’ exposure (Figure 2F and Additional file S3). ## 3.3. Application of M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF Signatures to the Identification of Swine Disease and Its Molecular Mechanism As macrophages play a key role in determining the activation or resolution of immune responses and can determine the fate of pathogen infection, we evaluated the robustness of our macrophage signatures (M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, M2_M-CSF) in discriminating macrophages infected with specific pathogens based on the enrichment analysis of selected genes. The association between the chosen clinical trials where infected pig macrophages were examined in 308 transcriptomes’ data obtained from the GEO (Table 2 and Additional file S4). The results indicated that genes from the SS2 infected macrophages were most significantly enriched in the M1_IFNγ + LPS set, genes from the PPRSV, ASFV, Mhp, and LPS infected were most significantly enriched in the M1_GM-CSF, genes from the PCV2 and T. gondii Me49 infections were most significantly enriched in the M2_M-CSF set, and genes from the different species of T. gondii infections were most significantly enriched in the M2_IL4 + IL10 set (Figure 3). Together, our results indicate that our macrophage signatures could characterize microarrays and RNA-seq from specific pathological scenarios. ## 4. Discussion Growing evidence shows macrophages have high plasticity and extensive polarization, which hinders the definition of macrophages obtained by different methods. The establishment of the porcine macrophage model is important for pig health and disease research. Transcriptomics is one of the primary tools in this investigation, however, we are aware of its shortcomings. Since the transcriptome is type-specific and changes over time and in response to stimulation, and some tissue-specific disorders cannot be diagnosed by RNA-seq, several studies have demonstrated that not all genes are expressed solely in particular cells [32]. Of course, considering the operability of the laboratory and the better differentiation effect of young piglets, we only used the 7-day-old DLY, which has certain limitations. In the present study, we generated four porcine macrophage phenotypes, namely M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF and created four macrophage molecular signatures by transcriptomic analysis. First, we identified 730 DEGs by comparing M1 (merging M1_IFNγ + LPS and M1_GM-CSF) with M2 (mering M2_IL4 + IL10 and M2_M-CSF). In line with previous reports, the porcine macrophages had similar gene profiles to human and mouse classical macrophage phenotypes and alternative phenotypes, respectively. For example, porcine M1 highly expressed CXCR4, S100A9, FCRL3, JAK3, and SLAMF1, while M2 highly expressed CCL11, POSTN, PTX3 MFAP4, PTH1R, and AGTR1. Apart from the well-known macrophage markers, we noticed that porcine M1 highly expressed MPO, S100A12, CTSG, CCR2, CAMP, KIT, and CYBB, and M2 highly expressed ANK2, ALDH1A1, PTHLH, and CACNA1H. MPO is a marker of macrophage aggregation in inflammatory species and promotes the recognition of macrophage scavenger receptors [33]. S100A12, CSTG, CAMP, and KIT are antimicrobial response genes. CSTG, CAMP, and CYBB are involved in macrophage phagocytosis [34,35,36]. Both CCR2 and CXCR4 are pro-inflammatory genes involved in macrophage migration [37]. ANK2 promotes the growth and invasion of pancreatic carcinoma, and the peptide derived from ANK2 is an effective and specific autophagy inhibitor binding to ATG8 [38,39]. Studies have shown that the expression level of ALDH1A1 is reduced in the inflammatory state, which is part of early inflammation [40]. PTHLH is involved in cell and organ growth, development, migration, and survival, and can be used as an independent marker of prognosis [41]. These genes may serve as specific porcine macrophage markers and potential therapeutic targets. It is worth mentioning that four MHC haplotype genes were detected expressing in porcine bone marrow-derived macrophages, such as CIITA, HLA-DOB, HLA-DRA, SLA-DMB, and two of them, HLA-DRA and SLA-DMB, are differentially expressed, comparing M1 versus M2 [42]. After revealing the transcriptomic difference between the two main macrophage phenotypes, we conducted a transcriptomic comparison of M1_IFNγ + LPS with M1_GM-CSF, and M2_IL4 + IL10 with M2_M-CSF. We found that porcine M1_IFNγ + LPS highly expressed S100A9, SELP, RETN, CAMP, and M1_GM-CSF highly expressed CD68, MRC1, CD28, HLA-DRA, PPARG, and A2M. S100A9 is related to the CD68 regulating macrophage function pathway and promoting macrophage migration, which can induce neutrophil chemotaxis and promote macrophage migration and adhesion under inflammatory conditions [43]. RETN has been reported to induce the production of pro-inflammatory cytokines and chemokines in PBMC [44]. CAMP can inhibit the phagocytosis of macrophages through the CAMP-1-activated CAMP-effect-exchange protein [45]. CD68 is highly expressed in macrophages and belongs to the scavenger family. It has the functions of clearing cell debris, promoting phagocytosis, and mediating the recruitment and activation of macrophages. MRC1 and A2M mediate the phagocytosis of macrophages [43]. CD28 can enhance the expression of RANTES and MIP-1α in T cells and MIP-1β, and increase the number and differentiation of macrophages in the wound healing stage [9,46]. HLA-DRA has been proven to inhibit the phagocytosis of macrophages in order to protect the intracellular niche from phagocytosis and killing of host macrophages, which is positively related to the regulation of GM-CSF [47,48,49]. We found that porcine M2_IL4 + IL10 highly expressed PPARG, NTRK1, ADIPOQ, FLT1, HMOX1, and M2_M-CSF highly expressed IL6, CD4, IGF1, IL10, PTGS2, FOS, PPARG [50,51,52]. PPARG has an anti-inflammatory effect, and its ligand is responsible for clearing the expression of genes of apoptotic cells and macrophage-mediated inflammatory responses [53,54]. Interestingly, PPARG was also specifically expressed in M1_GM-CSF. NTRK1 decreases malignancy and/or spontaneous degeneration of neuroblastoma cells [55]. ADIPOQ and FLT1 have anti-inflammatory functions [56,57,58]. According to reports, HMOX1 also has anti-cancer, anti-inflammatory, anti-apoptotic, anti-proliferation, and anti-oxidation effects [59]. Although IL6 is usually associated with pro-inflammatory function and involves many inflammatory diseases, it can increase the polarization of alternatively activated (AAM) [60]. AAM and IL6 together, lead to the release of IL10 [60,61]. At the same time, FOS can increase the expression level of IL10 and promote the formation of osteoclasts [62]. IGF1 and IL10 are also positively correlated with the inhibition of inflammation and wound healing. Interestingly, IL10 also promotes macrophage phagocyte debris [63]. CD4 has an anti-inflammatory effect, inhibits macrophage migration, and induces M2 polarization [64]. Interestingly, PTGS2 is a pro-inflammatory gene [65]. We have investigated the variations among various phenotypes and associated them with disease outcomes because we are interested in investigating the potential application of macrophage phenotypes. We proposed a consensus collection of markers describing major macrophages’ activation phenotypes, namely M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF, able to characterize robustly controlled in vitro and in vivo scenarios for macrophage induction. Our study confirmed that macrophages cultured in vitro have different enrichment pathways from macrophages infected with different pathogens. At the same time, it is confirmed that macrophages infected with the same pathogen in vivo or in vitro also have different enrichment pathways. Since the description of the polarization state of macrophages has not been unified, our study provides a framework to guide the definition of the phenotype of porcine macrophages in the disease state. Future research into the macrophage model in a disease setting will help us create medications and diagnostic tools for particular disorders. Nevertheless, we can never ignore the bias using public downloaded transcriptomic data. We conclude here that there are transcriptomic differences between and within two macrophage phenotypes in our porcine model. In general, we found the four types of macrophages (M1_IFNγ + LPS, M1_GM-CSF, M2_IL4 + IL10, and M2_M-CSF) have different functions. Compared with M1_GM-CSF, M1_IFNγ + LPS has a weaker phagocytic capacity, but stronger antibacterial and migratory capacity; M2_IL4 + IL10 has a stronger tissue repair function, while M2_M-CSF has a stronger wound healing ability [66]. At the same time, we used the four established gene characteristics to identify various pig infectious diseases with prognosis and predictive values. Pigs’ immunological parameters overlap with either mice or humans’ in particular ways more than mice and humans’ do with each other [21,23,24,25,26,27]. Indeed, it would be an interesting research direction to establish similar models of pig, mouse, and human macrophages for future studies. ## References 1. Takahashi K.. Development and Differentiation of Macrophages and Related Cells: Historical Review and Current Concepts. J. Clin. Exp. Hematop. (2001) 41 1-31. DOI: 10.3960/jslrt.41.1 2. Murray P.J., Wynn T.A.. Protective and Pathogenic Functions of Macrophage Subsets. Nat. Rev. Immunol. (2011) 11 723-737. DOI: 10.1038/nri3073 3. Sica A., Mantovani A.. Plasticity and Polarization. J. Clin. Investig. (2012) 122 787-795. DOI: 10.1172/JCI59643 4. Ma W., Gao F., Gu K., Chen D.. The Role of Monocytes and Macrophages in Autoimmune Diseases: A Comprehensive Review. Front. Immunol. (2019) 10 1140. DOI: 10.3389/fimmu.2019.01140 5. Rendra E., Riabov V., Mossel D.M., Sevastyanova T., Harmsen M.C., Kzhyshkowska J.. Reactive Oxygen Species (ROS) in Macrophage Activation and Function in Diabetes. Immunobiology (2019) 224 242-253. DOI: 10.1016/j.imbio.2018.11.010 6. Martinez F.O., Gordon S.. The M1 and M2 Paradigm of Macrophage Activation: Time for Reassessment. F1000Prime Rep. (2014) 6 13. DOI: 10.12703/P6-13 7. Bility M.T., Cheng L., Zhang Z., Luan Y., Li F., Chi L., Zhang L., Tu Z., Gao Y., Fu Y.-X.. Hepatitis B Virus Infection and Immunopathogenesis in a Humanized Mouse Model: Induction of Human-Specific Liver Fibrosis and M2-Like Macrophages. PLoS Pathog. (2014) 10. DOI: 10.1371/journal.ppat.1004032 8. Anderson N.R., Minutolo N.G., Gill S., Klichinsky M.. Macrophage-Based Approaches for Cancer Immunotherapy. Cancer Res. (2021) 81 1201-1208. DOI: 10.1158/0008-5472.CAN-20-2990 9. Lacey D.C., Achuthan A., Fleetwood A.J., Dinh H., Roiniotis J., Scholz G.M., Chang M.W., Beckman S.K., Cook A.D., Hamilton J.A.. Defining GM-CSF– and Macrophage-CSF–Dependent Macrophage Responses by In Vitro Models. J. Immunol. (2012) 188 5752-5765. DOI: 10.4049/jimmunol.1103426 10. Martinez F.O., Gordon S., Locati M., Mantovani A.. Transcriptional Profiling of the Human Monocyte-to-Macrophage Differentiation and Polarization: New Molecules and Patterns of Gene Expression. J. Immunol. (2006) 177 7303-7311. DOI: 10.4049/jimmunol.177.10.7303 11. Vogel D.Y.S., Vereyken E.J.F., Glim J.E., Heijnen P.D.A.M., Moeton M., van der Valk P., Amor S., Teunissen C.E., van Horssen J., Dijkstra C.D.. Macrophages in Inflammatory Multiple Sclerosis Lesions Have an Intermediate Activation Status. J. Neuroinflamm. (2013) 10 809. DOI: 10.1186/1742-2094-10-35 12. Swindle M.M., Makin A., Herron A.J., Clubb F.J., Frazier K.S.. Swine as Models in Biomedical Research and Toxicology Testing. Vet. Pathol. (2012) 49 344-356. DOI: 10.1177/0300985811402846 13. Sironi M.. Differential Regulation of Chemokine Production by Fc Receptor Engagement in Human Monocytes: Association of CCL1 with a Distinct Form of M2 Monocyte Activation (M2b, Type 2). J. Leukoc. Biol. (2006) 80 342-349. DOI: 10.1189/jlb.1005586 14. Mantovani A., Sozzani S., Locati M., Allavena P., Sica A.. Macrophage Polarization: Tumor-Associated Macrophages as a Paradigm for Polarized M2 Mononuclear Phagocytes. Trends Immunol. (2002) 23 549-555. DOI: 10.1016/S1471-4906(02)02302-5 15. Morris S.M.. Arginine Metabolism: Boundaries of Our Knowledge. J. Nutr. (2007) 137 1602S-1609S. DOI: 10.1093/jn/137.6.1602S 16. Perleberg C., Kind A., Schnieke A.. Genetically Engineered Pigs as Models for Human Disease. DMM Dis. Model. Mech. (2018) 11 dmm030783. DOI: 10.1242/dmm.030783 17. Pepe G., Calderazzi G., De Maglie M., Villa A.M., Vegeto E.. Heterogeneous Induction of Microglia M2a Phenotype by Central Administration of Interleukin-4. J. Neuroinflamm. (2014) 11 211. DOI: 10.1186/s12974-014-0211-6 18. Giraud E., Lestinova T., Derrick T., Martin O., Dillon R.J., Volf P., Műller I., Bates P.A., Rogers M.E.. Leishmania Proteophosphoglycans Regurgitated from Infected Sand Flies Accelerate Dermal Wound Repair and Exacerbate Leishmaniasis via Insulin-like Growth Factor 1-Dependent Signalling. PLoS Pathog. (2018) 14. DOI: 10.1371/journal.ppat.1006794 19. Leidi M., Gotti E., Bologna L., Miranda E., Rimoldi M., Sica A., Roncalli M., Palumbo G.A., Introna M., Golay J.. M2 Macrophages Phagocytose Rituximab-Opsonized Leukemic Targets More Efficiently than M1 Cells In Vitro. J. Immunol. (2009) 182 4415-4422. DOI: 10.4049/jimmunol.0713732 20. Harusato A., Geem D., Denning T.L.. Macrophage Isolation from the Mouse Small and Large Intestine. Methods Mol. Biol. (2016) 1422 171-180. DOI: 10.1007/978-1-4939-3603-8_16 21. Truty M.J., Smoot R.L.. Animal Models in Pancreatic Surgery: A Plea for Pork. Pancreatology (2008) 8 546-550. DOI: 10.1159/000160214 22. Martinez F.O., Helming L., Milde R., Varin A., Melgert B.N., Draijer C., Thomas B., Fabbri M., Crawshaw A., Ho L.P.. Genetic Programs Expressed in Resting and IL-4 Alternatively Activated Mouse and Human Macrophages: Similarities and Differences. Blood J. Am. Soc. Hematol. (2013) 121 e57-e69. DOI: 10.1182/blood-2012-06-436212 23. Gao J., Scheenstra M.R., van Dijk A., Veldhuizen E.J.A., Haagsman H.P.. A New and Efficient Culture Method for Porcine Bone Marrow-Derived M1-and M2-Polarized Macrophages. Vet. Immunol. Immunopathol. (2018) 200 7-15. DOI: 10.1016/j.vetimm.2018.04.002 24. Raes G., Van den Bergh R., De Baetselier P., Ghassabeh G.H.. Arginase-1 and Ym1 Are Markers for Murine, but Not Human, Alternatively Activated Myeloid Cells. J. Immunol. (2005) 174 6561-6562. DOI: 10.4049/jimmunol.174.11.6561 25. Schneemann M., Schoeden G.. Macrophage Biology and Immunology: Man Is Not a Mouse. J. Leukoc. Biol. (2007) 81 579. DOI: 10.1189/jlb.1106702 26. Meurens F., Summerfield A., Nauwynck H., Saif L., Gerdts V.. The Pig: A Model for Human Infectious Diseases. Trends Microbiol. (2012) 20 50-57. DOI: 10.1016/j.tim.2011.11.002 27. Mehle A., Doudna J.A.. Adaptive Strategies of the Influenza Virus Polymerase for Replication in Humans. Proc. Natl. Acad. Sci. USA (2009) 106 21312-21316. DOI: 10.1073/pnas.0911915106 28. Mclaughlin T., Ackerman S.E., Shen L., Engleman E.. Role of Innate and Adaptive Immunity in Obesity-Associated Metabolic Disease. J. Clin. Investig. (2017) 127 5-13. DOI: 10.1172/JCI88876 29. Björck P.. Isolation and Characterization of Plasmacytoid Dendritic Cells from Flt3 Ligand and Granulocyte-Macrophage Colony-Stimulating Factor-Treated Mice. Blood (2001) 98 3520-3526. DOI: 10.1182/blood.V98.13.3520 30. Bray N.L., Pimentel H., Melsted P., Pachter L.. Near-Optimal Probabilistic RNA-Seq Quantification. Nat. Biotechnol. (2016) 34 525-527. DOI: 10.1038/nbt.3519 31. Subramanian A., Tamayo P., Mootha V.K., Mukherjee S., Ebert B.L., Gillette M.A., Paulovich A., Pomeroy S.L., Golub T.R., Lander E.S.. Gene Set Enrichment Analysis: A Knowledge-Based Approach for Interpreting Genome-Wide Expression Profiles. Proc. Natl. Acad. Sci. USA (2005) 102 15545-15550. DOI: 10.1073/pnas.0506580102 32. Cirulli E.T., Singh A., Shianna K.V., Ge D., Smith J.P., Maia J.M., Heinzen E.L., Goedert J.J., Goldstein D.B.. Screening the Human Exome: A Comparison of Whole Genome and Whole Transcriptome Sequencing. Genome Biol. (2010) 11 R57. DOI: 10.1186/gb-2010-11-5-r57 33. Luo H., Ma C.. A Novel Ferroptosis-Associated Gene Signature to Predict Prognosis in Patients with Uveal Melanoma. Diagnostics (2021) 11. DOI: 10.3390/diagnostics11020219 34. Li X., Yu W., Wollenweber T., Lu X., Wei Y., Beitzke D., Wadsak W., Kropf S., Wester H.J., Haug A.R.. [68 Ga] Pentixafor PET/MR Imaging of Chemokine Receptor 4 Expression in the Human Carotid Artery. Eur. J. Nucl. Med. Mol. Imaging (2019) 46 1616-1625. DOI: 10.1007/s00259-019-04322-7 35. Tang C., Lei X., Xiong L., Hu Z., Tang B.. HMGA1B/2 Transcriptionally Activated-POU1F1 Facilitates Gastric Carcinoma Metastasis via CXCL12/CXCR4 Axis-Mediated Macrophage Polarization. Cell Death Dis. (2021) 12 1-15. DOI: 10.1038/s41419-021-03703-x 36. Emokpae M.A., Mrakpor B.A.. Do Sex Differences in Respiratory Burst Enzyme Activities Exist in Human Immunodeficiency Virus-1 Infection?. Med. Sci. (2016) 4. DOI: 10.3390/medsci4040019 37. Wu W., Hou B., Tang C., Liu F., Yang J., Pan T., Si K., Lu D., Wang X., Wang J.. (+)-Usnic Acid Inhibits Migration of c-KIT Positive Cells in Human Colorectal Cancer. Evid.-Based Complement. Altern. Med. (2018) 2018 5149436. DOI: 10.1155/2018/5149436 38. Van Rossum D., Hilbert S., Straßenburg S., Hanisch U., Brück W.. Myelin-phagocytosing Macrophages in Isolated Sciatic and Optic Nerves Reveal a Unique Reactive Phenotype. Glia (2008) 56 271-283. DOI: 10.1002/glia.20611 39. Cao W., Wei W., Zhan Z., Xie D., Xie Y., Xiao Q.. Regulation of Drug Resistance and Metastasis of Gastric Cancer Cells via the MicroRNA647-ANK2 Axis. Int. J. Mol. Med. (2018) 41 1958-1966. DOI: 10.3892/ijmm.2018.3381 40. Auburger G., Gispert S., Torres-Odio S., Jendrach M., Brehm N., Canet-Pons J., Key J., Sen N.-E.. SerThr-PhosphoProteome of Brain from Aged PINK1-KO+ A53T-SNCA Mice Reveals PT1928-MAP1B and PS3781-ANK2 Deficits, as Hub between Autophagy and Synapse Changes. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20133284 41. Wu X., Lv D., Cai C., Zhao Z., Wang M., Chen W., Liu Y.. A TP53-Associated Immune Prognostic Signature for the Prediction of Overall Survival and Therapeutic Responses in Muscle-Invasive Bladder Cancer. Front. Immunol. (2020) 11 1-15. DOI: 10.3389/fimmu.2020.590618 42. Foss D.L., Bennaars A.M., Pennell C.A., Moody M.D., Murtaugh M.P.. Differentiation of Porcine Dendritic Cells by Granulocyte-Macrophage Colony-Stimulating Factor Expressed in Pichia Pastoris. Vet. Immunol. Immunopathol. (2003) 91 205-215. DOI: 10.1016/S0165-2427(03)00002-3 43. Bézie S., Freuchet A., Sérazin C., Salama A., Vimond N., Anegon I., Guillonneau C.. IL-34 Actions on FOXP3+ Tregs and CD14+ Monocytes Control Human Graft Rejection. Front. Immunol. (2020) 11 1496. DOI: 10.3389/fimmu.2020.01496 44. Mossel D.M., Moganti K., Riabov V., Weiss C., Kopf S., Cordero J., Dobreva G., Rots M.G., Klüter H., Harmsen M.C.. Epigenetic Regulation of S100A9 and S100A12 Expression in Monocyte-Macrophage System in Hyperglycemic Conditions. Front. Immunol. (2020) 11 1071. DOI: 10.3389/fimmu.2020.01071 45. Pan L., Wei N., Jia H., Gao M., Chen X., Wei R., Sun Q., Gu S., Du B., Xing A.. Genome-Wide Transcriptional Profiling Identifies Potential Signatures in Discriminating Active Tuberculosis from Latent Infection. Oncotarget (2017) 8 112907. DOI: 10.18632/oncotarget.22889 46. Carroll R.G., Riley J.L., Levine B.L., Feng Y., Kaushal S., Ritchey D.W., Bernstein W., Weislow O.S., Brown C.R., Berger E.A.. Differential Regulation of HIV-1 Fusion Cofactor Expression by CD28 Costimulation of CD4+ T Cells. Science (1997) 276 273-276. DOI: 10.1126/science.276.5310.273 47. Gladow N., Hollmann C., Ramos G., Frantz S., Kerkau T., Beyersdorf N., Hofmann U.. Treatment of Mice with a Ligand Binding Blocking Anti-CD28 Monoclonal Antibody Improves Healing after Myocardial Infarction. PLoS ONE (2020) 15. PMID: 32298302 48. Vozza E.G., Mulcahy M.E., McLoughlin R.M.. Making the Most of the Host; Targeting the Autophagy Pathway Facilitates Staphylococcus Aureus Intracellular Survival in Neutrophils. Front. Immunol. (2021) 12 2304. DOI: 10.3389/fimmu.2021.667387 49. Perry S.E., Mostafa S.M., Wenstone R., Shenkin A., McLaughlin P.J.. HLA-DR Regulation and the Influence of GM-CSF on Transcription, Surface Expression and Shedding. Int. J. Med. Sci. (2004) 1 126. DOI: 10.7150/ijms.1.126 50. Peng P., Zhu H., Liu D., Chen Z., Zhang X., Guo Z., Dong M., Wan L., Zhang P., Liu G.. TGFBI Secreted by Tumor-Associated Macrophages Promotes Glioblastoma Stem Cell-Driven Tumor Growth via Integrin Avβ5-Src-Stat3 Signaling. Theranostics (2022) 12 4221. DOI: 10.7150/thno.69605 51. Pasqualetti F., Giampietro C., Montemurro N., Giannini N., Gadducci G., Orlandi P., Natali E., Chiarugi P., Gonnelli A., Cantarella M.. Old and New Systemic Immune-Inflammation Indexes Are Associated with Overall Survival of Glioblastoma Patients Treated with Radio-Chemotherapy. Genes (2022) 13. DOI: 10.3390/genes13061054 52. Li D., Zhang Q., Li L., Chen K., Yang J., Dixit D., Gimple R.C., Ci S., Lu C., Hu L.. Β2-Microglobulin Maintains Glioblastoma Stem Cells and Induces M2-like Polarization of Tumor-Associated Macrophages. Cancer Res. (2022) 82 3321-3334. DOI: 10.1158/0008-5472.CAN-22-0507 53. Vikhe P.P., Purnell T., Brown S.D.M., Hood D.W.. Cellular Immune Response against Nontypeable Haemophilus Influenzae Infecting the Preinflamed Middle Ear of the Junbo Mouse. Infect. Immun. (2019) 87 e00689-19. DOI: 10.1128/IAI.00689-19 54. Glas J., Seiderer J., Markus C., Pfennig S., Wetzke M., Paschos E., Göke B., Ochsenkühn T., Müller-Myhsok B., Diegelmann J.. Role of PPARG Gene Variants in Inflammatory Bowel Disease. Inflamm. Bowel Dis. (2011) 17 1057-1058. DOI: 10.1002/ibd.21425 55. Zahuczky G., Kristóf E., Majai G., Fésüs L.. Differentiation and Glucocorticoid Regulated Apopto-Phagocytic Gene Expression Patterns in Human Macrophages. Role of Mertk in Enhanced Phagocytosis. PLoS ONE (2011) 6. DOI: 10.1371/journal.pone.0021349 56. Pajtler K.W., Rebmann V., Lindemann M., Schulte J.H., Schulte S., Stauder M., Leuschner I., Schmid K., Köhl U., Schramm A.. Expression of NTRK1/TrkA Affects Immunogenicity of Neuroblastoma Cells. Int. J. Cancer (2013) 133 908-919. DOI: 10.1002/ijc.28096 57. Pereira G., Bexiga R., Silva J.C.E., Silva E., Ramé C., Dupont J., Guo Y., Humblot P., Lopes-da-Costa L.. Adipokines as Biomarkers of Postpartum Subclinical Endometritis in Dairy Cows. Reproduction (2020) 160 417-430. DOI: 10.1530/REP-20-0183 58. Otowa Y., Moriwaki K., Sano K., Shirakabe M., Yonemura S., Shibuya M., Rossant J., Suda T., Kakeji Y., Hirashima M.. Flt1/VEGFR1 Heterozygosity Causes Transient Embryonic Edema. Sci. Rep. (2016) 6 27186. DOI: 10.1038/srep27186 59. Lokki A.I., Heikkinen-Eloranta J.K., Laivuori H.. The Immunogenetic Conundrum of Preeclampsia. Front. Immunol. (2018) 9 2630. DOI: 10.3389/fimmu.2018.02630 60. Fernando M.R., Giembycz M.A., McKay D.M.. Bidirectional Crosstalk via IL-6, PGE2 and PGD2 between Murine Myofibroblasts and Alternatively Activated Macrophages Enhances Anti-Inflammatory Phenotype in Both Cells. Br. J. Pharmacol. (2016) 173 899-912. DOI: 10.1111/bph.13409 61. Keränen T., Hömmö T., Hämäläinen M., Moilanen E., Korhonen R.. Anti-Inflammatory Effects of Β2-Receptor Agonists Salbutamol and Terbutaline Are Mediated by MKP-1. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0148144 62. Barroso F.A.L., de Jesus L.C.L., de Castro C.P., Batista V.L., Ferreira Ê., Fernandes R.S., de Barros A.L.B., Leclerq S.Y., Azevedo V., Mancha-Agresti P.. Intake of Lactobacillus Delbrueckii (PExu: Hsp65) Prevents the Inflammation and the Disorganization of the Intestinal Mucosa in a Mouse Model of Mucositis. Microorganisms (2021) 9. DOI: 10.3390/microorganisms9010107 63. Bergmann C.B., Beckmann N., Salyer C.E., Hanschen M., Crisologo P.A., Caldwell C.C.. Potential Targets to Mitigate Trauma-or Sepsis-Induced Immune Suppression. Front. Immunol. (2021) 12 121. DOI: 10.3389/fimmu.2021.622601 64. Fernando M.R., Reyes J.L., Iannuzzi J., Leung G., McKay D.M.. The Pro-Inflammatory Cytokine, Interleukin-6, Enhances the Polarization of Alternatively Activated Macrophages. PLoS ONE (2014) 9. DOI: 10.1371/journal.pone.0094188 65. Messing M., Jan-Abu S.C., McNagny K.. Group 2 Innate Lymphoid Cells: Central Players in a Recurring Theme of Repair and Regeneration. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21041350 66. WERNER S., GROSE R.. Regulation of Wound Healing by Growth Factors and Cytokines. Physiol. Rev. (2003) 83 835-870. DOI: 10.1152/physrev.2003.83.3.835 67. Acosta J.R., Tavira B., Douagi I., Kulyté A., Arner P., Rydén M., Laurencikiene J.. Human-Specific Function of IL-10 in Adipose Tissue Linked to Insulin Resistance. J. Clin. Endocrinol. Metab. (2019) 104 4552-4562. DOI: 10.1210/jc.2019-00341 68. Al-Rubaie A., Wise A.F., Sozo F., De Matteo R., Samuel C.S., Harding R., Ricardo S.D.. The therapeutic effect of mesenchymal stem cells on pulmonary myeloid cells following neonatal hyperoxic lung injury in mice. Respir. Res. (2018) 19 1-11. DOI: 10.1186/s12931-018-0816-x 69. Awad F., Assrawi E., Jumeau C., Georgin-Lavialle S., Cobret L., Duquesnoy P., Piterboth W., Thomas L., Stankovic-Stojanovic K., Louvrier C.. Impact of human monocyte and macrophage polarization on NLR expression and NLRP3 inflammasome activation. PLoS ONE (2017) 12. DOI: 10.1371/journal.pone.0175336 70. Beceiro S., Radin J., Chatuvedi R., Piazuelo M., Horvarth D., Cortado H., Gu Y., Dixon B., Gu C., Lange I.. TRPM2 ion channels regulate macrophage polarization and gastric inflammation during Helicobacter pylori infection. Mucosal Immunol. (2017) 10 493-507. DOI: 10.1038/mi.2016.60 71. Binnemars-Postma K., Bansal R., Storm G., Prakash J.. Targeting the Stat6 pathway in tumor-associated macrophages reduces tumor growth and metastatic niche formation in breast cancer. FASEB J. (2018) 32 969-978. DOI: 10.1096/fj.201700629R 72. Camiolo G., Barbato A., Giallongo C., Vicario N., Romano A., Parrinello N.L., Parenti R., Sandoval J.C., García-Moreno D., Lazzarino G.. Iron regulates myeloma cell/macrophage interaction and drives resistance to bortezomib. Redox Biol. (2020) 36 101611. DOI: 10.1016/j.redox.2020.101611 73. Chen C., Perry T.L., Chitko-McKown C.G., Smith A.D., Cheung L., Beshah E., Urban J.F., Dawson H.D.. The regulatory actions of retinoic acid on M2 polarization of porcine macrophages. Dev. Comp. Immunol. (2019) 98 20-33. DOI: 10.1016/j.dci.2019.03.020 74. Chen P.-C., Cheng H.-C., Wang J., Wang S.-W., Tai H.-C., Lin C.-W., Tang C.-H.. Prostate cancer-derived CCN3 induces M2 macrophage infiltration and contributes to angiogenesis in prostate cancer microenvironment. Oncotarget (2014) 5 1595-1608. DOI: 10.18632/oncotarget.1570 75. Cheng Y., Si Y., Wang L., Ding M., Yu S., Lu L., Guo Y., Zong M., Fan L.. The regulation of macrophage polarization by hypoxia-PADI4 coordination in Rheumatoid arthritis. Int. Immunopharmacol. (2021) 99 107988. DOI: 10.1016/j.intimp.2021.107988 76. Chiba Y., Mizoguchi I., Furusawa J., Hasegawa H., Ohashi M., Xu M., Owaki T., Yoshimoto T.. Interleukin-27 exerts its antitumor effects by promoting differentiation of hematopoietic stem cells to M1 macrophages. Cancer Res. (2018) 78 182-194. DOI: 10.1158/0008-5472.CAN-17-0960 77. Cui Y., Chen F., Gao J., Lei M., Wang D., Jin X., Guo Y., Shan L., Chen X.. Comprehensive landscape of the renin-angiotensin system in Pan-cancer: A potential downstream mediated mechanism of SARS-CoV-2. Int. J. Biol. Sci. (2021) 17 3795. DOI: 10.7150/ijbs.53312 78. Cui Z., Feng Y., Li D., Li T., Gao P., Xu T.. Activation of aryl hydrocarbon receptor (AhR) in mesenchymal stem cells modulates macrophage polarization in asthma. J. Immunotoxicol. (2020) 17 21-30. DOI: 10.1080/1547691X.2019.1706671 79. Dong H., Xie C., Jiang Y., Li K., Lin Y., Pang X., Xiong X., Zheng J., Ke X., Chen Y.. Tumor-Derived Exosomal Protein Tyrosine Phosphatase Receptor Type O Polarizes Macrophage to Suppress Breast Tumor Cell Invasion and Migration. Front. Cell Dev. Biol. (2021) 2731. DOI: 10.3389/fcell.2021.703537 80. Fang X.Y., Zhan Y.X., Zhou X.M., Wu L.N., Lin J., Yi Y.T., Jiang C.M., Wang J., Liu J.. CXCL12/CXCR4 Mediates Orthodontic Root Resorption via Regulating the M1/M2 Ratio. J. Dent. Res. (2021) 101 569-579. DOI: 10.1177/00220345211050324 81. Guo Y., Lin C., Xu P., Wu S., Fu X., Xia W., Yao M.. AGEs induced autophagy impairs cutaneous wound healing via stimulating macrophage polarization to M1 in diabetes. Sci. Rep. (2016) 6 1-16. DOI: 10.1038/srep36416 82. Huang Y., Li Q., Hu R., Li R., Yang Y.. Five immune-related genes as diagnostic markers for endometriosis and their correlation with immune infiltration. Front. Endocrinol. (2022) 13 1-15. DOI: 10.3389/fendo.2022.1011742 83. Inoue T.. M1 macrophage triggered by Mincle leads to a deterioration of acute kidney injury. Kidney Int. (2017) 91 526-529. DOI: 10.1016/j.kint.2016.11.026 84. Jablonski K.A., Gaudet A.D., Amici S.A., Popovich P.G., Guerau-De-Arellano M.. Control of the inflammatory macrophage transcriptional signature by miR-155. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0159724 85. Joerink M., Rindsjö E., van Riel B., Alm J., Papadogiannakis N.. Placental macrophage (Hofbauer cell) polarization is independent of maternal allergen-sensitization and presence of chorioamnionitis. Placenta (2011) 32 380-385. DOI: 10.1016/j.placenta.2011.02.003 86. Ganta V.C., Choi M., Farber C.R., Annex B.H.. Antiangiogenic VEGF165b regulates macrophage polarization via S100A8/S100A9 in peripheral artery disease. Circulation (2019) 139 226-242. DOI: 10.1161/CIRCULATIONAHA.118.034165 87. Le Y., Gao H., Bleday R., Zhu Z.. The homeobox protein VentX reverts immune suppression in the tumor microenvironment. Nat. Commun. (2018) 9 1-10. DOI: 10.1038/s41467-018-04567-0 88. Liu M., Li F., Bin Liu B., Jian Y., Zhang D., Zhou H., Wang Y., Xu Z.. Profiles of immune cell infiltration and immune-related genes in the tumor microenvironment of esophageal squamous cell carcinoma. BMC Med. Genomics (2021) 14. DOI: 10.1186/s12920-021-00928-9 89. Lv D., Wu X., Chen X., Yang S., Chen W., Wang M., Liu Y., Gu D., Zeng G.. A novel immune-related gene-based prognostic signature to predict biochemical recurrence in patients with prostate cancer after radical prostatectomy. Cancer Immunol. Immunother. (2021) 70 3587-3602. DOI: 10.1007/s00262-021-02923-6 90. Müller E., Christopoulos P.F., Halder S., Lunde A., Beraki K., Speth M., Øynebråten I., Corthay A.. Toll-like receptor ligands and interferon-γ synergize for induction of antitumor M1 macrophages. Front. Immunol. (2017) 8 1383. DOI: 10.3389/fimmu.2017.01383 91. Oliveira L., McClellan S., Hansen P.J.. Differentiation of the endometrial macrophage during pregnancy in the cow. PLoS ONE (2010) 5. DOI: 10.1371/journal.pone.0013213 92. Pagie S., Gérard N., Charreau B.. Notch signaling triggered via the ligand DLL4 impedes M2 macrophage differentiation and promotes their apoptosis. Cell Commun. Signal. (2018) 16 1-12. DOI: 10.1186/s12964-017-0214-x 93. Quan Q., Xiong X., Wu S., Yu M.. Identification of Immune-Related Key Genes in Ovarian Cancer Based on WGCNA. Front. Genet. (2021) 12 760225. DOI: 10.3389/fgene.2021.760225 94. Quero L., Tiaden A.N., Hanser E., Roux J., Laski A., Hall J., Kyburz D.. miR-221-3p Drives the Shift of M2-Macrophages to a Pro-Inflammatory Function by Suppressing JAK3/STAT3 Activation. Front. Immunol. (2020) 10 3087. DOI: 10.3389/fimmu.2019.03087 95. Sanjurjo L., Aran G., Téllez É., Amézaga N., Armengol C., López D., Prats C., Sarrias M.-R.. CD5L promotes M2 macrophage polarization through autophagy-mediated upregulation of ID3. Front. Immunol. (2018) 9 480. DOI: 10.3389/fimmu.2018.00480 96. Sato K., Yamashita T., Shirai R., Shibata K., Okano T., Yamaguchi M., Mori Y., Hirano T., Watanabe T.. Adropin contributes to anti-atherosclerosis by suppressing monocyte-endothelial cell adhesion and smooth muscle cell proliferation. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19051293 97. Seneviratne A., Cave L., Hyde G., Moestrup S.K., Carling D., Mason J., O Haskard D., Boyle J.J.. Metformin directly suppresses atherosclerosis in normoglycaemic mice via haematopoietic adenosine monophosphate-activated protein kinase. Cardiovasc. Res. (2021) 117 1295-1308. DOI: 10.1093/cvr/cvaa171 98. Shapouri-Moghaddam A., Mohammadian S., Vazini H., Taghadosi M., Esmaeili S.-A., Mardani F., Seifi B., Mohammadi A., Afshari J.T., Sahebkar A.. Macrophage plasticity, polarization, and function in health and disease. J. Cell. Physiol. (2018) 233 6425-6440. DOI: 10.1002/jcp.26429 99. Song W.-M., Agrawal P., Von Itter R., Fontanals-Cirera B., Wang M., Zhou X., Mahal L.K., Hernando E., Zhang B.. Network models of primary melanoma microenvironments identify key melanoma regulators underlying prognosis. Nat. Commun. (2021) 12 1-14. DOI: 10.1038/s41467-021-21457-0 100. Talamonti E., Pauter A.M., Asadi A., Fischer A.W., Chiurchiù V., Jacobsson A.. Impairment of systemic DHA synthesis affects macrophage plasticity and polarization: Implications for DHA supplementation during inflammation. Cell. Mol. Life Sci. (2017) 74 2815-2826. DOI: 10.1007/s00018-017-2498-9 101. Tan Y., Sun R., Liu L., Yang D., Xiang Q., Li L., Tang J., Qiu Z., Peng W., Wang Y.. Tumor suppressor DRD2 facilitates M1 macrophages and restricts NF-κB signaling to trigger pyroptosis in breast cancer. Theranostics (2021) 11 5214. DOI: 10.7150/thno.58322 102. Bossche J.V.D., Laoui D., Morias Y., Movahedi K., Raes G., De Baetselier P., Van Ginderachter J.A.. Claudin-1, Claudin-2 and Claudin-11 Genes Differentially Associate with Distinct Types of Anti-inflammatory Macrophages In vitro and with Parasite- and Tumour-elicited Macrophages In vivo. Scand. J. Immunol. (2012) 75 588-598. DOI: 10.1111/j.1365-3083.2012.02689.x 103. Wang L., Liu X., Song X., Dong L., Liu D.. MiR-202-5p promotes M2 polarization in allergic rhinitis by targeting MATN2. Int. Arch. Allergy Immunol. (2019) 178 119-127. DOI: 10.1159/000493803 104. Wu D., Weng Y., Feng Y., Liang B., Wang H., Li L., Wang Z.. Trem1 Induces Periodontal Inflammation via Regulating M1 Polarization. J. Dent. Res. (2021) 101 437-447. DOI: 10.1177/00220345211044681 105. Wu M.F., Lin C.A., Yuan T.H., Yeh H.Y., Su S.F., Guo C.L., Chang G.C., Li K.C., Ho C.C., Chen H.W.. The M1/M2 spectrum and plasticity of malignant pleural effusion-macrophage in advanced lung cancer. Cancer Immunol. Immunother. (2021) 70 1435-1450. DOI: 10.1007/s00262-020-02781-8 106. Wu X., Giobbie-Hurder A., Liao X., Connelly C., Connolly E.M., Li J., Manos M.P., Lawrence D., McDermott D., Severgnini M.. Angiopoietin-2 as a Biomarker and Target for Immune Checkpoint Therapy. Cancer Immunol. Res. (2017) 5 17-28. DOI: 10.1158/2326-6066.CIR-16-0206 107. Yan C., Zhou Q.-Y., Wu J., Xu N., Du Y., Li J., Liu J.-X., Koda S., Zhang B.-B., Yu Q.. Csi-let-7a-5p delivered by extracellular vesicles from a liver fluke activates M1-like macrophages and exacerbates biliary injuries. Proc. Natl. Acad. Sci. USA (2021) 118 e2102206118. DOI: 10.1073/pnas.2102206118 108. Yan J., Wu X., Yu J., Zhu Y., Cang S.. Prognostic Role of Tumor Mutation Burden Combined With Immune Infiltrates in Skin Cutaneous Melanoma Based on Multi-Omics Analysis. Front. Oncol. (2020) 10 2403. DOI: 10.3389/fonc.2020.570654 109. Zhang A.Z., Yuan X., Liang W.H., Zhang H.J., Li Y., Xie Y.F., Li J.F., Jiang C.H., Li F.P., Shen X.H.. Immune Infiltration in Gastric Cancer Microenvironment and Its Clinical Significance. Front. Cell Dev. Biol. (2021) 9 762029. DOI: 10.3389/fcell.2021.762029
--- title: 'Symptomatic COVID-19 in Pregnancy: Hospital Cohort Data between May 2020 and April 2021, Risk Factors and Medicolegal Implications' authors: - Marianna Maranto - Simona Zaami - Vincenzo Restivo - Donatella Termini - Antonella Gangemi - Mario Tumminello - Silvia Culmone - Valentina Billone - Gaspare Cucinella - Giuseppe Gullo journal: Diagnostics year: 2023 pmcid: PMC10047111 doi: 10.3390/diagnostics13061009 license: CC BY 4.0 --- # Symptomatic COVID-19 in Pregnancy: Hospital Cohort Data between May 2020 and April 2021, Risk Factors and Medicolegal Implications ## Abstract Pregnancy does not appear to increase susceptibility to SARS-CoV-2 infection, but some physiological changes, such as the reduction of residual functional volumes, elevation of the diaphragm, and impaired cellular immunity, may increase the risk of severe disease and result in a higher risk of complications. The article’s primary objective is to evaluate the factors associated with symptomatic COVID-19 disease in pregnancy. The secondary objective is to describe maternal and neonatal outcomes and cases of vertical transmission of the infection. All pregnant women hospitalized with SARS-CoV2 infection were included in a prospective study in the UOC of Obstetrics and Gynecology, AOOR Villa Sofia—Cervello, Palermo, between May 2020 and April 2021. The patients who requested the termination of the pregnancy according to Law $\frac{194}{78}$ were excluded. We included 165 pregnancies with a total number of 134 deliveries. Overall, $88.5\%$ of the patients were asymptomatic at the time of admission and $11.5\%$ were symptomatic. Of them, $1.8\%$ of the patients required hospital admission in the intensive care unit. Symptoms occurrences were positively associated with the increase in maternal BMI (OR 1.17; $$p \leq 0.002$$), the prematurity (OR 4.71; $$p \leq 0.022$$), and at a lower birth weight (OR 0.99; $$p \leq 0.007$$). One infant tested positive for SARS-CoV2 nasopharyngeal swab; $11.4\%$ of newborns had IgG anti SARS-CoV2 at birth; IgM was positive in $2.4\%$ of newborns. There was no difference statistically significant difference in the vertical transmission of the infection among the group of symptomatic pregnant women and that of asymptomatic pregnant women. ## 1. Introduction Coronavirus Disease-19 (COVID-19) is caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) and includes several characterizations, from asymptomatic patients to respiratory failure, cardiac and cardiovascular complications, thromboembolic and inflammatory complications. Pregnancy does not appear to increase susceptibility to this infection, even if the entry into respiratory cells of SARS-CoV-2 is mediated by ACE2, and its expression increases during pregnancy, which may provide favorable conditions for SARS-CoV-2 infection [1]. Physiological changes during pregnancy, such as reduced functional residual volumes, diaphragm elevation, and altered cell immunity, may be at increased risk for severe disease, necessitating maternal intensive care unit admission, mechanical ventilation, and, in rare cases, extracorporeal membrane oxygenation [2,3]. Deaths have been reported equally in pregnant and non-pregnant women of reproductive age [4]. Among pregnant women, especially those who develop COVID-19 pneumonia, there is an increased risk of preeclampsia, preterm, and cesarean delivery due to fever and hypoxemia [5,6]. In this regard, it is worth mentioning a multinational cohort study which enrolled a total of 2130 women during the first phase of the pandemic, 706 of whom had COVID-19 and 1424 who were uninfected. Women diagnosed with COVID-19 were at higher risk of preeclampsia/eclampsia, serious infections, need for intensive care, maternal mortality, preterm delivery including iatrogenic, perinatal morbidity, and mortality. Although the presence of fever and dyspnea was associated with a higher risk of serious maternal and perinatal complications, asymptomatic patients were also at higher risk for maternal morbidity and preeclampsia [7,8,9,10]. Vertical transmission of SARS-CoV-2 appears to be limited, although it should not be ruled out altogether in light of the viral presence in placental villi and fetal membranes, which points to the possibility that the virus may be able to access the placenta and affect fetal development [1]. At any rate, the SARS-CoV-2-related cytokine storm could bring about higher morbidity and mortality rates in pregnant women, and possibly even pose a threat to the developing fetus and neonate, even in the absence of vertical viral transmission. For these reasons, more effective, evidence-based strategies, models, and targets need to be outlined, for the ultimate purpose of mitigating the impact of viral infection and improving maternal and fetal outcomes [1]. Furthermore, this evaluation should be useful in terms of raising awareness and clinical management abilities in the case of a new pandemic infection. Information on SARS-CoV2-infected pregnancies is evolving rapidly and it is critical to collect data to plan for best practice. The main purpose of this study is to evaluate maternal and neonatal outcomes of pregnant women with SARS-CoV-2 infection. The secondary objective is to describe cases of vertical transmission of SARS-CoV2 infection. ## 2.1. Objectives This study accounted for 165 pregnant women, with a total number of 134 deliveries, while 31 pregnant women were discharged before delivery due to improvement of clinical symptoms. The patients were divided into groups based on the “NIH COVID-19 Treatment Guidelines” [11,12]:✓Asymptomatic or presymptomatic infection: positive test for SARS-CoV-2 but no symptoms;✓Mild illness: any signs and symptoms (e.g., fever, cough, sore throat, malaise, headache, muscle pain) without shortness of breath, dyspnea, or abnormal chest imaging;✓Moderate illness: evidence of lower respiratory disease by clinical assessment or imaging and a saturation of oxygen (SaO2) ≥ 94 percent at room temperature at sea level—severe illness: respiratory frequency > 30 breaths per minute, SaO2 < 94 percent on room air at sea level, ratio of arterial partial pressure of oxygen to fraction of inspired oxygen (PaO2/FiO2) < 300, or lung infiltrates > 50 percent;✓Critical illness: respiratory failure, septic shock, and/or multiple organ dysfunction. ## 2.2. Inclusion and Exclusion Criteria In this retrospective study, all pregnant women hospitalized with SARS-CoV-2 infection in the Department of Obstetrics and Gynecology, Villa Sofia—Cervello Hospitals, Palermo, Sicily, between May 2020 and April 2021, were included. Pregnant women requiring termination of pregnancy as well as non-pregnant women hospitalized for other gynecological conditions were excluded. ## 2.3. Maternal and Fetal Outcomes We collected personal and anamnestic data, reason and duration of hospitalization, duration of positivity to the nasopharyngeal molecular swab, symptoms, saturation, need for oxygen therapy or access to the intensive care unit, diagnostic tests and therapy. Fetal status was assessed by ultrasound and, in third-trimester pregnancies, with cardiotocography. We also evaluated pathologies of pregnancies and outcomes such as live birth, miscarriage, and stillbirth. For patients who gave birth with SARS-CoV-2 infection, data on delivery and any postpartum complications were collected. ## 2.4. Neonatal Outcomes All newborns underwent a thorough review of their history, including gestational age, sex, birth weight, Apgar Score, type of feeding, length of hospitalization, symptoms for up to one week of age, such as respiratory distress, oxygen desaturation, feeding problems, fever, and/or seizures, heart rate anomalies thorough clinical examination, review of laboratory evaluations, rRT-PCR analysis of nasopharyngeal swab samples, and COVID-19 serology to determine SARS-CoV-2 status for those born to SARS-CoV-2-positive mothers. ## 2.5. Evaluation Vertical Transmission All newborns from SARS-CoV-2 positive mothers were tested via a SARS-CoV-2 quantitative rRT-PCR nasopharyngeal swab at birth on day 3 and/or day 7 during their hospital stay. In case of positive result, neonates were re-tested on day 14. ## 2.6. Statistical Analysis Socio-demographic and clinical characteristics of all the recruited pregnant women were summarized using frequencies and percentages. In order to evaluate the distribution of quantitative variables such as age, the skewness and kurtosis test was performed. Mean and standard deviation (SD) were chosen for the normal distribution of these variables, while median and interquartile range (IQR) were used for the non-normal distribution. The differences in quantitative variables normally and not normally distributed among pregnant with COVID-19 infection were evaluated, respectively, with the Student’s t test and with the Wilcoxon and Mann–Whitney test, while the Chi2 test was used for the qualitative variables. Bivariable analyses were performed to assess the associations between factors allegedly linked to symptomatic (mild, moderate, and critical illness) COVID (Odds ratio (OR) with a confidence interval of $95\%$). The significant ($p \leq 0.05$) factors associated in bivariable analyses were run into a multivariable logistic regression model in order to identify predictors of symptomatic COVID-19. A p-value of <0.05 was considered statistically significant. Statistical analyses were performed using Stata/SE 14.2 (Copyright 1985–2015, StataCorp LLC, 4905 Lakeway Drive, College Station, TX 77845, U.S. Revision 29 January 2018). ## 3. Results As shown in Table 1, 12 patients ($7.3\%$) were hospitalized for COVID-related symptoms, 107 ($64.9\%$) for obstetric reasons, 31 ($18.8\%$) were new mothers transferred from other structures in the region due to a positive swab found in time of delivery, 9 ($5.4\%$) were for abortion, and 6 ($3.6\%$) were for other reasons. At admission, 146 patients ($88.5\%$) were asymptomatic, 8 ($4.8\%$) had mild disease, 9 ($5.5\%$) had moderate/severe disease, and 2 ($1.2\%$) had critical disease. During hospitalization, 11 ($6.7\%$) patients needed oxygen therapy, while the remaining 154 ($93.3\%$) did not. Finally, 3 patients ($1.8\%$) required admission to the intensive care unit (Table 1). The patients were then subdivided into two main groups: asymptomatic and symptomatic. As shown in Table 2, the symptomatic group showed a higher mean BMI than the asymptomatic group (33.9 vs. 28.9, $$p \leq 0.001$$); thus, the mode of delivery, preterm delivery and complications during hospitalization showed a statistically significant difference in the two groups (Table 2). Regarding the 123 newborns admitted to the Department of Neonatology and Neonatal Intensive Care Unit, Villa Sofia Cervello Hospital, a statistically significant difference in birth weight was found between the group of symptomatic versus asymptomatic patients, as shown in Table 3. Overall, 104 babies were born at term and 26 were born preterm; $66.6\%$ of them were adequate for gestational age; $12.2\%$ were small for gestational age (SGA) and $9.8\%$ were large for gestational age (LGA). No neonatal respiratory distress was reported, and only $9\%$ of newborns required noninvasive respiratory support at birth. During their hospital stay, all infants remained asymptomatic, with normal temperature and vital parameters. Only one newborn was positive at nasopharyngeal swab for SARS-CoV2: the first SARS-CoV-2 test resulted positive at birth at their 25th hour of life. Repeated SARS-CoV-2 tests at 25, 48 h, 7 days, and 14 days of life were positive. The SARS-CoV-2 test resulted negative at 21 days of life. Soon after birth, the baby’s serology tested positive for both SARS-CoV-2 immunoglobulin (Ig)-G and Ig-M titers. Overall, $11.4\%$ ($$n = 14$$) of newborns had anti SARS-CoV2 IgG at birth; IgM was positive in $2.4\%$ ($$n = 3$$) of newborns. There was no statistically significant difference in vertical transmission of the infection between the symptomatic and asymptomatic pregnant groups (Table 3). A univariable analysis shows COVID-19 symptoms to be positively associated among maternal characteristics with increased maternal BMI (OR = 1.18, $95\%$ CI 1.06–1.29, p value 0.002) and prematurity (OR 4.71; $$p \leq 0.022$$). After controlling for factors and statistical significance at multivariable analysis, only the unit increase of BMI (OR = 1.18, CI$95\%$ 1.04–1.35, $$p \leq 0.011$$) was associated with being symptomatic for COVID-19 among the maternal outcome (Table 4). At univariable analysis, COVID-19 symptoms were positively associated among neonatal characteristics with birth weight (OR = 0.99, CI$95\%$ 0.99–1.00, $$p \leq 0.007$$). After controlling for other factors, COVID-19 symptoms of the mother were associated with the use of ventilation by the infants (OR = 26.95 CI$95\%$ = 1.26–574.29, $$p \leq 0.035$$) (Table 5). ## 4.1. Risk-Factor Assessment The article relies on data from the first cases of SARS-CoV2 positive pregnancies. The main results were that high BMI values were associated with a higher risk of symptomatic disease. These data are in agreement with the scientific literature that associates obesity with adverse outcomes from COVID-19 in the general population. Indeed, SARS-CoV-2 had the ability to gain entry into human cells by direct binding to ACE2 receptors on host cells [13]. Obese individuals have been found to be more susceptible to COVID-19 infection, which is likely due to the higher density of ACE2 in adipose tissue [14]. Furthermore, in animal models, tumor necrosis factor-α (TNF-α) is a multifunctional cytokine expressed in adipose tissue capable of influencing insulin-induced signaling and preventing glucose transporter type 4 (GLUT-4) expression, which gives rise to higher levels of free fatty acids (FFA), and worsening insulin resistance [15]. Immune inflammation pathways can be triggered by overly high levels of FFAs via several signaling pathways, ultimately leading to TNF-α, interleukin-6 (IL-6), leptin, and resistin [16], all of which play a direct role in the differentiation of monocytes into activated M1 macrophages. Inflammatory cytokines, active oxygen radicals, and nitric oxide (NO) can originate from M1 macrophages, and can negatively impact the endogenous immune response to pathogens [16]. The inflammatory response caused by obesity thus results in more pronounced cell aggregation and higher levels of cytokine production. Nutritional status therefore has a major role in the development of COVID-19 complications, hospitalization length, and mortality rates, as pointed out by research findings from diverse populations [17]. Ultimately, obesity has been linked to higher rates of major complications due to its considerable impact on immune responses. In particular, an interesting review article has pointed to obesity as a considerable factor in the likelihood of incurring severe COVID-19 complications and the need for ICU admission, intubation, and even higher mortality rates. Such evidence makes it necessary to keep overweight and obese patients under close observation and monitoring at all times [18]. The second fundamental aspect which seems to emerge from the multivariate analysis is the linkage between symptomaticity and prematurity, and consequently low weight at birth. Recent findings arising from an analysis of data gathered between April to May 2017 to 2019 and April to May 2020 pointed to a lower rate of prematurity (from $5.31\%$ to $4.91\%$, $p \leq 0.01$). Furthermore, a decrease in the rate of prematurity was still observed after the end of lockdown (from June to September 2020) for singleton pregnancies. However, among the 1752 SARS-CoV-2-positive patients with singleton pregnancies, a higher prematurity rate was reported in 2020 than in 2017 to 2019 ($9.93\%$ vs. $5.32\%$; $p \leq 0.01$), regardless of the severity of prematurity. On the other hand, a lower prematurity rate was reported in uninfected or untested in 202 patients compared to those who gave birth in the 2017–2019 period ($4.67\%$ vs. $5.32\%$; $p \leq 0.01$), irrespective of prematurity severity [19]. A lower rate of preterm births during the COVID-19 pandemic has been reported by various sources [20,21,22,23], although it is worth noting that such an overall decrease does not account for SARS-CoV positive and negative patients, unlike the present study. In any case, some studies have reported that such a decrease was limited to deliveries to white patients residing in more affluent neighborhoods, and deliveries at non-outpatient care facilities; such findings may be due to the fact that COVID-19 response measures may have benefited women with more indicators of advantage [24]. A recent noteworthy retrospective cohort study has focused on the clinical manifestations, complications, and maternal–fetal outcomes in women with SARS-CoV-2 infection during delivery and divided patients into two groups: symptomatic and asymptomatic. Compared to asymptomatic patients, symptomatic pregnant women at the time of delivery were found to have slightly higher, though not significant, preterm delivery and cesarean section rates, in addition to lower neonatal birthweights and Apgar score, [25]. As for the high rate of caesarean sections, a retrospective review of case records in India compared outcomes (cesarean section rate, maternal and neonatal ICU admission, and feto-maternal mortality) in positive and negative pregnant women at delivery. Similar to our results, considerably higher cesarean section rates were reported among women with COVID-19. Furthermore, viral RNA was detected in the cord blood and nasopharyngeal swab of one infant [26]. Other studies also point to cesarean section as the most widespread delivery modality in parturient women with COVID-19. In particular, another Indian study accounting for 44 women undergoing cesarean section during the study period, with elective and emergency surgeries of 22 each, showed that no indication other than COVID-19 status was reported in 13 out of 44 patients [27]. In order to prevent the host COVID-19 complication herein laid out, the anti-SARS-CoV-2 vaccination is a safe and effective tool even in pregnancy. In this regard, an interesting mathematical model proposed by an Indian group that studied the transmission dynamics was associated with the decrease of COVID-19, underlining the importance of non-pharmaceutical interventions and vaccination as a strategy for the control of COVID-19 [28]. However, a part of the population of pregnant women still shows vaccine hesitancy, for which suitable counseling by gynecologists is certainly a valuable option worth pursuing [29]. A systematic review of a small sample of 6 early pandemic studies shows that, although vertical transmission of severe acute respiratory syndrome arising from coronavirus infection has so far been ruled out, and maternal and neonatal outcomes have been favorable overall, preterm delivery rates by cesarean section are still worrisome [30]. In any case, COVID-19, which is linked to respiratory insufficiency in late pregnancies, can undoubtedly give rise to a complex clinical scenario [6,31]. A review centered around 36 research studies has focused on deliveries in 203 SARS-CoV-2 positive pregnant women. Rather similar levels of disease severity in pregnant as opposed to non-pregnant women were reported. The majority of patients, $68.9\%$, ultimately gave birth via cesarean section, with COVID-19 status as the sole common indication [32]. As for the management of newborns at our hospital, the decision to separate newborns is necessarily made on a case-by-case basis, and is shared and agreed upon by mothers and the healthcare professionals based on a thorough risks vs. benefits evaluation. Mother and newborn were separated in the case of maternal severe clinical symptoms, or after surgery in cases of caesarean section, due to the impossibility of taking care of the newborn independently. Mothers were counseled prior to discharge about home isolation and precautions, according to guidelines from the Italian Society of Neonatology. Clinical follow-up for infants was provided remotely and a repeat test for SARS-CoV-2 was administered 7 or 14 days after discharge, and then at 1 month after discharge. Only one newborn tested positive at our facility. Furthermore, the mother wore a face mask throughout the hospital stay, had no skin-to-skin contact with the baby, no direct breastfeeding, no visitors (including parents) were allowed during the newborn’s first 14 days of life, and strict droplet isolation precautions took place. All such precautions notwithstanding, the newborn tested positive for SARS-CoV-2 at birth. In other studies, SARS-CoV-2 positive infants were also observed, but no definitive evidence of vertical transmission remains because available data are still insufficient [30]. Several studies now show that the SARS-CoV-2 genome can be detected in umbilical cord blood and placenta at term, and infants demonstrate elevated levels of SARS-CoV-2 specific IgG and IgM antibodies [33,34]. Although specimens of placental tissue or amniotic fluid found positive for any pathogen are considered a diagnostic sign of maternal infection, further confirmatory testing is needed before it can be deemed a sign of neonatal congenital infection [35]. Hence, it stands to reason that even though SARS-CoV-2 has been reportedly found by RT-PCR in the placenta, as reflected by various findings, a positive RT-PCR test in the fetus/neonate does not necessarily follow [36,37,38]. By the same token, positivity detected in the amniotic fluid does not necessarily entail fetal positivity. Even though RT-PCR in amniotic fluid have detected the presence of SARS-CoV-2 in a relatively limited number of case reports, not all infants were confirmed to be infected [39,40,41,42]. By virtue of such findings, a positive SARS-CoV-2 assay on amniotic fluid or placenta alone is not enough to provide a reliable level of confirmatory proof as to actual in-utero infection. As far as the contamination of umbilical cord blood is concerned, that is thought to possibly take place because of cross-contamination with maternal blood during sample collection, or blood cells from the mother getting into the fetal bloodstream through the placenta over the course of gestation, or most commonly, maternal blood cells getting into fetal circulation during labor, due to contractions of the uterus [43,44,45]. In light of such dynamics, confirmatory testing via fetal/neonatal peripheral blood sample or testing of another sterile or non-sterile sample is needed in addition to PCR. It is likely that studies centered around placental tissue are the most significant in terms of providing insight as to SARS-CoV-2 vertical transmission. A Brazilian study on five pregnant women infected with SARS-CoV-2 before vaccination who gave birth to a stillborn child investigated placental alterations compared to a prepandemic sample. It is worth noting that RT-PCRq found SARS-CoV-2 RNA in three out of five placentas at least two to twenty weeks following primary pregnancy infection symptoms. Moreover, immunoperoxidase assays showed SARS-CoV-2 spike protein in all placental samples. Ultrastructural aspects of the infected placentas showed similar alteration patterns between the samples, with predominant characteristics of apoptosis and detection of virus-like particles [46]. ## 4.2. Medicolegal Implications In light of the risk factors associated with COVID-19 in pregnancy, both in terms of maternal and neonatal outcomes, it is worth briefly elaborating on the medicolegal repercussions that might arise from non-compliance with specific evidence based recommendations, guidelines, and best practices. The need to take into account the host of immunological changes which occur during pregnancy, over the third trimester especially, stems from the fact that such adjustments make women more vulnerable and more likely to develop major severe symptoms from SARS-CoV-2 infection; similar dynamics have also been shown with previous similar epidemics such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) [47,48]. COVID-19 vaccines based on mRNA have been shown to pose no safety concerns during pregnancy or breastfeeding, and they do not affect fertility [49]. A possible element complicating the decision-making process and the implementation of suitable treatment options is due the fact that COVID-19 during pregnancy often shows signs and symptoms similar to those in non-pregnant patients [50,51], although one systematic review found that pregnant and recently pregnant people were less likely to manifest fever, cough, dyspnea, and myalgia than non-pregnant females of reproductive age [52]. When assessing pregnant symptomatic people without fever, it is worth bearing in mind that it may be difficult to differentiate between several COVID-19 manifestations and common pregnancy symptoms, such as nausea, shortness of breath, and fatigue. Providing thorough counseling to pregnant patients as to the risks of COVID-19 infection is also key: the increased risk for severe disease from SARS-CoV-2 during pregnancy ought to be discussed and recommendations for the effective protection from infection should be given [53]. If a pregnant patient should need hospitalization due to COVID-19 infection, it ought to be at a hospital capable of guaranteeing maternal and fetal monitoring should the need arise. COVID-19 in pregnancy should be managed by ensuring fetal and uterine contraction monitoring, based on gestational age, whenever deemed advisable. In addition, delivery planning should be adequately designed and implemented on a case-by-case basis by relying on a multidisciplinary, team-based approach that may include consultation with obstetric, maternal-fetal medicine, infectious disease, pulmonary-critical care, and pediatric specialists, whenever deemed necessary. Provable and documented compliance with recommendations and evidence-based criteria outlined and released by scientific societies and institutions (such as the Centers for Disease Control and Prevention [54,55], the American College of Obstetricians and Gynecologists [56], and the Society for Maternal-Fetal Medicine [57], among others) can greatly contribute to ensuring that care for pregnant patients with COVID-19 is delivered in a viable fashion, from a medicolegal perspective, in order to shield healthcare professionals from negligence-based malpractice allegations in case of adverse outcomes [58,59,60]. Novel telemedicine-based methods of providing care and counseling to pregnant women need to take into account the relevant norms and regulations [61,62], as well as the unique complexities that such innovative practices entail from a legal and ethical standpoint [63,64]. ## 5. Conclusions The present study is based on the identification of patients at greater risk of contracting a symptomatic form of COVID 19 in order to reduce the onset of complications and, consequently, stem the increase in preterm deliveries, with possible sequelae in the short, medium and long term. Overall, findings from the present study support the claim that neonates born to mothers with confirmed or suspected SARS-CoV-2 are mostly asymptomatic, and therefore their status is not associated with worse clinical outcomes. However, neonatal critical illness is still a possibility; administering a nasopharyngeal swab at least at 24 h after birth and monitoring the infants for possible symptoms for 14 days after birth are necessary precautions and major contributors to medicolegal viability in case of adverse outcomes, as is long-term follow-up. Parents must be thoroughly counseled and directly involved in the decision-making process with options for rooming in, skin-to-skin contact, and breastfeeding with appropriate protective equipment, taking into account the organization of the hospital that hosts the pregnant patient and the newborn. ## References 1. Verma S., Carter E.B., Mysorekar I.U.. SARS-CoV-2 and Pregnancy: An Invisible Enemy?. Am. J. Reprod. Immunol. (2020.0) 84 e13308. DOI: 10.1111/aji.13308 2. Dashraath P., Wong J.L.J., Lim M.X.K., Lim L.M., Li S., Biswas A., Choolani M., Mattar C., Su L.L.. Coronavirus Disease 2019 (COVID-19) Pandemic and Pregnancy. Am. J. Obstet. Gynecol. (2020.0) 222 521-531. DOI: 10.1016/j.ajog.2020.03.021 3. Kasraeian M., Zare M., Vafaei H., Asadi N., Faraji A., Bazrafshan K., Roozmeh S.. COVID-19 Pneumonia and Pregnancy; a Systematic Review and Meta-Analysis. J. Matern. Fetal Neonatal Med. (2022.0) 35 1652-1659. DOI: 10.1080/14767058.2020.1763952 4. Ellington S., Strid P., Tong V.T., Woodworth K., Galang R.R., Zambrano L.D., Nahabedian J., Anderson K., Gilboa S.M.. Characteristics of Women of Reproductive Age with Laboratory-Confirmed SARS-CoV-2 Infection by Pregnancy Status–United States, January 22–June 7, 2020. MMWR Morb. Mortal. Wkly. Rep. (2020.0) 69 769-775. DOI: 10.15585/mmwr.mm6925a1 5. Khoury R., Bernstein P.S., Debolt C., Stone J., Sutton D.M., Simpson L.L., Limaye M.A., Roman A.S., Fazzari M., Penfield C.A.. Characteristics and Outcomes of 241 Births to Women with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection at Five New York City Medical Centers. Obstet. Gynecol. (2020.0) 136 273-282. DOI: 10.1097/AOG.0000000000004025 6. Gullo G., Cucinella G., Tumminello M., Renda B., Donzelli M., Lo Bue V., Termini D., Maranto M., De Tommasi O., Tarantino F.. Convalescent Plasma Use in Pregnant Patients with COVID-19 Related ARDS: A Case Report and Literature Review. Ital. J. Gynaecol. Obs. (2022.0) 34 228. DOI: 10.36129/jog.2022.32 7. Villar J., Ariff S., Gunier R.B., Thiruvengadam R., Rauch S., Kholin A., Roggero P., Prefumo F., Do Vale M.S., Cardona-Perez J.A.. Maternal and Neonatal Morbidity and Mortality among Pregnant Women with and without COVID-19 Infection: The INTERCOVID Multinational Cohort Study. JAMA Pediatr. (2021.0) 175 817. DOI: 10.1001/jamapediatrics.2021.1050 8. Papageorghiou A.T., Deruelle P., Gunier R.B., Rauch S., García-May P.K., Mhatre M., Usman M.A., Abd-Elsalam S., Etuk S., Simmons L.E.. Preeclampsia and COVID-19: Results from the INTERCOVID Prospective Longitudinal Study. Am. J. Obstet. Gynecol. (2021.0) 225 289.e1-289.e17. DOI: 10.1016/j.ajog.2021.05.014 9. Villar J., Soto Conti C.P., Gunier R.B., Ariff S., Craik R., Cavoretto P.I., Rauch S., Gandino S., Nieto R., Winsey A.. Pregnancy Outcomes and Vaccine Effectiveness during the Period of Omicron as the Variant of Concern, INTERCOVID-2022: A Multinational, Observational Study. Lancet (2023.0) 401 447-457. DOI: 10.1016/S0140-6736(22)02467-9 10. Eskenazi B., Rauch S., Iurlaro E., Gunier R.B., Rego A., Gravett M.G., Cavoretto P.I., Deruelle P., García-May P.K., Mhatre M.. Diabetes Mellitus, Maternal Adiposity, and Insulin-Dependent Gestational Diabetes Are Associated with COVID-19 in Pregnancy: The INTERCOVID Study. Am. J. Obstet. Gynecol. (2022.0) 227 74.e1-74.e16. DOI: 10.1016/j.ajog.2021.12.032 11. Clinical Spectrum of SARS-CoV-2 Infection. Issued on 26 September 2022 12. Bertino E., Spada E., Occhi L., Coscia A., Giuliani F., Gagliardi L., Gilli G., Bona G., Fabris C., De Curtis M.. Neonatal Anthropometric Charts: The Italian Neonatal Study Compared with Other European Studies. J. Pediatr. Gastroenterol. Nutr. (2010.0) 51 353-361. DOI: 10.1097/MPG.0b013e3181da213e 13. Ni W., Yang X., Yang D., Bao J., Li R., Xiao Y., Hou C., Wang H., Liu J., Yang D.. Role of Angiotensin-Converting Enzyme 2 (ACE2) in COVID-19. Crit. Care (2020.0) 24 422. DOI: 10.1186/s13054-020-03120-0 14. Couselo-Seijas M., Almengló C., Agra-Bermejo R.M., Luis Fernandez Á., Alvarez E., R González-Juanatey J., Eiras S.. Higher ACE2 Expression Levels in Epicardial Cells than Subcutaneous Stromal Cells from Patients with Cardiovascular Disease: Diabetes and Obesity as Possible Enhancer. Eur. J. Clin. Investig. (2021.0) 51 e13463. DOI: 10.1111/eci.13463 15. Jang D.-I., Lee A.-H., Shin H.-Y., Song H.-R., Park J.-H., Kang T.-B., Lee S.-R., Yang S.-H.. The Role of Tumor Necrosis Factor Alpha (TNF-α) in Autoimmune Disease and Current TNF-α Inhibitors in Therapeutics. Int. J. Mol. Sci. (2021.0) 22. DOI: 10.3390/ijms22052719 16. Makki K., Froguel P., Wolowczuk I.. Adipose Tissue in Obesity-Related Inflammation and Insulin Resistance: Cells, Cytokines, and Chemokines. ISRN Inflamm. (2013.0) 2013 139239. DOI: 10.1155/2013/139239 17. Albarrán-Sánchez A., Ramírez-Rentería C., Anda-Garay J.C., Noyola-García M.E., Alberti-Minutti P., Flores-Padilla G., Guizar-García L.A., Contreras-García C.E., Marrero-Rodríguez D., Taniguchi-Ponciano K.. Differences in Mortality Rate among Patients Hospitalized with Severe COVID-19 According to Their Body Mass Index. Obes. Sci. Pract. (2022.0) 8 423-432. DOI: 10.1002/osp4.584 18. Vassilopoulou E., Bumbacea R.S., Pappa A.K., Papadopoulos A.N., Bumbacea D.. Obesity and Infection: What Have We Learned from the COVID-19 Pandemic. Front. Nutr. (2022.0) 9 931313. DOI: 10.3389/fnut.2022.931313 19. Simon E., Cottenet J., Mariet A.-S., Bechraoui-Quantin S., Rozenberg P., Gouyon J.-B., Quantin C.. Impact of the COVID-19 Pandemic on Preterm Birth and Stillbirth: A Nationwide, Population-Based Retrospective Cohort Study. Am. J. Obstet. Gynecol. (2021.0) 225 347-348. DOI: 10.1016/j.ajog.2021.05.015 20. Been J.V., Burgos Ochoa L., Bertens L.C.M., Schoenmakers S., Steegers E.A.P., Reiss I.K.M.. Impact of COVID-19 Mitigation Measures on the Incidence of Preterm Birth: A National Quasi-Experimental Study. Lancet Public Health (2020.0) 5 e604-e611. DOI: 10.1016/S2468-2667(20)30223-1 21. Meyer R., Bart Y., Tsur A., Yinon Y., Friedrich L., Maixner N., Levin G.. A Marked Decrease in Preterm Deliveries during the Coronavirus Disease 2019 Pandemic. Am. J. Obstet. Gynecol. (2021.0) 224 234-237. DOI: 10.1016/j.ajog.2020.10.017 22. Philip R.K., Purtill H., Reidy E., Daly M., Imcha M., McGrath D., O’Connell N.H., Dunne C.P.. Unprecedented Reduction in Births of Very Low Birthweight (VLBW) and Extremely Low Birthweight (ELBW) Infants during the COVID-19 Lockdown in Ireland: A “natural Experiment” Allowing Analysis of Data from the Prior Two Decades. BMJ Glob. Health (2020.0) 5 e003075. DOI: 10.1136/bmjgh-2020-003075 23. Hedermann G., Hedley P.L., Bækvad-Hansen M., Hjalgrim H., Rostgaard K., Poorisrisak P., Breindahl M., Melbye M., Hougaard D.M., Christiansen M.. Danish Premature Birth Rates during the COVID-19 Lockdown. Arch. Dis. Child. Fetal Neonatal Ed. (2021.0) 106 93-95. DOI: 10.1136/archdischild-2020-319990 24. Lemon L., Edwards R.P., Simhan H.N.. What Is Driving the Decreased Incidence of Preterm Birth during the Coronavirus Disease 2019 Pandemic?. Am. J. Obstet. Gynecol. MFM (2021.0) 3 100330. DOI: 10.1016/j.ajogmf.2021.100330 25. Grgić G., Cerovac A., Hudić I., Laganà A.S., Favilli A., Garzon S., Chiantera V., Margioula-Siarkou C., Hadžimehmedović A., Mandžić A.. Clinical Manifestation and Obstetric Outcomes in Pregnant Women with SARS-CoV-2 Infection at Delivery: A Retrospective Cohort Analysis. J. Pers. Med. (2022.0) 12. DOI: 10.3390/jpm12091480 26. Ahmad S.N., Sameen D., Dar M.A., Jallu R., Shora T.N., Dhingra M.. Do SARS-CoV-2-Infected Pregnant Women Have Adverse Pregnancy Outcomes as Compared to Non-Infected Pregnant Women?. Int. J. Women’s Health (2022.0) 14 1201-1210. DOI: 10.2147/IJWH.S375739 27. Venkateswaran V., Parida R., Khanna P., Bhoi D., Singh A.K., Mathur P., Sahoo D., Dass C., Gupta A., Aravindan A.. Maternal and Neonatal Characteristics, Operative Details and Outcomes in COVID-19 Positive Parturients Undergoing Cesarean Sections: A Retrospective Observational Study. J. Anaesthesiol. Clin. Pharmacol. (2022.0) 38 S52-S57. DOI: 10.4103/joacp.joacp_358_21 28. Das P., Upadhyay R.K., Misra A.K., Rihan F.A., Das P., Ghosh D.. Mathematical model of COVID-19 with comorbidity and controlling using non-pharmaceutical interventions and vaccination. Nonlinear Dyn. (2021.0) 106 1213-1227. DOI: 10.1007/s11071-021-06517-w 29. Maranto M., Gullo G., Bruno A., Minutolo G., Cucinella G., Maiorana A., Casuccio A., Restivo V.. Factors Associated with Anti-SARS-CoV-2 Vaccine Acceptance among Pregnant Women: Data from Outpatient Women Experiencing High-Risk Pregnancy. Vaccines (2023.0) 11. DOI: 10.3390/vaccines11020454 30. Della Gatta A.N., Rizzo R., Pilu G., Simonazzi G.. Coronavirus Disease 2019 during Pregnancy: A Systematic Review of Reported Cases. Am. J. Obstet. Gynecol. (2020.0) 223 36-41. DOI: 10.1016/j.ajog.2020.04.013 31. Lau C.-H., Mao C.-L., Chang Y.-K., Chiu S.-K., Lan C.-C., Zhao L.-L., Lin H.-Y., Huang S.-C., Chiu H.-C.. Emergent Cesarean Section in a Preterm Pregnant Woman with Severe COVID-19 Pneumonia in Taiwan: A Case Report. Taiwan. J. Obstet. Gynecol. (2022.0) 61 868-872. DOI: 10.1016/j.tjog.2022.02.047 32. Debrabandere M.L., Farabaugh D.C., Giordano C.. A Review on Mode of Delivery during COVID-19 between December 2019 and April 2020. Am. J. Perinatol. (2021.0) 38 332-341. DOI: 10.1055/s-0040-1721658 33. Pomorski M., Trzeszcz M., Matera-Witkiewicz A., Krupińska M., Fuchs T., Zimmer M., Zimmer-Stelmach A., Rosner-Tenerowicz A., Budny-Wińska J., Tarczyńska-Podraza A.. SARS-CoV-2 Infection and Pregnancy: Maternal and Neonatal Outcomes and Placental Pathology Correlations. Viruses (2022.0) 14. DOI: 10.3390/v14092043 34. Boncompagni A., De Agostini M., Lugli L., Ternelli G., Colonna V., Biagioni E., Bonasoni M.P., Salviato T., Gabrielli L., Falconi M.. Unexpected Vertical Transmission of SARS-CoV-2: Discordant Clinical Course and Transmission from Mother to Newborn. Microorganisms (2022.0) 10. DOI: 10.3390/microorganisms10091718 35. Definition and Categorization of the Timing of Mother-to-Child Transmission of SARS-CoV-2. Scientific Brief. Issued on 8th February 2021 36. Yu N., Li W., Kang Q., Zeng W., Feng L., Wu J.. No SARS-CoV-2 Detected in Amniotic Fluid in Mid-Pregnancy. Lancet Infect. Dis. (2020.0) 20 1364. DOI: 10.1016/S1473-3099(20)30320-0 37. Michailidou D., Stavridou A., Panagouli E.D., Sergentanis T.N., Psaltopoulou T., Bacopoulou F., Baltag V., Greydanus D.E., Mastorakos G., Chrousos G.P.. The Impact of COVID-19 during Pregnancy on Maternal and Neonatal Outcomes: A Systematic Review. EMBnet J. (2021.0) 26 e969. DOI: 10.14806/ej.26.1.969 38. McDevitt K.E.M., Ganjoo N., Mlangeni D., Pathak S.. Outcome of Universal Screening of Neonates for COVID-19 from Asymptomatic Mothers. J. Infect. (2020.0) 81 452-482. DOI: 10.1016/j.jinf.2020.06.037 39. Zamaniyan M., Ebadi A., Aghajanpoor S., Rahmani Z., Haghshenas M., Azizi S.. Preterm Delivery, Maternal Death, and Vertical Transmission in a Pregnant Woman with COVID-19 Infection. Prenat. Diagn. (2020.0) 40 1759-1761. DOI: 10.1002/pd.5713 40. Palalioglu R.M., Mahammadaliyeva A., Erbiyik H.I., Muhcu M.. COVID-19 in Third Trimester May Not Be as Scary as You Think, It Can Be Innocent: Evaluating Vertical Transmission from a COVID-19 Positive Asymptomatic Pregnant Woman with Early Membrane Rupture. J. Obstet. Gynaecol. Res. (2021.0) 47 838-842. DOI: 10.1111/jog.14584 41. Schwartz D.A., Mohagheghi P., Beigi B., Zafaranloo N., Moshfegh F., Yazdani A.. Spectrum of Neonatal COVID-19 in Iran: 19 Infants with SARS-CoV-2 Perinatal Infections with Varying Test Results, Clinical Findings and Outcomes. J. Matern. Fetal Neonatal Med. (2022.0) 35 2731-2740. DOI: 10.1080/14767058.2020.1797672 42. Angelidou A., Sullivan K., Melvin P.R., Shui J.E., Goldfarb I.T., Bartolome R., Chaudhary N., Vaidya R., Culic I., Singh R.. Association of Maternal Perinatal SARS-CoV-2 Infection with Neonatal Outcomes during the COVID-19 Pandemic in Massachusetts. JAMA Netw. Open (2021.0) 4 e217523. DOI: 10.1001/jamanetworkopen.2021.7523 43. Masuzaki H., Miura K., Miura S., Yoshiura K.-I., Mapendano C.K., Nakayama D., Yoshimura S., Niikawa N., Ishimaru T.. Labor Increases Maternal DNA Contamination in Cord Blood. Clin. Chem. (2004.0) 50 1709-1711. DOI: 10.1373/clinchem.2004.036517 44. Kumar M., Abbas Z., Azami M., Belopolskaya M., Dokmeci A.K., Ghazinyan H., Jia J., Jindal A., Lee H.C., Lei W.. Asian Pacific Association for the Study of Liver (APASL) Guidelines: Hepatitis B Virus in Pregnancy. Hepatol. Int. (2022.0) 16 211-253. DOI: 10.1007/s12072-021-10285-5 45. Penfield C.A., Brubaker S.G., Limaye M.A., Lighter J., Ratner A.J., Thomas K.M., Meyer J.A., Roman A.S.. Detection of Severe Acute Respiratory Syndrome Coronavirus 2 in Placental and Fetal Membrane Samples. Am. J. Obstet. Gynecol. MFM (2020.0) 2 100133. DOI: 10.1016/j.ajogmf.2020.100133 46. Parcial A.L.N., Salomão N.G., Portari E.A., Arruda L.V., de Carvalho J.J., de Matos Guedes H.L., Conde T.C., Moreira M.E., Batista M.M., Paes M.V.. SARS-CoV-2 Is Persistent in Placenta and Causes Macroscopic, Histopathological, and Ultrastructural Changes. Viruses (2022.0) 14. DOI: 10.3390/v14091885 47. COVID-19 and Pregnancy. BMJ (2020.0) 369 m1672. DOI: 10.1136/bmj.m1672 48. Di Mascio D., Khalil A., Saccone G., Rizzo G., Buca D., Liberati M., Vecchiet J., Nappi L., Scambia G., Berghella V.. Outcome of Coronavirus Spectrum Infections (SARS, MERS, COVID-19) during Pregnancy: A Systematic Review and Meta-Analysis. Am. J. Obstet. Gynecol. MFM (2020.0) 2 100107. DOI: 10.1016/j.ajogmf.2020.100107 49. COVID-19, Pregnancy, Childbirth, and Breastfeeding: Answers from Ob-Gyns. Issued on 13 October 2022 50. Khan D.S.A., Hamid L.-R., Ali A., Salam R.A., Zuberi N., Lassi Z.S., Das J.K.. Differences in Pregnancy and Perinatal Outcomes among Symptomatic versus Asymptomatic COVID-19-Infected Pregnant Women: A Systematic Review and Meta-Analysis. BMC Pregnancy Childbirth (2021.0) 21. DOI: 10.1186/s12884-021-04250-1 51. Jamieson D.J., Rasmussen S.A.. An Update on COVID-19 and Pregnancy. Am. J. Obstet. Gynecol. (2022.0) 226 177-186. DOI: 10.1016/j.ajog.2021.08.054 52. Allotey J., Stallings E., Bonet M., Yap M., Chatterjee S., Kew T., Debenham L., Llavall A.C., Dixit A., Zhou D.. Clinical Manifestations, Risk Factors, and Maternal and Perinatal Outcomes of Coronavirus Disease 2019 in Pregnancy: Living Systematic Review and Meta-Analysis. BMJ (2020.0) 370 m3320. DOI: 10.1136/bmj.m3320 53. Rasmussen S.A., Smulian J.C., Lednicky J.A., Wen T.S., Jamieson D.J.. Coronavirus Disease 2019 (COVID-19) and Pregnancy: What Obstetricians Need to Know. Am. J. Obstet. Gynecol. (2020.0) 222 415-426. DOI: 10.1016/j.ajog.2020.02.017 54. Coronavirus Disease 2019 (COVID-19) Treatment Guidelines. National Institutes of Health 55. Considerations for Inpatient Obstetric Healthcare Settings. Issued on 19 November 2021 56. COVID-19 FAQs for Obstetrician-Gynecologists, Obstetrics 57. COVID-19 and Pregnancy: What Maternal-Fetal Medicine Subspecialists Need to Know 58. Montanari Vergallo G., Zaami S.. Guidelines and Best Practices: Remarks on the Gelli-Bianco Law. Clin. Ter. (2018.0) 169 e82-e85. DOI: 10.7417/T.2018.2059 59. Pressman K., Müller R., Krstić N., Običan S.. A Novel Clinic Structure for Exposure Counseling during Pregnancy. Birth Defects Res. (2022.0) 114 855-862. DOI: 10.1002/bdr2.2069 60. Malvasi A., Marinelli E., Ghi T., Zaami S.. ISUOG Practice Guidelines for Intrapartum Ultrasound: Application in Obstetric Practice and Medicolegal Issues. Ultrasound Obstet. Gynecol. (2019.0) 54 421. DOI: 10.1002/uog.20399 61. Gil Membrado C., Barrios V., Cosín-Sales J., Gámez J.M.. Telemedicine, Ethics, and Law in Times of COVID-19. A Look towards the Future. Rev. Clin. Esp. (2021.0) 221 408-410. DOI: 10.1016/j.rce.2021.03.002 62. Aziz A., Zork N., Aubey J.J., Baptiste C.D., D’Alton M.E., Emeruwa U.N., Fuchs K.M., Goffman D., Gyamfi-Bannerman C., Haythe J.H.. Telehealth for High-Risk Pregnancies in the Setting of the COVID-19 Pandemic. Am. J. Perinatol. (2020.0) 37 800-808. DOI: 10.1055/s-0040-1712121 63. Marinelli S., Basile G., Zaami S.. Telemedicine, Telepsychiatry and COVID-19 Pandemic: Future Prospects for Global Health. Healthcare (2022.0) 10. DOI: 10.3390/healthcare10102085 64. Kaplan B.. Revisiting Health Information Technology Ethical, Legal, and Social Issues and Evaluation: Telehealth/Telemedicine and COVID-19. Int. J. Med. Inform. (2020.0) 143 104239. DOI: 10.1016/j.ijmedinf.2020.104239
--- title: Evaluation of the Corneal Endothelium Following Cataract Surgery in Diabetic and Non-Diabetic Patients authors: - Adela-Laura Ciorba - George Roiu - Amir Mohamed Abdelhamid - Sameh Saber - Simona Cavalu journal: Diagnostics year: 2023 pmcid: PMC10047116 doi: 10.3390/diagnostics13061115 license: CC BY 4.0 --- # Evaluation of the Corneal Endothelium Following Cataract Surgery in Diabetic and Non-Diabetic Patients ## Abstract The aim of this study was to evaluate the influence of phacoemulsification cataract surgery on the state of the corneal endothelium in diabetic versus non-diabetic patients. We compared the corneal cell morphology in 48 diabetics with good glycemic control and 72 non-diabetic patients before and after uneventful phacoemulsification. Corneal cell density, central corneal thickness, and hexagonality were measured preoperatively and post-surgery (at 1 and 4 weeks) by specular microscopy. The effect of age, gender, axial length, and anterior chamber depth on the parameters of the corneal endothelium were evaluated. We noticed a drop in the endothelial density in both groups postoperatively: a mean endothelial cell loss of 472.7 ± 369.1 in the diabetic group was recorded versus 165.7 ± 214.6 mean loss in the non-diabetic group after the first week. A significant increase in central corneal thickness was also noticed in both groups one week after phacoemulsification, but no statistical significance after 4 weeks in the diabetic group. In terms of cell hexagonality, statistically significant differences were noticed after 4 weeks in both groups. Overall, a significant difference between diabetic and non-diabetic population was noticed in terms of corneal endothelial cell loss after uneventful phacoemulsification cataract surgery. Routine specular microscopy and HbA1c evaluation is recommended before cataract surgery, while intraoperative precautions and high monitorisation in terms of pacho power intensity and ultrasound energy, along with a proper application of the dispersive viscoelastic substances are essential to reduce the risk of endothelial damage. ## 1. Introduction In 2020, the leading cause of blindness all over the world was cataract, which affected 15.2 million people, among other pathologies such as uncorrected refractive error, glaucoma, age-related macular degeneration, and diabetic retinopathy [1]. Aging is the most common cause of cataract development (senile cataracts over the age of 65), but other factors, such as traumatic injuries (concussion or perforation), systemic diseases, endocrine disorders (diabetes, hypoparathyroidism), radiation, chemicals, or local diseases, such as uveitis and retinal detachment, can also cause cataract formation [2,3]. It is well known that the ability of corneal self-repair occurs within a mechanism characterized by enlargement of the endothelial cells. However, the posterior corneal surface is maintained by a gradual increase in the size of the remaining healthy cells, which result in increased cellular pleomorphism accompanied by a decrease in the percentage of hexagonal cells with age [4,5,6,7]. Any injury that causes loss or damage to the corneal endothelial cells may threaten vision, including traumatism or intraocular surgery [8]. Moreover, the presence of diabetes in elderly patients undergoing cataract surgery makes them even more vulnerable to greater endothelial damage, a hypothesis that has been demonstrated by several studies [9,10,11,12,13]. They are five times more likely than non-diabetic to develop cataract at an early age. One of the mechanisms responsible for the pathogenesis of cataract among diabetics seems to be the polyol pathway, through which the enzyme aldose reductase (AR) catalyses the reduction of glucose into sorbitol [14]. Diabetic patients with poor glycemic control have been found to have pleomorphism and polymegethism, which can be translated into diabetic corneal endotheliopathy [15,16]. The avascular cornea is very sensitive to prolonged hyperglycaemia, and, therefore, many diabetic patients develop corneal complications, such as recurrent erosions, delayed epithelial wound healing, sensitivity loss, or tear film alterations [17,18]. Anterior capsular phimosis is more common in diabetic eye patients, and, therefore, a larger capsulorhexis should be performed, but at the same time it should be smaller than the optic diameter to avoid another possible postoperative complication, posterior capsular opacification [15,16]. As a consequence, the endothelium of diabetic patients is more susceptible to trauma associated with surgery, as they possess a smaller pupil. Due to these aspects, there is a lack of room for handling the intraocular instruments, and a potential damage to the cornea is more likely to occur. Hence, difficulties in dilation and maintaining it with mydriatics are often mainly due to autonomic neuropathy, which predominantly involves the sympathetic innervation of the iris dilator [10]. Phacoemulsification surgery with intraocular lens (IOL) implantation is a newer and appealing technique for surgeons because it allows a faster and safer surgical procedure which involves smaller incisions for lens nucleus removing. The visual recovery time is shortened and the intraoperative complications, such as expulsive haemorrhage or suture-induced astigmatism are fewer compared to conventional extracapsular cataract extraction [19,20]. Fragmentation and emulsification of the lens involves the use of high-intensity ultrasound energy during phacoemulsification, which can damage the corneal endothelium due to elevated localised temperature or a prolonged time of phacoemulsification [21]. Upon applying the phacoemulsification technique, two main heat sources are involved: [1] heat due to the conversion of electrical energy into mechanical energy and [2] heat due to friction when the phaco needle vibrates against the sleeve that contains the probe. The thermal damage to the collagen fibres may occur when the temperature reaches 60 °C, but continuous irrigation applied during surgery has a cooling effect by preventing the contracture of the incision site and surrounding tissue [22,23,24,25,26]. Specular microscopy is used for in vivo analyzation of the corneal endothelium, being an excellent non-contact and non-invasive tool used for assessing the health of the corneal endothelium before and after surgery [27]. This allows a direct investigation of the morphology, distribution, and density of the endothelial cells, supporting the surgeon’s decision either that it is safe or to not proceed with the surgery, as the minimal cell density required for maintaining corneal transparency and not developing irreversible corneal oedema or bullous keratopathy is 700 cells/mm2 [28,29]. The average value for cell density in a healthy adult is 2400 cells/mm2; if there is a coefficient of variation above 0.40 or a hexagonality percentage below 50 it might not tolerate intraocular surgery [27]. It has an extremely valuable diagnostic role for different corneal dystrophies, such as Fuchs endothelial dystrophy, iridocorneal endothelial syndrome, or posterior polymorphous dystrophy [30]. In this context, we present our results related to the effect of phacoemulsification cataract surgery on the morphology of corneal endothelium in diabetic versus non-diabetic patients, aiming to evaluate the influence of age, gender, axial length, and anterior chamber depth on the characteristic parameters of the corneal endothelium. ## 2. Materials and Methods This cross-sectional, retrospective study was conducted on 120 eyes from 120 patients in order to compare the corneal endothelium alterations in terms of corneal cell density (CD), central corneal thickness (CCT), and hexagonality (HEX) of the cells of 48 diabetic patients with a 10-year maximum history of the disease and good glycemic control (HbA1c < $7\%$) compared to 72 non-diabetic patients, before and after uneventful phacoemulsification surgery (1 and 4 weeks post-surgery) in the ophthalmology department of Emergency County Hospital Oradea, Bihor County, Romania. The study was conducted in tenets of the Declaration of Helsinki. Ethical approval was obtained from the Institutional Review Board of the hospital (no. $\frac{4238}{11.02.2021}$). ## 2.1. Pre-Operatory Ophthalmological Examination All patients underwent a rigorous ocular examination including visual acuity, slit-lamp examination, intraocular pressure using Goldman applanation tonometry, and a dilated fundus examination using a 90D lens. Corneal endothelial cell density (CD), central corneal thickness (CCT), and percentage of hexagonality (HEX) were measured both preoperatively and after surgery (at 1 and 4 weeks) using a noncontact specular microscope (Topcon Specular Microscope, Topcon Corporation Itabashi-ku, Tokyo, Japan 2017). A representative photograph recorded with the specular microscope showing the features of a non-diabetic cornea 1 week after surgery is displayed in Figure 1. The axial length (AXL) and the anterior chamber depth (ACD) were noted before surgery for each patient with immersion A-scan biometry, while the ultrasound energy consumption of the phaco machine and the effective phaco time were observed intraoperatively. The inclusion criteria of the study group consisted of patients with: [1] ages 50–90 years; [2] surgeries conducted by the same surgeon; and [3] phacoemulsification surgery technique, while the exclusion criteria were patients with: [1] pathological or traumatic cataracts; [2] pachymetry > 0.70 mm; and [3] corneal endothelial cells < 1200 cells/mm2 (Figure 2). ## 2.2. Surgical Technique All surgical procedures were performed by an experienced surgeon (author G.R.) using the phacoemulsification machine (Stellaris PC, Bausch & Lomb, Bloomfield, CT, USA, 2016) with the “divide and conquer“ technique shifting into a “stop and chop”. Preoperatively, all the eyes were dilated using tropicamide $1\%$ and phenylephrine $10\%$, followed by peribulbar anaesthesia. A main incision of 2.2 mm on the temporal side and two side-port incisions of 1.2 mm were applied at the limbus. A cohesive viscoelastic substance (Protectalon $2\%$) was injected into the anterior chamber for space maintenance, and a dispersive viscoelastic fluid was applied for corneal endothelium protection (Etacoat $2.4\%$ HPMC—Hydroxypropyl Methyl Cellulose), then a curvilinear capsulorhexis was performed. Using phacoemulsification, the nucleus lens was removed along with the residual cortex, while irrigation–aspiration was applied. Different foldable IOLs (intraocular lenses), such as hydrophobic monofocal IOLs with a biconvex aspheric optic that ensures that the lens is virtually aberration neutral (Hyflex) or trifocal hydrophobic lenses (Acriva Trinova) were implanted in the capsular bag. The excess of viscoelastic materials was washed out and the corneal incisions were hydro-sealed with a special patch. ## 2.3. Statistical Analysis Statistical analysis was carried out using GraphPad Prism (version 9.3.1). Numerical variables are presented as the mean ± standard deviation (SD). The correlations between different parametric variables before the operation were analyzed using Pearson correlation test. Repeated measure ANOVA following Tukey’s multiple comparisons test was used to compare the variables in the pre- and postoperative periods in one group. Unpaired t-test was used to compare the difference between diabetic and non-diabetic group. Correlation analysis was used to assess the univariate associations of the quantitative variables with endothelial cell loss. Multiple regression analysis was used to determine which variables independently contributed to the amount of endothelial cell loss. Significance was accepted at ($p \leq 0.05$). ## 3. Results A total of 120 eyes were included in this study, respectively, 48 from diabetic and 72 from non-diabetic patients; the socio-demographic characteristics are presented in Figure 3. The mean age of the study population in the diabetic group was 70.06 ± 11.08 while in the non-diabetic group it was 68.65 ± 9.71 years, ranging from 48–88 years. The ratio between men and women in the diabetic group was 1:1, while in the non-diabetic group a slightly higher percent of males was noted. Based on the records, in the non-diabetic group, $22.22\%$ of the patients presented with values of axial length (AXL) above 25 mm and $29.2\%$ under 22 mm, while in the diabetic group, $27.1\%$ of patients presented with AXL values over 25 mm and $20.8\%$ under 22 mm. The anterior chamber depth (ACD) was over 3 mm in about $60\%$ of both groups and under 3 mm in $30\%$ of the groups. Of the operated eyes, 67 were right and 53 were left eyes; the clinical features of each are also summarized in Figure 3. There was no statistically significant difference between the two groups in terms of age or gender. Intraoperatively, the mean effective phaco time in the diabetic group was 10.15 ± 6.18 s, while in the non-diabetic group it was 8.23 ± 5.64 s, with a p-value of 0.0814 and no statistical difference between groups (Figure 4). The correlation matrix between different variables before surgery reveals that in the diabetic group, there was a moderate negative correlation between age and AXL, ACD, and HEX, whereas in the non-diabetic group, age showed a moderate negative correlation with CD and HEX. The HEX also had a moderate positive correlation with ACD and AXL in the diabetic group and a moderate positive correlation with CD in non-diabetics. Also, a very strong positive correlation between AXL and ACD was noticed in both diabetic and non-diabetic groups (Figure 5). ## 3.1. Corneal Endothelial Cell Count The mean preoperative cell density (CD) was 2238 ± 369.6 in the diabetic group and 2537 ± 469 in the non-diabetic group. Comparing the postoperative cell loss in the non-diabetic group versus the diabetic group, a significantly higher endothelial cell loss was noted in the diabetic group after the first-week postoperative measurements (Table 1). More precisely, one week after the surgery the mean value of endothelial cell loss was 472.7 ± 369.1 in the diabetic group compared to 165.7 ± 214.6 in the non-diabetic group. It is also important to underscore the fact that diabetic patients lost more endothelial cells, progressively, from the 1st week post-surgical visit to the 4th week follow-up evaluation compared to their non-diabetic counterparts (Figure 6). ## 3.2. Central Corneal Thickness (CCT) The central corneal thickness increased in both groups after surgery, with a greater statistical significance in the first week post phacoemulsification (Figure 6). There was no statistical difference when the results of the non-diabetic and the diabetic group were compared by applying the unpaired t-test. ## 3.3. Percentage of Hexagonal Cells (HEX) The percentage of hexagonal cells dropped in both groups after surgery. The first week after surgery a statistically significant difference ($p \leq 0.05$) between non-diabetic and diabetic groups was noted, with a greater difference observed at week 4 ($p \leq 0.01$) (Figure 7). Corneal cells do not possess the ability to replicate, and hence, when their number decreases, a compensatory mechanism is initiated, causing enlargement of the remaining cells and loss of their hexagonal shape [28]. ## 3.4. Univariate Analysis By applying the univariate analysis, it could be observed that only preoperative cells density was significantly associated with endothelial cells loss in the diabetic group. On the contrary, the univariate analysis showed that no variable had any influence on the endothelial cell loss in the non-diabetic group (Table 2). ## 3.5. Multiple Regression Analysis The analysis by multiple regression model was applied to find the best predictors of endothelial cell loss. The final model explained $22\%$ of the variation in endothelial cell loss. The final model identified that being diabetic was an independent predictor of endothelial cell loss, while gender, age, ultrasound energy (U/S), axial length (AXL), anterior chamber depth (ACD) and effective phaco time were not independently associated with endothelial cell loss (Table 3). ## 4. Discussion In the present study, the influence of phacoemulsification cataract surgery on the morphology of corneal endothelium was evaluated in diabetic versus non-diabetic patients. Our results revealed a reduction in the number of endothelial cells, highlighting the fact that cataract surgery is a traumatic procedure for the cornea, regardless of the presence or absence of diabetes. We also noticed an increase in the central corneal thickness in both groups after surgery, with a greater statistical significance in the first week post phacoemulsification, while the percentage of hexagonal cells dropped in both groups after surgery, with a greater difference observed at week 4 ($p \leq 0.01$). High endothelial cell loss after phacoemulsification was described as being related to age, ocular comorbidities such as cornea guttata or other corneal dystrophies [31], ocular trauma, anterior chamber depth [32], eyes with occludable angles, or the inexperience of surgeons [33]. The patients with diabetes seem to be more prone to endothelium damage. In our study, after phacoemulsification, a decrease in the number of endothelial cells was noticed in each group individually, but comparatively, the cells loss in the diabetic group was statistically more significant than in the non-diabetic group (a mean endothelial cell loss of 472.7 ± 369.1 versus 165.7 ± 214.6). Similar to our study, the research conducted by Yang et al. [ 34] and Maadane et al. [ 35] reported significant corneal endothelial cell loss after cataract surgery, both in the diabetic group and the control group. Sharma et al. [ 36] also obtained similar results: the CD was lower in the diabetic patients than in the control group, respectively, 2550 ± 326 versus 2634 ± 256, but no difference was noted in the mean pachymetry or hexagonality. Similar results regarding the postoperative CD values were also found in Muhammad Khalid’s study [37]. They included 80 patients with type 2 diabetes and 80 patients without diabetes, with the results showing CD loss of $14.88\%$ (±10.79) in the diabetic group and $9.86\%$ (±10.20) in the control group ($$p \leq 0.003$$). The study conducted by Lee and Choi showed a mean CCT that was significantly higher in the diabetic group (588 ± 2.7 µm) and also a lower cell density compared to the control group (2577.2 ± 27.3 cell/mm2 vs. 2699.9 ± 38.7 cell/mm2) [38]. Amira and Shams [39] investigated 57 diabetic eyes and 45 non-diabetic eyes and found a lower endothelial cell density in the diabetic cornea group. A lower percentage of hexagonal cells ($33.24\%$ ± $10.25\%$) compared to that of the control group ($34.24\%$ ± $8.73\%$) was observed, but with no statistically significant difference ($$p \leq 0.603$$). Central corneal thickness was also noted to increase in the diabetic group after cataract surgery, with values of 545.61 ± 30.39 μm compared to 539.42 ± 29.22 μm in the control group, but it was not statistically significant ($$p \leq 0.301$$). A large study by Sudhir et al. on 1191 type 2 diabetic patients and 121 controls showed that in the diabetic group the endothelial cell count was lower when compared to the non-diabetic group, but no differences were found between the groups regarding the pachymetry values, hexagonality %, or coefficient of variation of cell size [36]. Leem et al. [ 40] reported that central corneal thickness was increased and endothelial cell density was decreased in patients with diabetes mellitus, and contact lens usage significantly affected corneal morphology in diabetic patients, although we did not include contact lenses users in our study. Xi Liu et al. evaluated the tear film of 25 diabetic patients and 20 non-diabetic patients after phacoemulsification cataract surgery. Alterations in the tear film and tear secretion were observed in both diabetic and non-diabetic patients after phacoemulsification, but with a greater reduction of the Schirmer I test in the diabetic group. The tear film break-up time values in both groups significantly decreased the first day after cataract surgery compared with the preoperative values, but when measured 180 days after phacoemulsification, the values returned to their preoperative ones in both groups [41]. Contrary, in a prospective study conducted by Storr-Paulsen et al. [ 6], the authors reported no significant differences between type 2 diabetic patients with good glycemic index and nondiabetic control subjects in terms of CD or hexagonality values [42]. Enlargement of endothelial cells, reduction of their hexagonality, and mobilisation of the remaining existing cells represent a mechanism of self-repair as a secondary effect to CD loss [6,40]. Many studies show that the corneal endothelial cells of diabetic patients have a decreased hexagonality and an increased coefficient of variation compared to those of non-diabetic patients [43,44,45], whereas others show no differences [42,46]. The greater sensitivity of the entire corneal endothelium of the diabetic patients may be explained by the accumulation of sugar alcohol in cells, which is converted by the excessive presence of glucose, as well by the fact that diabetic patients possess a reduced activity of Na/K-ATPase, which translates into structural instability and loss of the regular hexagonal pattern [47]. The endothelial cell loss was not significant enough in any of the patients included in our study, diabetic or non-diabetic, for postoperative complications to take place, such as corneal decompensation or bullous keratopathy. This is probably due to the use of viscoelastic substances that protect the corneal endothelium, careful use of ultrasound energy used for nucleus lens removal, a reduced effective phaco time, and also a good HbA1c ($7\%$) in the diabetic group patients. However, as a general approach, the risk of corneal decompensation should always be the main concern in diabetic patients as the severity of the disease increases; additionally, in patients with diabetic retinopathy and nephropathy, extreme caution should be exercised. There were a few limitations in our study. It was a cross-sectional, retrospective study, and hence, the measurements were made at a certain time point, and not all clinical variances could be precisely determined, while a prospective, controlled, and blinded design study would have allowed better quality interpretation of data. A single surgeon performed all surgeries, while with multiple surgeons we would probably have had a wider palette of postoperative results. Another limitation was the short-term follow-up of the study groups, although several studies showed that 4 weeks after uneventful cataract surgery is enough time for follow-ups, as the visual outcome at the end of 1 month is optimal and the postoperative complications usually occur in the first 2 weeks [48,49]. ## 5. Conclusions In our study, the influence of phacoemulsification cataract surgery on corneal cell density, central corneal thickness, and hexagonality was evaluated preoperatively and post-surgery (at 1 and 4 weeks) using specular microscopy. The effects of age, gender, axial length, and anterior chamber depth on the parameters of corneal endothelium of diabetic and non-diabetic patients were examined. We noticed significant differences between pre-surgical and postoperative CD values in both diabetic and non-diabetic patients. Despite good glycemic control, diabetic patients had more pronounced morphological abnormalities compared to those of non-diabetics, but visual outcomes after phacoemulsification with IOL implantation were similar in both groups. A drop in the postoperative endothelial density was noted after the first week, in both groups. A significant increase in central corneal thickness was also noted in both groups one week after phacoemulsification, but there was no statistical significance after 4 weeks in the diabetic group. In terms of cell hexagonality, statistically significant differences were noted after 4 weeks in both groups. A major finding in our study is that, although an advanced loss of CD was noted, along with an increased CCT and a reduction of hexagonality (especially in the diabetic group), there were no cases of postoperative bullous keratopathy, probably due to several factors, such as surgeon’s experience and the use of viscoelastic substances with a protective role, as well as a careful preoperative evaluation and a good glycemic index (HbA1c < $7\%$). We strongly recommend routine specular microscopy and HbA1c evaluation before all cataract surgeries. Regarding intraoperative precautions, a high level of monitoring is necessary in terms of pacho power intensity and ultrasound energy, along with a proper application of the dispersive viscoelastic substances to reduce the risk of endothelial damage for a successful surgical procedure. ## References 1. Burton M.J., Ramke J., Marques A.P., A Bourne R.R., Congdon N., Jones I., Tong B.A.M.A., Arunga S., Bachani D., Bascaran C.. The Lancet Global Health Commission on Global Eye Health: Vision beyond 2020. Lancet Glob. Health (2021) 9 e489-e551. DOI: 10.1016/S2214-109X(20)30488-5 2. Abrahamson I.A.. Cataract update. Am. Fam. Physician (1981) 24 111-119 3. Trevor-Roper P.D.. Cataracts. Br. Med. J. (1970) 3 33-35. DOI: 10.1136/bmj.3.5713.33 4. Busted N., Olsen T., Schmitz O.. Clinical observations on the corneal thickness and the corneal endothelium in diabetes mellitus. Br. J. Ophthalmol. (1981) 65 687-690. DOI: 10.1136/bjo.65.10.687 5. Joyce N.C.. Proliferative capacity of the corneal endothelium. Prog. Retin. Eye Res. (2003) 22 359-389. DOI: 10.1016/S1350-9462(02)00065-4 6. Bourne W.M.. Corneal Endothelium—Past, Present, and Future. Eye Contact Lens (2010) 36 310-314. DOI: 10.1097/ICL.0b013e3181ee1450 7. Abdellah M.M., Ammar H.G., Anbar M., Mostafa E.M., Farouk M.M., Sayed K., Alsmman A.H., Elghobaier M.G.. Corneal Endothelial Cell Density and Morphology in Healthy Egyptian Eyes. J. Ophthalmol. (2019) 2019 e637024. DOI: 10.1155/2019/6370241 8. Tsai M.-C., Daniels J.T.. The impact of biomechanics on corneal endothelium tissue engineering. Exp. Eye Res. (2021) 209 108690. DOI: 10.1016/j.exer.2021.108690 9. Beato J.N., Esteves-Leandro J., Reis D., Falcão M., Rosas V., Carneiro A., Reis F.F.. Corneal structure and endothelial morphological changes after uneventful phacoemulsification in type 2 diabetic and nondiabetic patients. Arq. Bras. Oftalmol. (2021) 84 454-461. DOI: 10.5935/0004-2749.20210071 10. Grzybowski A., Kanclerz P., Huerva V., Ascaso F.J., Tuuminen R.. Diabetes and Phacoemulsification Cataract Surgery: Difficulties, Risks and Potential Complications. J. Clin. Med. (2019) 8. DOI: 10.3390/jcm8050716 11. Zhai J., Su W., Tang Z., Lu L., Huang X., Wei L.. Phacoemulsification cataract surgery with different cumulative energy composite parameters in patients with type 2 diabetes mellitus:therapeutic effect and complications. J. South. Med. Univ. (2019) 39 500-504. DOI: 10.12122/j.issn.1673-4254.2019.04.19 12. Shakya K., Pokharel S., Karki K., Pradhananga C., Pokharel R., Malla O.. Corneal edema after phacoemulsification surgery in patients with type II diabetes mellitus. Nepal. J. Ophthalmol. (2013) 5 230-234. DOI: 10.3126/nepjoph.v5i2.8734 13. Chen Z., Song F., Sun L., Zhao C., Gao N., Liu P., Ge H.. Corneal integrity and thickness of central fovea after phacoemulsification surgery in diabetic and nondiabetic cataract patients. Arch. Med. Sci. (2018) 14 818-825. DOI: 10.5114/aoms.2016.64036 14. Kiziltoprak H., Tekin K., Inanc M., Goker Y.S.. Cataract in diabetes mellitus. World J. Diabetes (2019) 10 140-153. DOI: 10.4239/wjd.v10.i3.140 15. Kelkar A., Kelkar J., Mehta H., Amoaku W.. Cataract surgery in diabetes mellitus: A systematic review. Indian J. Ophthalmol. (2018) 66 1401-1410. DOI: 10.4103/ijo.IJO_1158_17 16. Taşlı N.G., Icel E., Karakurt Y., Ucak T., Ugurlu A., Yilmaz H., Akbas E.M.. The findings of corneal specular microscopy in patients with type-2 diabetes mellitus. BMC Ophthalmol. (2020) 20. DOI: 10.1186/s12886-020-01488-9 17. Yu F.-S.X., Lee P.S., Yang L., Gao N., Zhang Y., Ljubimov A.V., Yang E., Zhou Q., Xie L.. The impact of sensory neuropathy and inflammation on epithelial wound healing in diabetic corneas. Prog. Retin. Eye Res. (2022) 89 101039. DOI: 10.1016/j.preteyeres.2021.101039 18. Pont C., Ascaso F., Grzybowski A., Huerva V.. Corneal endothelial cell density during diabetes mellitus and ocular diabetes complications treatment. J. Fr. Ophtalmol. (2020) 43 794-798. DOI: 10.1016/j.jfo.2019.12.003 19. Asbell P.A., Dualan I., Mindel J., Brocks D., Ahmad M., Epstein S.. Age-related cataract. Lancet (2005) 365 599-609. DOI: 10.1016/S0140-6736(05)70803-5 20. Linebarger E.J., Hardten D.R., Shah G.K., Lindstrom R.L.. Phacoemulsification and Modern Cataract Surgery. Surv. Ophthalmol. (1999) 44 123-147. DOI: 10.1016/S0039-6257(99)00085-5 21. Takahashi H.. Corneal Endothelium and Phacoemulsification. Cornea (2016) 35 S3-S7. DOI: 10.1097/ICO.0000000000000990 22. Dewan T., Malik P., Tomar P.. Comparison of effective phacoemulsification time and corneal endothelial cell loss using three different ultrasound frequencies: A randomized controlled trial. Indian J. Ophthalmol. (2022) 70 1180-1185. DOI: 10.4103/ijo.IJO_2163_21 23. Holst A., Rolfsen W., Svensson B., Öllinger K., Lundgren B.. Formation of free radicals during phacoemulsification. Curr. Eye Res. (1993) 12 359-365. DOI: 10.3109/02713689308999460 24. Bradley M.J., Olson R.J.. A Survey about Phacoemulsification Incision Thermal Contraction Incidence and Causal Relationships. Am. J. Ophthalmol. (2006) 141 222-224. DOI: 10.1016/j.ajo.2005.08.018 25. Sorensen T., Chan C.C., Bradley M., Braga-Mele R., Olson R.J.. Ultrasound-induced corneal incision contracture survey in the United States and Canada. J. Cataract. Refract. Surg. (2012) 38 227-233. DOI: 10.1016/j.jcrs.2011.08.039 26. Klein B.E.K., Klein R., Wang Q., Moss S.E.. Older-onset diabetes and lens opacities. The Beaver Dam Eye Study. Ophthalmic Epidemiol. (1995) 2 49-55. DOI: 10.3109/09286589509071451 27. Martin R.. Cornea and anterior eye assessment with slit lamp biomicroscopy, specular microscopy, confocal microscopy, and ultrasound biomicroscopy. Indian J. Ophthalmol. (2018) 66 195-201. DOI: 10.4103/ijo.IJO_649_17 28. Chaurasia S., Vanathi M.. Specular microscopy in clinical practice. Indian J. Ophthalmol. (2021) 69 517-524. DOI: 10.4103/ijo.IJO_574_20 29. Inoue K., Tokuda Y., Inoue Y., Amano S., Oshika T., Inoue J.. Corneal Endothelial Cell Morphology in Patients Undergoing Cataract Surgery. Cornea (2002) 21 360-363. DOI: 10.1097/00003226-200205000-00006 30. Rio-Cristobal A., Martin R.. Corneal assessment technologies: Current status. Surv. Ophthalmol. (2014) 59 599-614. DOI: 10.1016/j.survophthal.2014.05.001 31. Giasson C.J., Solomon L.D., Polse K.A.. Morphometry of Corneal Endothelium in Patients with Corneal Guttata. Ophthalmology (2007) 114 1469-1475. DOI: 10.1016/j.ophtha.2006.11.022 32. El Ibiary H.A.N.Y., Badran T., Fawzy H., Mostafa A.. Endothelial Cell Loss after Phacoemulsification According to Different Anterior Chamber Depths in Hard Cataract. Med. J. Cairo Univ. (2022) 90 573-582. DOI: 10.21608/mjcu.2022.239612 33. O’Brien P.D., Fitzpatrick P., Kilmartin D.J., Beatty S.. Risk factors for endothelial cell loss after phacoemulsification surgery by a junior resident. J. Cataract. Refract. Surg. (2004) 30 839-843. DOI: 10.1016/S0886-3350(03)00648-5 34. Yang R., Sha X., Zeng M., Tan Y., Zheng Y., Fan F.. The influence of phacoemulsification on corneal endothelial cells at varying blood glucose levels. Eye Sci. (2011) 26 91-95. DOI: 10.3969/j.issn.1000-4432.2011.02.018 35. Maadane A., Boutahar H., Bourakba S., Sekhsoukh R.. Corneal endothelial evaluation of diabetic patients after phacoemulsification. J. Fr. Ophtalmol. (2019) 42 381-386. DOI: 10.1016/j.jfo.2018.09.012 36. Sudhir R.R., Raman R., Sharma T.. Changes in the Corneal Endothelial Cell Density and Morphology in Patients with Type 2 Diabetes Mellitus: A population-based study, Sankara Nethralaya Diabetic Retinopathy and Molecular Genetics Study (SN-DREAMS, Report 23). Cornea (2012) 31 1119-1122. DOI: 10.1097/ICO.0b013e31823f8e00 37. Khalid M., Hanif M.K., Islam Q.U., Mehboob M.A.. Change in corneal endothelial cell density after phacoemulsification in patients with type II diabetes mellitus. Pak. J. Med. Sci. (2019) 35 1366-1369. DOI: 10.12669/pjms.35.5.596 38. Lee J.S., Oum B.S., Choi H.Y., Lee J.E., Cho B.M.. Differences in corneal thickness and corneal endothelium related to duration in Diabetes. Eye (2006) 20 315-318. DOI: 10.1038/sj.eye.6701868 39. El-Agamy A., Alsubaie S.. Corneal endothelium and central corneal thickness changes in type 2 diabetes mellitus. Clin. Ophthalmol. (2017) 11 481-486. DOI: 10.2147/OPTH.S126217 40. Leem H.S., Lee K.J., Shin K.C.. Central Corneal Thickness and Corneal Endothelial Cell Changes Caused by Contact Lens Use in Diabetic Patients. Yonsei Med. J. (2011) 52 322-325. DOI: 10.3349/ymj.2011.52.2.322 41. Liu X., Gu Y.-S., Xu Y.-S.. Changes of tear film and tear secretion after phacoemulsification in diabetic patients. J. Zhejiang Univ. Sci. B (2008) 9 324-328. DOI: 10.1631/jzus.B0710359 42. Storr-Paulsen A., Singh A., Jeppesen H., Norregaard J.C., Thulesen J.. Corneal endothelial morphology and central thickness in patients with type II diabetes mellitus. Acta Ophthalmol. (2014) 92 158-160. DOI: 10.1111/aos.12064 43. Leelawongtawun W., Surakiatchanukul B., Kampitak K., Leelawongtawun R.. Study of Corneal Endothelial Cells Related to Duration in Diabetes. J. Med. Assoc. Thail. Chotmaihet Thangphaet. (2016) 99 S182-S188 44. Urban B., Raczyńska R., Bakunowicz-Łazarczyk A., Raczynska K., Krętowska M.. Evaluation of Corneal Endothelium in Children and Adolescents with Type 1 Diabetes Mellitus. Mediat. Inflamm. (2013) 2013 913754. DOI: 10.1155/2013/913754 45. Itoi M.M., Nakamura T.M., Mizobe K.M., Kodama Y.M., Nakagawa N.M., Itoi M.M.. Specular Microscopic Studies of the Corneal Endothelia of Japanese Diabetics. Cornea (1989) 8 2-6. DOI: 10.1097/00003226-198903000-00002 46. Islam Q.U., Mehboob M.A., Amin Z.A.. Comparison of corneal morphological characteristics between diabetic and non diabetic population. Pak. J. Med. Sci. (2017) 33 1307-1311. DOI: 10.12669/pjms.336.13628 47. Tang Y., Chen X., Zhang X., Tang Q., Liu S., Yao K.. Clinical evaluation of corneal changes after phacoemulsification in diabetic and non-diabetic cataract patients, a systematic review and meta-analysis. Sci. Rep. (2017) 7 14128. DOI: 10.1038/s41598-017-14656-7 48. Deshpande R., Vora U., Mangiraj V., Dole K., Deshpande M.. Can the postoperative follow-up visit be deferred up to four weeks after an uneventful cataract surgery?—A randomized controlled trial. Indian J. Ophthalmol. (2021) 69 1409-1413. DOI: 10.4103/ijo.IJO_2390_20 49. Saeed A., Guerin M., Khan I., Keane P., Stack J., Hayes P., Tormey P., Mullhern M., Beatty S.. Deferral of first review after uneventful phacoemulsification cataract surgery until 2 weeks: Randomized controlled study. J. Cataract. Refract. Surg. (2007) 33 1591-1596. DOI: 10.1016/j.jcrs.2007.05.022
--- title: Complete Freund’s Adjuvant Induces a Fibroblast-like Synoviocytes (FLS) Metabolic and Migratory Phenotype in Resident Fibroblasts of the Inoculated Footpad at the Earliest Stage of Adjuvant-Induced Arthritis authors: - Susana Aideé González-Chávez - Eduardo Chaparro-Barrera - María Fernanda Alvarado-Jáquez - Rubén Cuevas-Martínez - Rosa Elena Ochoa-Albíztegui - César Pacheco-Tena journal: Cells year: 2023 pmcid: PMC10047124 doi: 10.3390/cells12060842 license: CC BY 4.0 --- # Complete Freund’s Adjuvant Induces a Fibroblast-like Synoviocytes (FLS) Metabolic and Migratory Phenotype in Resident Fibroblasts of the Inoculated Footpad at the Earliest Stage of Adjuvant-Induced Arthritis ## Abstract The fibroblast-like synoviocytes (FLS) have a crucial role in the pathogenesis of Rheumatoid Arthritis (RA); however, its precise mechanisms remain partially unknown. The involvement of the fibroblast in activating adjuvant-induced arthritis (AA) has not been previously reported. The objective was to describe the participation of footpads’ fibroblasts in the critical initial process that drives the AA onset. Wistar rats were injected with Complete Freund’s Adjuvant (CFA) or saline solution in the hind paws’ footpads and euthanized at 24 or 48 h for genetic and histological analyses. Microarrays revealed the differentially expressed genes between the groups. The CFA dysregulated RA-linked biological processes at both times. Genes of MAPK, Jak-STAT, HIF, PI3K-Akt, TLR, TNF, and NF-κB signaling pathways were altered 24 h before the arrival of immune cells (CD4, CD8, and CD68). Key markers TNF-α, IL-1β, IL-6, NFκB, MEK-1, JAK3, Enolase, and VEGF were immunodetected in fibroblast in CFA-injected footpads at 24 h but not in the control group. Moreover, fibroblasts in the CFA inoculation site overexpressed cadherin-11, which is linked to the migration and invasion ability of RA-FLS. Our study shows that CFA induced a pathological phenotype in the fibroblast of the inoculation site at very early AA stages from 24 h, suggesting a prominent role in arthritis activation processes. ## 1. Introduction Fibroblast-like synoviocytes (FLS) are relevant players in the pathogenic process of Rheumatoid Arthritis (RA) [1,2,3]; they mediate the altered dynamics of the rheumatoid synovium and perpetuate RA. RA-FLS phenotype requires the transformation of an otherwise quiescent structural cell into an aggressive and proliferative cell, which interacts with- and activates immune cells [4,5,6,7,8]. Currently, FLS is considered a potential therapeutic target for RA patients [9,10,11]. RA-FLS exhibit a metabolic profile parallel to some tumors, thriving in a hypoxic environment with increased production of oxygen radicals and cytokines [12,13,14]. Under stressful conditions, FLS has the potential to initiate, coordinate, and perpetuate inflammatory processes through complex interactions with adaptive immune response cells [5,15,16,17,18]. Several surface markers of FLS and its pathogenic signaling pathways have been characterized [19,20,21]. Nevertheless, the origin of RA-FLS and the mechanisms driving the transformation of their precursors remain largely undefined. Very likely, FLS derive either from mesenchymal stem cells (MSC) or synovial resident fibroblasts. Furthermore, the fact that FLS and other mesenchymal-derived cells can migrate and disseminate in the bloodstream raises the possibility that either FLS or activated precursors can be induced elsewhere and settle on joints, expanding the inflammatory process [22,23,24]. Adjuvant-induced arthritis (AA) rat model resembles human RA. Complete Freund’s Adjuvant (CFA) injections in the footpads are enough to induce joint swelling, synovial lymphoid infiltrate, and joint destruction [25,26]. AA is a standard testing field for novel treatments for RA [27,28,29,30], confirming the similarity of their pathogenesis. However, despite understanding the downstream consequences of the CFA injection, the initial mechanisms to induce arthritis onset remain unknown, especially at the injection site at the earliest events. Since it is a controlled scenario, AA can allow the detection of essential mediators at the earliest stages that lead to arthritis and which are probably undetectable once arthritis sets on. Therefore, the present study aimed to describe the metabolic and immune profile triggered by the CFA in the subcutaneous injection site, within the resident cells, including fibroblast in the AA model’s earliest stages. The transcriptome modifications were evaluated by microarray technology and its subsequent bioinformatic analysis. Additionally, the role of local fibroblasts in the injected site was confirmed by the immunodetection of metabolic and immune markers. ## 2.1. Animal Model The study included 24 8-week-old male Wistar rats that were initially randomly divided into 2 groups of 12 rats each: the CFA group that received the injection of 0.3 mL of CFA (1 mg Mycobacterium tuberculosis (H 37RA, ATCC 25177), heat-killed and dried, 0.85 mL paraffin oil, and 0.15 mL mannide monooleate; Sigma Aldrich Chemic, cat. No. F5881) in the hind paws’ footpads [31]; and the control group that received the same volume of saline solution (SS). Rats were kept under controlled luminosity conditions (12 h light/12 h dark) and temperature (23 ± 2 °C) and received food and water ad libitum. Six rats from each group were euthanized with isoflurane 24 h after the CFA or SS injection, and the other 6 from each group 48 h after the injections. Finally, 4 study groups were formed: CFA-24 h, CFA-48 h, SS-24 h, and SS-48 h. Footpads were dissected for RNA extraction and histological analysis (1 of the hind paws of each rat for each experimental strategy). The entire protocol, including manipulation of the animals, complied with the institutional ethics committee and Institutional Animal Care and Use Committee (IACUC), ID numbers: UACH, CEI-B/$\frac{329}{15}$, UACH, FM-FM-EXT-B-$\frac{339}{17.}$ ## 2.2. DNA Microarray and Bioinformatics Analysis After the euthanasia, injected footpads (6 per study group) were pulverized on liquid nitrogen using a mortar and pestle. Total RNA was purified using the RNeasy® Lipid Tissue Mini Kit (Qiagen, Hilden, Germany) extraction kit, following the manufacturer’s protocol. RNA concentration and integrity were verified on a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). For the microarray assay, the RNA of each study group was mixed in equimolar quantities to form the pools used in the DNA microarrays. The DNA microarrays were performed Institute of Cellular Physiology, Autonomous University of Mexico (UNAM), Mexico. The expression profiles of CFA (experiment) vs. SS (reference) at 24 and 48 h were evaluated. The microarrays included the reverse transcription-polymerase chain reaction (RT-PCR), and the resulting cDNAs were labeled with Cy5 (CFA group) or Cy3 (SS groups). Hybridization was done in the Rn5K (UNAM, Mexico) chip containing 5000 rat genes. The signal was scanned and acquired using the ScanArray 4000 (Packard BioChips Technologies, Billerica, MA, USA) and analyzed in the GenArise Microarray Analysis Tool software (UNAM, Mexico). The lists of differentially expressed genes (DEGs) [Z-score ≥ 1.5 standard deviations (SD)] in the CIA-injected respect SS-injected footpads were obtained. The list of DEGs was further analyzed in DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/ accessed on 2 March 2023), an open-resource platform that classifies genes list into functional biological processes and KEGG (Kyoto Encyclopedia of Genes and Genomes) signaling pathways [32]. Furthermore, DEGs were also analyzed on the STRING 11.5 database (https://string-db.org/ accessed on 2 March 2023) to obtain the analysis and integration of direct and indirect protein-protein interactions (IPP) centered on the functional association [33]. The DEGs identified in the microarray were loaded, and the interactions with minimal confidence (interaction score > 0.4) were selected. Finally, the obtained IPP network was analyzed more thoroughly to obtain primary clusters of sub-networks using the Cytoscape software (version 3.9.1) with the Molecular Complex Detection (MCODE) complement (node score cutoff = 0.4) [34,35]. The main clusters were analyzed in STRING to obtain the IPP networks and their associated KEGG signaling pathways. Those pathways and genes of relevance in arthritis were identified. ## 2.3. Histological Analysis The injected hind paws (six per study group) were fixed in $10\%$ formaldehyde, decalcified with $5\%$ nitric acid for 24 h, dehydrated in ethanol, and embedded in paraffin [36]. Sections of 3 μm were obtained and stained with hematoxylin and eosin (H&E) for histological evaluation. The images were acquired with a digital camera coupled to the optical microscope (AxioStar plus, Carl Zeiss, Berlin, Germany). The histological analysis was performed in the subcutaneous injection site and not the synovial structures because we wanted to explore the effect of CFA within the injected tissues. Immunohistochemistry was done with specific antibodies against rat cell surface receptors as well as intracellular signaling molecules such as CD4 [11-0042-82], CD8 [14-0081-82], CD68 (14-0689-82; eBioscienceTM, Invitrogen, Waltham, MA, USA), tumor necrosis factor (TNF)-α (sc-52746), Interleukin (IL)-1β (sc-32294), IL-6 (sc-130326), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB; sc-8414), Toll-like receptor (TLR)-4 (sc-10741), mitogen-activated protein kinase kinase (MEK)-1 (sc-6250), Janus kinase (Jak)-3 (sc-513), enolase (Eno; sc-100812), and vascular endothelial growth factor (VEGF; sc-7269; Santa Cruz Biotechnology, Dallas, TX, USA). Tissue sections were deparaffinized in xylene and dehydrated in descending ethanol until water. Antigen retrieval was done using $0.05\%$ trypsin (T1426-250 mg, SIGMA Life Science, St. Louis, MO, USA) for 30 min at 37 °C or 1 mM EDTA pH 8.0 for 30 min at 95 °C. The slides were treated with $0.2\%$ Triton-X100 (Bio-Rad, Hercules, CA, USA). After blocking with $10\%$ bovine serum albumin (BSA; A9647-100G SIGMA Life Science, St. Louis, MO, USA) for 1 h at room temperature in a humidified chamber, the tissues were incubated with the primary antibody in a 1:200 dilution at 4 °C overnight. The corresponding isotype’s biotin-streptavidin-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) were used in a 1:400 dilution. Immunodetection was carried out using the Pierce® streptavidin horseradish peroxidase-conjugated (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and Diaminobenzidine (DAB; D4293-50SET, SIGMA-ALDRICH, USA) as the chromogen. The primary antibody was replaced with PBS buffer to establish negative controls. Images were acquired using a digital camera (AmScope MU1803, Irvine, CA, USA) and an optical microscope (AxioStar Plus, Carl Zeiss, Berlin, Germany). The expression of CD4, CD8, and CD68 at 24 h and 48 h was quantified with the ImageJ program and the IHC toolbox. The DAB color was extracted from each image, and the maximum and mean gray values were obtained. Each image’s optical density (OD) was obtained with log10 (maximum gray value/mean gray value). The OD’s means and SD were calculated and graphed per study group. The double indirect-immunofluorescence (IIF) was performed to co-localize the TNFα, IL-1β, IL-6, NFκB, TLR-4, MEK-1, JAK3, Eno, VEGF, and cadherin-11 (CDH11) markers in the fibroblasts of the CFA-injected hind paws. The immunofluorescence was performed sequentially. First, the tissues were deparaffinized, and antigen retrieval was performed with $0.05\%$ trypsin for 30 min at 37 °C or 1 mM EDTA pH 8.0 for 30 min at 95 °C. Subsequently, the tissues were permeabilized with $0.2\%$ Triton-X100 and blocked with $5\%$ normal donkey serum. Next, sections were incubated with the first primary antibody (1:200 dilution): CD4 [11-0042-82], CD8 [14-0081-82], CD68 (14-0689-82; eBioscienceTM, Invitrogen, USA), TNFα (sc-52746), IL-1β (sc-32294), IL-6 (sc-130326), NF-κB (sc-8414), TLR-4 (sc-10741), MEK-1 (sc-6250) Jak-3 (sc-513), Eno (sc-100812), VEGF (sc-7269), and CDH11 (sc-365867)(Santa Cruz Biotechnology, USA). After washing in PBS, tissues were incubated with the AF488-labeled secondary antibody (1:200 dilution): Donkey Anti-Mouse IgG [715-545-150] or Donkey Anti-Rabbit IgG (711-545-152; Jackson ImmunoResearch, USA). A second blocking was performed, and then tissues were incubated with the second primary antibody (1:200 dilution): Fibroblast Marker, ER-TR7 [against Collagen type VI (ColVI) [37,38] (sc-73355, Santa Cruz Biotechnology), washed in PBS, and incubated with the Cy5-labeled secondary antibody (1:200 dilution): Donkey Anti-Rat IgG (712-175-153, Jackson ImmunoResearch, USA). Labeling was evaluated by epifluorescence microscopy (Zeiss Axio Imager A1), and images were acquired using a digital camera (AmScope MU1203-FL, USA). ## 2.4. Statistical Analysis The bioinformatics analysis of the microarray data included their statistical analysis. In DAVID, Fisher’s exact test measures gene enrichment in annotation terms. Fisher’s Exact p-values are computed by summing probabilities p over defined sets of tables (Prob = ∑Ap) [32]. In the STRING database, the PPI enrichment p-value indicates that the nodes are not random and that the observed number of edges is significant; for the associated-KEGG pathways, the false discovery rate (FDR) is defined as FDR = E(V/R\|R > 0) P(R > 0) [39]. In Cytoscape-MCODE, the complex score is defined as the product of the complex subgraph, C = (V, E), density, and the number of vertices in the complex subgraph (DC × \|V\|) [35]. For OD measures in IHC, statistical analysis was made in SPSS statistics v22 software (SPSS Science Inc., Chicago, IL, USA). The Shapiro-Wilk and Kolmogorov-Smirnov tests were used to determine the data normality. In addition, means and SD were estimated and graphed, and a Student’s t-test was used to compare the expression of CD4, CD8, and CD68 at 24 h and 48 h. Differences were considered significant when p ≤ 0.05. ## 3.1. DNA Microarray and Bioinformatic Analysis The microarray resulted in 663 DEGs (162 up/501 down) at 24 h after CFA injection and 689 DEGs (413 up/276 down) at 48 h. Furthermore, the bioinformatic analysis in the DAVID platform showed that these genes were significantly associated with biological processes (Figure 1A and Figure 2A) and KEGG pathways (Figure 1B and Figure 2B) related to RA, such as the response to hypoxia. The apoptosis regulation processes were associated with the highest number of genes at both times. In the combined analysis with Cytoscape-MCODE at 24 h, the comparison between CFA and SS resulted in two relevant clusters. The first, which included 20 nodes, 75 edges, and a score of 7.89 (Figure 1C), mainly showed activation of intracellular signaling pathways, including neuroactive ligand-receptor interaction and calcium signaling pathways. Carbohydrate metabolism signaling pathways, including glycolysis, insulin secretion, and carbon metabolism, were also found in this cluster. The most relevant genes in this cluster were cholinergic receptor muscarinic 3 (Chrm3) and glutamate metabotropic receptor 2 (Grm2). The second cluster included 50 nodes, 163 edges, and a score of 6.65 (Figure 1D), which confirmed the relevance of metabolic pathways and showed several immune-relevant pathways mainly related to innate immunity, including TNF, NOD-like receptor, Chemokine, and Cytokine-cytokine receptor interaction signaling pathways. MAPK signaling pathway was also found dysregulated. The most prominent proteins in the network included Il1b, Mapk9, and several chemokines. The Jak3 and Eno3 were also relevant proteins in the network. At 48 h, cluster 1, which included 25 nodes, 132 edges, and a score of 11.00 (Figure 2C), showed that both MAPK and PI3K-Akt signaling pathways were active. In addition, other pathways linked to human RA were present in the network, including HIF-1, TNF, VEGF, TLR, NOD-like receptor, Jak-STAT, and Wnt signaling pathways. The most interactive proteins in the network included several Mapk [1,13,7], Nfkb1, Jun, protein phosphatase three catalytic (Ppp3c) subunit alpha and beta, and protein kinase c (Prkc) beta and gamma. The second cluster included 33 nodes, 166 edges, and a score of 10.37 (Figure 2D), and the most significant signaling pathway was the neuroactive ligand-receptor interaction with 18 associated genes. This network also highlights the multiple interactions of fibroblast growth factors (Fgf) 3, 4, and 21 with the Rap1, Ras, MAPK, and PI3K-Akt signaling pathways. The genes of clusters 1 and 2, both at 24 h and 48 h, were selected and analyzed in the STRING platform to construct the IPP network and highlight the related KEGG pathways at the earliest stages of the AA model. Figure 3 shows several pathways linked to RA, including HIF-1, TNF, MAPK, Chemokine, T-cell receptor, Jak-STAT, and Glycolysis/Gluconeogenesis, were associated with these genes. In addition, Mapk proteins (1 and 13) linked several protein circuits, and the Il1b was also a link between MAPK, TNF, TLR, and Cytokine-cytokine receptor interaction signaling pathways. ## 3.2. Histological Analysis CFA-injected footpads showed an inflammatory infiltrate 24 h post-injection, while those injected with SS had no infiltrate (Figure 4). In the tissues injected with CFA, the expression of CD4, CD8, and CD68 at 24 h was scarce and limited to the cells surrounding the oil drops, whereas, at 48 h, all of them were markedly increased (Figure 4). According to the analysis of the OD of the IHC images, the expression of these three markers was significantly higher at 48 h post-CFA injection. Some relevant markers in the RA signaling pathways highlighted in the bioinformatic analysis were present in the CFA-injected footpads at 24 h post-injection. The positive immunodetections of TNFα, IL-1β, IL-6, NF-κB, TLR-4, MEK-1, Jak-3, Eno, and VEGF are shown in Figure 5. The staining patterns were different for each protein. Its expression was not exclusive to the inflammatory cells, showing clear patterns of expression in the skin, adipose tissue, and muscle. The double IIF stains confirmed that, except for the TLR4 marker, all those proteins were expressed in the fibroblast of CFA-injected footpads’ (Figure 5). Moreover, the expression of CDH11 and ColVI, molecules associated with the pathogenic phenotype of FLS, were overexpressed in the cells from the CFA-injection site at 24 h (Figure 6). In the case of CDH11, an increased number of cells expressing it was noted compared to the SS-injected group. At the same time, the expression of ColVI only was confirmed in CFA-injected footpads. ## 4. Discussion The present article describes the effect of CFA within the resident cells surrounding the injection in the subcutaneous footpad at the earliest time. Our results show that the CFA induces the activation of quiescent subcutaneous fibroblasts, which express a hypoxic metabolic profile and can produce key pathogenic inflammatory mediators for both AA and RA. This immunometabolic profile recreates the FLS phenotype [9,21,40] and suggests its critical role in the processes that explain AA onset. Although the involvement of FLS and its pathogenic phenotype in RA within the synovium has been widely studied, the potential participation and transformation of fibroblasts residing in distant tissues as precursors of these FLS remain undefined. Our findings confirm that in AA, from 24 h, the fibroblasts at the CFA-inoculation site expressed genes and proteins explicitly linked to RA. Moreover, we found that the footpad’s fibroblast overexpressed migration and invasion molecules related to RA-FLS. Our results show that just 24 h after the injection, CFA dysregulated key signaling pathways linked to RA pathogenesis, including MAPK [41], Jak-STAT [42], HIF [43], PI3K-Akt [44], TLR [45], TNF [46], and NF-kB [47]. Likewise, our bioinformatic analysis revealed the most relevant genes in these pathways, including Mapk 1, 9, and 13, IL-1b, -2 and -6, Pik3r2, and NfκB. These findings are consistent with those described by Stolina M. et al. [ 48], which assessed the presence of biomarkers in 14 stages of the AA model, the earliest the day -5 concerning the onset of clinical arthritis. On the other hand, Yu H. et al. [ 49] reported that the most dramatic changes in gene expression were observed in the preclinical phase of the disease (day 7 post-injection). Exploring earlier times than previously reported, our study allows us to conclude that the processes that explain arthritis are established much earlier than was known. The transcriptomic profiles at 24 and 48 h presented some differences worth noting. At 24 h, the transcriptional modifications were primarily associated with metabolic rather than inflammatory signaling pathways. Although at 24 h, the differential expression of some inflammatory cytokines such as Il1b was observed, the bioinformatic analysis indicated that, at this time, the associations with the high number of genes were of metabolic processes. In contrast, the transcriptome analyses at 48 h resulted in the most substantial identification of inflammatory processes and pathways. Indeed, the expression profile at 48 h was highly similar to what we previously observed in the joints, weeks after CFA injection, in other of our previous works carried out in this model [50]. Moreover, from 24 h, the neuroactive ligand-receptor interaction signaling pathway was deregulated by the CFA. Nine genes with an FDR of 1.9 × 10−10 were associated with this pathway at 24 h; notably, at 48 h, this pathway doubled the number of genes, and the FDR value decreased to 4 × 10−24. The neuroactive ligand-receptor interaction signaling pathway is directly related to neuro function. Neuroactive ligands influence neuronal function by binding to intracellular receptors, which can bind transcription factors and regulate gene expressions [51]. Recent studies have demonstrated that RA progression is closely related to abnormal function of the central nervous system. The nervous system (NS) can receive stimulation from immune cells. At the same time, the signals of the central NS act on immune cells through the peripheral NS to regulate the inflammatory response. In RA, cytokines such as IL-6, IL-1β, IL-17, and TNF can interact with joint nociceptors and activate and sensitize them. IL-1β may also be involved in RA-induced pain and hypersensitivity [52]. The above could suggest that the AA activation that begins at the inoculation site in the footpad is closely linked to neuronal processes from the early times of the disease. At 24 h post-injection, our IHC analysis showed that CD4 and CD8 positive cells were virtually absent, and CD68 cells were scarce and limited to the periphery of the oil drops in CFA-injected footpads. Since our main objective was to describe the activation profile of fibroblasts at the CFA-injected footpads, we analyzed the expression of RA-metabolic and RA-inflammation markers at 24 h to isolate them as much as possible from the immune cells. The markers selection for IHC analysis resulted mainly from the microarray analyses. We choose markers highlighted at 24 and 48 h, such as IL-1β, MEK-1 (MAPK activator), JAK3, NF-κB, and Eno3. Other markers were selected because they are critical in RA-associated pathways, such as TNFα for the TNF signaling pathway, TLR4 for the TLR signaling pathway, and VEGF for the VEGF signaling pathway. Compared with saline solution-injected footpads, at 24 h in the CFA-injected, the TNFα, IL-1β, IL-6, NF-κB, MEK-1, JAK-3, Eno, and VEGF proteins were overexpressed. The IHC showed that different types of cells expressed these markers; therefore, we performed the colocalization staining using a fibroblast marker. Our IIF confirmed that the footpad’s resident fibroblast expressed the markers of interest, suggesting their phenotype transformation resembling descriptions of FLS. In RA, FLS exhibits a significant phenotype transformation; studying the metabolic and regulatory changes that drive this transformation is a promising field for understanding its etiopathogenesis. The FLS activation by hypoxia, platelet-derived growth factor, TNF, and other inflammatory mediators increases glucose metabolism and transforms the FLS from quiescent to aggressive and metabolically active cells. Glucose metabolism is increased in activated FLS, and glycolytic inhibition reduces FLS aggressive phenotype in vitro and decreases bone and cartilage damage in several murine models of arthritis [53,54,55]. In our study, at 24 h and 48 h, the CFA-injected footpads had dysregulated the HIF-1, TNF, and glycolysis signaling pathways. Essential genes, including Hk1, Gck, Aldoc, and Eno3, were increased by CFA injection; Eno3 expression was also confirmed in the fibroblast of footpads by IIF. FLS also produces MMP-3, VEGF, and IL-6, which contribute to the worsening of arthritic conditions through the recruitment and activation of inflammatory cells and angiogenesis [12]. Moreover, angiogenesis is also linked to hypoxia and oxidative stress in RA [43,56]. Here, we also demonstrated that angiogenesis’ transcriptome was altered in the CFA-injected footpads; moreover, we confirmed that resident fibroblast expressed VEGF and IL-6. Increasing evidence has demonstrated the role of mitochondrial alterations in RA mainly due to the interplay between metabolism and innate immunity and the modulation of the inflammatory response of FLS. Mitochondrial dysfunction derived from several danger signals could activate tricarboxylic acid (TCA) disruption, creating a vicious cycle of oxidative-mitochondrial stress [57]. Our microarray analyses showed that several mitochondrial metabolism genes, including Ak3, Idh3a, Idh3B, Idh3g, Asl, and Acyl, were deregulated by the CFA, suggesting the alteration of TCA. In RA-FLS, secreted Frizzle-related protein-1 (SFPR1) regulates pyroptosis through WNT/beta-catenin and Notch signaling pathways [20]. Consistent with this description, the WNT pathway was dysregulated in CFA-injected footpads. In addition, other pathways widely recognized as altered in FLS, including Jak-STAT [58], PI3K-Akt, MAPK, and Toll-like receptors [21], were dysregulated by the CFA in our study. Furthermore, critical markers of these pathways, including MEK-1, JAK-3, and TLR4, are overexpressed by rat footpad fibroblasts after adjuvant injection. Regarding the role of specific proinflammatory cytokines in the FLS pathogenic phenotype, it is known that IL-1β induces the proliferation of FLS through the mediation of the NF-κB signaling pathway [59]. The dysregulation of proinflammatory pathways was also found in our transcriptomic analyses. We found that Il1b and Nfkb genes were overexpressed in the footpads at 24 h and 48 h post-CFA injection. Moreover, we confirmed that at 24 h, the resident fibroblasts expressed both proteins. Although arthritis is the hallmark of RA, it is a systemic disease, and the inflammation involves other organs and structures; therefore, the specific origin of the illness is a matter of great interest. Indeed, the possibility that the initial inflammatory process could start away from the joints is a valid research hypothesis; and the mucosal structures such as the gut [60,61], the lungs [62,63,64,65], or the oral cavity [66,67,68] are candidate triggering sites. If that is the case, a prerequisite is the migration of this initial priming to the susceptible joints. The ability of MSC, fibrocytes, and fibroblasts to migrate and induce the pathogenic phenotype of FLS has gained evidence. In experimental arthritis, joint inflammation is preceded by infiltration of MSC, which contributes to synovial membrane hyperplasia. On the other hand, fibroblast migration in human disease has only recently been reported [22]. Fibrocytes were the first cells with fibroblastic properties described to migrate from the circulation into the joint [23]. Even activated fibrocytes are considered FLS precursors [24]. Some authors also supposed that FLS could migrate from one joint to another and potentially spread the disease; however, this idea remains unproven [69]. Recently, the presence of circulating fibroblast-like cells in the blood of patients with RA weeks before the disease flare-up was demonstrated. These cells are identified as PRe-Inflammatory MEsenchymal (PRIME) cells and share some markers with FLS, such as CDH11 [70]. CDH11 is mainly expressed in MSC and is essential for tissue migration and organization during embryogenesis [71]. In joints, CDH11 is primarily expressed in FLS and cooperates with inflammatory factors to promote the migration, invasion, and degradation of joint tissue in RA [72,73,74]. CDH11 was previously linked to ColVI in human MSC differentiation towards the adipogenic lineage [75] and adipose tissue fibroblasts [76]. Human MSCs lacking CDH11 had decreased the expression of ColVI and increased the expression of fibronectin through the TGFβ1 pathway [75]. Furthermore, in adipose tissue fibroblasts, CDH11 deficiency reduced their production of ColIII and ColVI, resulting in substantially less adipose tissue fibrosis in obesity [76]. ColVI is a structural and signaling protein that may act as an early sensor of the injury/repair response and regulate fibrogenesis by modulating cell-cell interactions and stimulating the proliferation of MSC. Moreover, ColVI regulates pericellular matrix properties, chondrocyte swelling, and mechanotransduction in articular cartilage [77,78]. In the RA synovium, ColVI is extensively deposited in the interstitial connective tissue and along the synovial membrane lining [79]. Remarkably, this interaction between CDH11 and ColVI (the antigen recognized by the ER-TR7 antibody [37,38] was observed in our study at CFA injection sites from 24 h, suggesting that footpad cells acquired a phenotype previously described for FLS. Considering that CDH11 has been associated with the fibroblast’s ability to migrate and invade the joint, our finding supports the possibility that these footpad’s fibroblasts initiate the pathogenic mechanisms. Interestingly, although CDH11 expression was markedly higher in cells from CFA-injected footpads, it was also detected in cells from the control group. Conversely, ColVI expression was only detected in CFA-injected foot pads and not controls. This finding suggests that CFA induces a fibrotic phenotype. Furthermore, fibrosis is associated with ColVI overexpression in a model of lung sepsis in rats within the first 24 h of LPS administration [80]. The above also allows us to suppose that the mycobacterial components of CFA could induce the fibrotic phenotype in our study. ## 5. Conclusions The earliest events in AA include the dysregulation of several key pathogenic signaling pathways in the footpad residing tissular cells, including fibroblasts. These pathways are linked to established pathogenic pathways that explain joint inflammation and destruction. Aside from an evident metabolic dysregulation, the CFA induces a severe challenge to the tissue environment, which drives a significant adaptation in the cells’ behavior, with a widespread activation of protective mechanisms, including those that result in inflammation, migration, and fibrosis. The fact that this transformation occurs in tissular fibroblast and that endorses it to acquire a phenotype shared with FLS becomes a matter of interest because it opens the potential to study its dynamics. It also reinforces its potential role as a critical player, under the logic of the Danger model, as a potential driving engine for the abnormal immune response resulting in arthritis. ## References 1. Marsh L., Kemble S., Reis Nisa P., Singh R., Croft A.P.. Fibroblast Pathology in Inflammatory Joint Disease. Immunol. Rev. (2021) 302 163-183. DOI: 10.1111/imr.12986 2. Chu C.-Q.. Highlights of Strategies Targeting Fibroblasts for Novel Therapies for Rheumatoid Arthritis. Front. Med. (2022) 9 846300. DOI: 10.3389/fmed.2022.846300 3. Fearon U., Hanlon M.M., Floudas A., Veale D.J.. Cellular Metabolic Adaptations in Rheumatoid Arthritis and Their Therapeutic Implications. Nat. Rev. Rheumatol. (2022) 18 398-414. DOI: 10.1038/s41584-022-00771-x 4. Bartok B., Firestein G.S.. Fibroblast-like Synoviocytes: Key Effector Cells in Rheumatoid Arthritis. Immunol. Rev. (2010) 233 233-255. DOI: 10.1111/j.0105-2896.2009.00859.x 5. Croft A.P., Campos J., Jansen K., Turner J.D., Marshall J., Attar M., Savary L., Wehmeyer C., Naylor A.J., Kemble S.. Distinct Fibroblast Subsets Drive Inflammation and Damage in Arthritis. Nature (2019) 570 246-251. DOI: 10.1038/s41586-019-1263-7 6. Pap T., Dankbar B., Wehmeyer C., Korb-Pap A., Sherwood J.. Synovial Fibroblasts and Articular Tissue Remodelling: Role and Mechanisms. Semin. Cell Dev. Biol. (2020) 101 140-145. DOI: 10.1016/j.semcdb.2019.12.006 7. Lefèvre S., Knedla A., Tennie C., Kampmann A., Wunrau C., Dinser R., Korb A., Schnäker E.-M., Tarner I.H., Robbins P.D.. Synovial Fibroblasts Spread Rheumatoid Arthritis to Unaffected Joints. Nat. Med. (2009) 15 1414-1420. DOI: 10.1038/nm.2050 8. Németh T., Nagy G., Pap T.. Synovial Fibroblasts as Potential Drug Targets in Rheumatoid Arthritis, Where Do We Stand and Where Shall We Go?. Ann. Rheum. Dis. (2022) 81 1055-1064. DOI: 10.1136/annrheumdis-2021-222021 9. Nygaard G., Firestein G.S.. Restoring Synovial Homeostasis in Rheumatoid Arthritis by Targeting Fibroblast-like Synoviocytes. Nat. Rev. Rheumatol. (2020) 16 316-333. DOI: 10.1038/s41584-020-0413-5 10. Sandhu G., Thelma B.K.. New Druggable Targets for Rheumatoid Arthritis Based on Insights from Synovial Biology. Front. Immunol. (2022) 13 834247. DOI: 10.3389/fimmu.2022.834247 11. Filer A.. The Fibroblast as a Therapeutic Target in Rheumatoid Arthritis. Curr. Opin. Pharmacol. (2013) 13 413-419. DOI: 10.1016/j.coph.2013.02.006 12. Bottini N., Firestein G.S.. Duality of Fibroblast-like Synoviocytes in RA: Passive Responders and Imprinted Aggressors. Nat. Rev. Rheumatol. (2013) 9 24-33. DOI: 10.1038/nrrheum.2012.190 13. Yang W., Wei X., Jiao Y., Bai Y., Sam W.N., Yan Q., Sun X., Li G., Ma J., Wei W.. STAT3/HIF-1α/Fascin-1 Axis Promotes RA FLSs Migration and Invasion Ability under Hypoxia. Mol. Immunol. (2022) 142 83-94. DOI: 10.1016/j.molimm.2021.12.004 14. Hu Z., Chen J., Wang M., Weng W., Chen Y., Pan Y.. Rheumatoid Arthritis Fibroblast-like Synoviocytes Maintain Tumor-like Biological Characteristics through CiRS-7-Dependent Regulation of MiR-7. Mol. Biol. Rep. (2022) 49 8473-8483. DOI: 10.1007/s11033-022-07666-w 15. Pacheco-Tena C., González-Chávez S.A.. The Danger Model Approach to the Pathogenesis of the Rheumatic Diseases. J. Immunol. Res. (2015) 2015 506089. DOI: 10.1155/2015/506089 16. Tu J., Huang W., Zhang W., Mei J., Zhu C.. Two Main Cellular Components in Rheumatoid Arthritis: Communication Between T Cells and Fibroblast-Like Synoviocytes in the Joint Synovium. Front. Immunol. (2022) 13 922111. DOI: 10.3389/fimmu.2022.922111 17. Qin Y., Cai M.-L., Jin H.-Z., Huang W., Zhu C., Bozec A., Huang J., Chen Z.. Age-Associated B Cells Contribute to the Pathogenesis of Rheumatoid Arthritis by Inducing Activation of Fibroblast-like Synoviocytes via TNF-α-Mediated ERK1/2 and JAK-STAT1 Pathways. Ann. Rheum. Dis. (2022) 81 1504-1514. DOI: 10.1136/ard-2022-222605 18. Tang Y., Wang B., Sun X., Li H., Ouyang X., Wei J., Dai B., Zhang Y., Li X.. Rheumatoid Arthritis Fibroblast-like Synoviocytes Co-Cultured with PBMC Increased Peripheral CD4+ CXCR5+ ICOS+ T Cell Numbers. Clin. Exp. Immunol. (2017) 190 384-393. DOI: 10.1111/cei.13025 19. Wu Z., Ma D., Yang H., Gao J., Zhang G., Xu K., Zhang L.. Fibroblast-like Synoviocytes in Rheumatoid Arthritis: Surface Markers and Phenotypes. Int. Immunopharmacol. (2021) 93 107392. DOI: 10.1016/j.intimp.2021.107392 20. Jiang P., Wei K., Chang C., Zhao J., Zhang R., Xu L., Jin Y., Xu L., Shi Y., Guo S.. SFRP1 Negatively Modulates Pyroptosis of Fibroblast-Like Synoviocytes in Rheumatoid Arthritis: A Review. Front. Immunol. (2022) 13 903475. DOI: 10.3389/fimmu.2022.903475 21. Bustamante M.F., Garcia-Carbonell R., Whisenant K.D., Guma M.. Fibroblast-like Synoviocyte Metabolism in the Pathogenesis of Rheumatoid Arthritis. Arthritis Res. Ther. (2017) 19 110. DOI: 10.1186/s13075-017-1303-3 22. Mahmoud D.E., Kaabachi W., Sassi N., Tarhouni L., Rekik S., Jemmali S., Sehli H., Kallel-Sellami M., Cheour E., Laadhar L.. The Synovial Fluid Fibroblast-like Synoviocyte: A Long-Neglected Piece in the Puzzle of Rheumatoid Arthritis Pathogenesis. Front. Immunol. (2022) 13 942417. DOI: 10.3389/fimmu.2022.942417 23. Herzog E.L., Bucala R.. Fibrocytes in Health and Disease. Exp. Hematol. (2010) 38 548-556. DOI: 10.1016/j.exphem.2010.03.004 24. Galligan C.L., Fish E.N.. Circulating Fibrocytes Contribute to the Pathogenesis of Collagen Antibody-Induced Arthritis. Arthritis Rheum. (2012) 64 3583-3593. DOI: 10.1002/art.34589 25. Choudhary N., Bhatt L.K., Prabhavalkar K.S.. Experimental Animal Models for Rheumatoid Arthritis. Immunopharmacol. Immunotoxicol. (2018) 40 193-200. DOI: 10.1080/08923973.2018.1434793 26. Fischer B.D., Adeyemo A., O’Leary M.E., Bottaro A.. Animal Models of Rheumatoid Pain: Experimental Systems and Insights. Arthritis Res. Ther. (2017) 19 146. DOI: 10.1186/s13075-017-1361-6 27. Bordy R., Verhoeven F., Tournier-Nappey M., Wendling D., Demougeot C., Totoson P.. Methotrexate Did Not Improve Endothelial Function in Rheumatoid Arthritis: A Study in Rats with Adjuvant-Induced Arthritis. Clin. Exp. Rheumatol. (2019) 37 81-88 28. Abdel-Maged A.E.-S., Gad A.M., Abdel-Aziz A.K., Aboulwafa M.M., Azab S.S.. Comparative Study of Anti-VEGF Ranibizumab and Interleukin-6 Receptor Antagonist Tocilizumab in Adjuvant-Induced Arthritis. Toxicol. Appl. Pharmacol. (2018) 356 65-75. DOI: 10.1016/j.taap.2018.07.014 29. Wang Z.-Z., Liu F., Gong Y.-F., Huang T.-Y., Zhang X.-M., Huang X.-Y.. Antiarthritic Effects of Sorafenib in Rats with Adjuvant-Induced Arthritis: Antiarthritic effects of sorafenib. Anat. Rec. (2018) 301 1519-1526. DOI: 10.1002/ar.23856 30. Vidal B., Cascão R., Finnilä M.A.J., Lopes I.P., da Glória V.G., Saarakkala S., Zioupos P., Canhão H., Fonseca J.E.. Effects of Tofacitinib in Early Arthritis-Induced Bone Loss in an Adjuvant-Induced Arthritis Rat Model. Rheumatology (2018) 57 1461-1471. DOI: 10.1093/rheumatology/kex258 31. Stevenson C.S., Marshall L.A., Morgan D.W.. In Vivo Models of Inflammation (2006) 32. Huang D.W., Sherman B.T., Tan Q., Kir J., Liu D., Bryant D., Guo Y., Stephens R., Baseler M.W., Lane H.C.. DAVID Bioinformatics Resources: Expanded Annotation Database and Novel Algorithms to Better Extract Biology from Large Gene Lists. Nucleic Acids Res. (2007) 35 W169-W175. DOI: 10.1093/nar/gkm415 33. Szklarczyk D., Morris J.H., Cook H., Kuhn M., Wyder S., Simonovic M., Santos A., Doncheva N.T., Roth A., Bork P.. The STRING Database in 2017: Quality-Controlled Protein–Protein Association Networks, Made Broadly Accessible. Nucleic Acids Res. (2017) 45 D362-D368. DOI: 10.1093/nar/gkw937 34. Isserlin R., Merico D., Voisin V., Bader G.D.. Enrichment Map—A Cytoscape App to Visualize and Explore OMICs Pathway Enrichment Results. F1000Research (2014) 3 141. DOI: 10.12688/f1000research.4536.1 35. Bader G.D., Hogue C.W.V.. An Automated Method for Finding Molecular Complexes in Large Protein Interaction Networks. BMC Bioinform. (2003) 4. DOI: 10.1186/1471-2105-4-2 36. González-Chávez S.A., Pacheco-Tena C., Macías-Vázquez C.E., Luévano-Flores E.. Assessment of Different Decalcifying Protocols on Osteopontin and Osteocalcin Immunostaining in Whole Bone Specimens of Arthritis Rat Model by Confocal Immunofluorescence. Int. J. Clin. Exp. Pathol. (2013) 6 1972-1983. PMID: 24133575 37. Schiavinato A., Przyklenk M., Kobbe B., Paulsson M., Wagener R.. Collagen Type VI Is the Antigen Recognized by the ER-TR7 Antibody. Eur. J. Immunol. (2021) 51 2345-2347. DOI: 10.1002/eji.202149263 38. Malhotra D., Fletcher A.L., Astarita J., Lukacs-Kornek V., Tayalia P., Gonzalez S.F., Elpek K.G., Chang S.K., Knoblich K., Hemler M.E.. Transcriptional Profiling of Stroma from Inflamed and Resting Lymph Nodes Defines Immunological Hallmarks. Nat. Immunol. (2012) 13 499-510. DOI: 10.1038/ni.2262 39. Szklarczyk D., Gable A.L., Nastou K.C., Lyon D., Kirsch R., Pyysalo S., Doncheva N.T., Legeay M., Fang T., Bork P.. The STRING Database in 2021: Customizable Protein–Protein Networks, and Functional Characterization of User-Uploaded Gene/Measurement Sets. Nucleic Acids Res. (2021) 49 D605-D612. DOI: 10.1093/nar/gkaa1074 40. Falconer J., Murphy A.N., Young S.P., Clark A.R., Tiziani S., Guma M., Buckley C.D.. Review: Synovial Cell Metabolism and Chronic Inflammation in Rheumatoid Arthritis. Arthritis Rheumatol. (2018) 70 984-999. DOI: 10.1002/art.40504 41. Paunovic V., Harnett M.M.. Mitogen-Activated Protein Kinases as Therapeutic Targets for Rheumatoid Arthritis. Drugs (2013) 73 101-115. DOI: 10.1007/s40265-013-0014-6 42. Hu L., Liu R., Zhang L.. Advance in Bone Destruction Participated by JAK/STAT in Rheumatoid Arthritis and Therapeutic Effect of JAK/STAT Inhibitors. Int. Immunopharmacol. (2022) 111 109095. DOI: 10.1016/j.intimp.2022.109095 43. Quiñonez-Flores C.M., González-Chávez S.A., Pacheco-Tena C.. Hypoxia and Its Implications in Rheumatoid Arthritis. J. Biomed. Sci. (2016) 23 62. DOI: 10.1186/s12929-016-0281-0 44. Liu S., Ma H., Zhang H., Deng C., Xin P.. Recent Advances on Signaling Pathways and Their Inhibitors in Rheumatoid Arthritis. Clin. Immunol. Orlando Fla (2021) 230 108793. DOI: 10.1016/j.clim.2021.108793 45. Liu J., Wang X., Wang S., Liu F.. Therapeutic Potential of Non-Coding RNAs and TLR Signalling Pathways in Rheumatoid Arthritis. Curr. Pharm. Biotechnol. (2021) 22 1490-1500. DOI: 10.2174/1389201021666201001142829 46. Yao Z., Getting S.J., Locke I.C.. Regulation of TNF-Induced Osteoclast Differentiation. Cells (2021) 11. DOI: 10.3390/cells11010132 47. Li J., Tang R.-S., Shi Z., Li J.-Q.. Nuclear Factor-ΚB in Rheumatoid Arthritis. Int. J. Rheum. Dis. (2020) 23 1627-1635. DOI: 10.1111/1756-185X.13958 48. Stolina M., Bolon B., Middleton S., Dwyer D., Brown H., Duryea D., Zhu L., Rohner A., Pretorius J., Kostenuik P.. The Evolving Systemic and Local Biomarker Milieu at Different Stages of Disease Progression in Rat Adjuvant-Induced Arthritis. J. Clin. Immunol. (2009) 29 158-174. DOI: 10.1007/s10875-008-9238-8 49. Yu H., Lu C., Tan M.T., Moudgil K.D.. The Gene Expression Profile of Preclinical Autoimmune Arthritis and Its Modulation by a Tolerogenic Disease-Protective Antigenic Challenge. Arthritis Res. Ther. (2011) 13 R143. DOI: 10.1186/ar3457 50. González-Chávez S.A., Quiñonez-Flores C.M., Espino-Solís G.P., Vázquez-Contreras J.Á., Pacheco-Tena C.. Exercise Exacerbates the Transcriptional Profile of Hypoxia, Oxidative Stress and Inflammation in Rats with Adjuvant-Induced Arthritis. Cells (2019) 8. DOI: 10.3390/cells8121493 51. Wei J., Liu J., Liang S., Sun M., Duan J.. Low-Dose Exposure of Silica Nanoparticles Induces Neurotoxicity via Neuroactive Ligand–Receptor Interaction Signaling Pathway in Zebrafish Embryos. Int. J. Nanomed. (2020) 15 4407-4415. DOI: 10.2147/IJN.S254480 52. Gao D., Gao X., Yang F., Wang Q.. Neuroimmune Crosstalk in Rheumatoid Arthritis. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23158158 53. Garcia-Carbonell R., Divakaruni A.S., Lodi A., Vicente-Suarez I., Saha A., Cheroutre H., Boss G.R., Tiziani S., Murphy A.N., Guma M.. Critical Role of Glucose Metabolism in Rheumatoid Arthritis Fibroblast-like Synoviocytes: Fls glycolytic metabolism in rheumatoid arthritis. Arthritis Rheumatol. (2016) 68 1614-1626. DOI: 10.1002/art.39608 54. de Oliveira P.G., Farinon M., Sanchez-Lopez E., Miyamoto S., Guma M.. Fibroblast-Like Synoviocytes Glucose Metabolism as a Therapeutic Target in Rheumatoid Arthritis. Front. Immunol. (2019) 10 1743. DOI: 10.3389/fimmu.2019.01743 55. Masoumi M., Mehrabzadeh M., Mahmoudzehi S., Mousavi M.J., Jamalzehi S., Sahebkar A., Karami J.. Role of Glucose Metabolism in Aggressive Phenotype of Fibroblast-like Synoviocytes: Latest Evidence and Therapeutic Approaches in Rheumatoid Arthritis. Int. Immunopharmacol. (2020) 89 107064. DOI: 10.1016/j.intimp.2020.107064 56. Quiñonez-Flores C.M., González-Chávez S.A., Del Río Nájera D., Pacheco-Tena C.. Oxidative Stress Relevance in the Pathogenesis of the Rheumatoid Arthritis: A Systematic Review. BioMed Res. Int. (2016) 2016 6097417. DOI: 10.1155/2016/6097417 57. López-Armada M.J., Fernández-Rodríguez J.A., Blanco F.J.. Mitochondrial Dysfunction and Oxidative Stress in Rheumatoid Arthritis. Antioxidants (2022) 11. DOI: 10.3390/antiox11061151 58. Hammaker D., Nygaard G., Kuhs A., Ai R., Boyle D.L., Wang W., Firestein G.S.. Joint Location–Specific. ACR Open Rheumatol. (2019) 1 640-648. DOI: 10.1002/acr2.11093 59. Xing X.-W., Shi H.-Y., Liu S., Feng S.-X., Feng S.-Q., Gong B.-Q.. MiR-496/MMP10 Is Involved in the Proliferation of IL-1β-Induced Fibroblast-Like Synoviocytes Via Mediating the NF-ΚB Signaling Pathway. Inflammation (2021) 44 1359-1369. DOI: 10.1007/s10753-021-01421-2 60. Longman R.S., Littman D.R.. The Functional Impact of the Intestinal Microbiome on Mucosal Immunity and Systemic Autoimmunity. Curr. Opin. Rheumatol. (2015) 27 381-387. DOI: 10.1097/BOR.0000000000000190 61. Scher J.U., Sczesnak A., Longman R.S., Segata N., Ubeda C., Bielski C., Rostron T., Cerundolo V., Pamer E.G., Abramson S.B.. Expansion of Intestinal Prevotella Copri Correlates with Enhanced Susceptibility to Arthritis. eLife (2013) 2 e01202. DOI: 10.7554/eLife.01202 62. Quirke A.-M., Perry E., Cartwright A., Kelly C., De Soyza A., Eggleton P., Hutchinson D., Venables P.J.. Bronchiectasis Is a Model for Chronic Bacterial Infection Inducing Autoimmunity in Rheumatoid Arthritis. Arthritis Rheumatol. (2015) 67 2335-2342. DOI: 10.1002/art.39226 63. Reynisdottir G., Karimi R., Joshua V., Olsen H., Hensvold A.H., Harju A., Engström M., Grunewald J., Nyren S., Eklund A.. Structural Changes and Antibody Enrichment in the Lungs Are Early Features of Anti-Citrullinated Protein Antibody-Positive Rheumatoid Arthritis. Arthritis Rheumatol. Hoboken NJ (2014) 66 31-39. DOI: 10.1002/art.38201 64. Demoruelle M.K., Weisman M.H., Simonian P.L., Lynch D.A., Sachs P.B., Pedraza I.F., Harrington A.R., Kolfenbach J.R., Striebich C.C., Pham Q.N.. Brief Report: Airways Abnormalities and Rheumatoid Arthritis-Related Autoantibodies in Subjects without Arthritis: Early Injury or Initiating Site of Autoimmunity?. Arthritis Rheum. (2012) 64 1756-1761. DOI: 10.1002/art.34344 65. Klareskog L., Stolt P., Lundberg K., Källberg H., Bengtsson C., Grunewald J., Rönnelid J., Harris H.E., Ulfgren A.-K., Rantapää-Dahlqvist S.. A New Model for an Etiology of Rheumatoid Arthritis: Smoking May Trigger HLA-DR (Shared Epitope)-Restricted Immune Reactions to Autoantigens Modified by Citrullination. Arthritis Rheum. (2006) 54 38-46. DOI: 10.1002/art.21575 66. Kaneko C., Kobayashi T., Ito S., Sugita N., Murasawa A., Ishikawa H., Tabeta K.. Association among Periodontitis Severity, Anti-Agalactosyl Immunoglobulin G Titer, and the Disease Activity of Rheumatoid Arthritis. J. Periodontal Res. (2021) 56 702-709. DOI: 10.1111/jre.12867 67. Li Y., Guo R., Oduro P.K., Sun T., Chen H., Yi Y., Zeng W., Wang Q., Leng L., Yang L.. The Relationship Between Porphyromonas Gingivalis and Rheumatoid Arthritis: A Meta-Analysis. Front. Cell. Infect. Microbiol. (2022) 12 956417. DOI: 10.3389/fcimb.2022.956417 68. Pischon N., Pischon T., Kröger J., Gülmez E., Kleber B.-M., Bernimoulin J.-P., Landau H., Brinkmann P.-G., Schlattmann P., Zernicke J.. Association among Rheumatoid Arthritis, Oral Hygiene, and Periodontitis. J. Periodontol. (2008) 79 979-986. DOI: 10.1902/jop.2008.070501 69. Køster D., Egedal J.H., Lomholt S., Hvid M., Jakobsen M.R., Müller-Ladner U., Eibel H., Deleuran B., Kragstrup T.W., Neumann E.. Phenotypic and Functional Characterization of Synovial Fluid-Derived Fibroblast-like Synoviocytes in Rheumatoid Arthritis. Sci. Rep. (2021) 11 22168. DOI: 10.1038/s41598-021-01692-7 70. Orange D.E., Yao V., Sawicka K., Fak J., Frank M.O., Parveen S., Blachere N.E., Hale C., Zhang F., Raychaudhuri S.. RNA Identification of PRIME Cells Predicting Rheumatoid Arthritis Flares. N. Engl. J. Med. (2020) 383 218-228. DOI: 10.1056/NEJMoa2004114 71. Sfikakis P.P., Vlachogiannis N.I., Christopoulos P.F.. Cadherin-11 as a Therapeutic Target in Chronic, Inflammatory Rheumatic Diseases. Clin. Immunol. (2017) 176 107-113. DOI: 10.1016/j.clim.2017.01.008 72. Chen X., Xiang H., Yu S., Lu Y., Wu T.. Research Progress in the Role and Mechanism of Cadherin-11 in Different Diseases. J. Cancer (2021) 12 1190-1199. DOI: 10.7150/jca.52720 73. Cao C., Wu F., Niu X., Hu X., Cheng J., Zhang Y., Li C., Duan X., Fu X., Zhang J.. Cadherin-11 Cooperates with Inflammatory Factors to Promote the Migration and Invasion of Fibroblast-like Synoviocytes in Pigmented Villonodular Synovitis. Theranostics (2020) 10 10573-10588. DOI: 10.7150/thno.48666 74. Beckmann D., Römer-Hillmann A., Krause A., Hansen U., Wehmeyer C., Intemann J., de Gorter D.J.J., Dankbar B., Hillen J., Heitzmann M.. Lasp1 Regulates Adherens Junction Dynamics and Fibroblast Transformation in Destructive Arthritis. Nat. Commun. (2021) 12 3624. DOI: 10.1038/s41467-021-23706-8 75. Passanha F.R., Geuens T., LaPointe V.L.S.. Cadherin-11 Influences Differentiation in Human Mesenchymal Stem Cells by Regulating the Extracellular Matrix Via the TGFβ1 Pathway. Stem Cells (2022) 40 669-677. DOI: 10.1093/stmcls/sxac026 76. Chang S.K., Kohlgruber A.C., Mizoguchi F., Michelet X., Wolf B.J., Wei K., Lee P.Y., Lynch L., Duquette D., Ceperuelo-Mallafré V.. Stromal Cell Cadherin-11 Regulates Adipose Tissue Inflammation and Diabetes. J. Clin. Investig. (2017) 127 3300-3312. DOI: 10.1172/JCI86881 77. Sun S., Karsdal M.A.. Type VI Collagen. Biochemistry of Collagens, Laminins and Elastin (2016) 49-55 78. Gilbert S.J., Bonnet C.S., Blain E.J.. Mechanical Cues: Bidirectional Reciprocity in the Extracellular Matrix Drives Mechano-Signalling in Articular Cartilage. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222413595 79. Wolf J., Carsons S.E.. Distribution of Type VI Collagen Expression in Synovial Tissue and Cultured Synoviocytes: Relation to Fibronectin Expression. Ann. Rheum. Dis. (1991) 50 493-496. DOI: 10.1136/ard.50.7.493 80. Okawa S., Unuma K., Yamada A., Aki T., Uemura K.. Lipopolysaccharide Induces Expression of Collagen VI in the Rat Lung. J. Toxicol. Pathol. (2015) 28 37-41. DOI: 10.1293/tox.2014-0053
--- title: The Effect of a Diet Enriched with Jerusalem artichoke, Inulin, and Fluoxetine on Cognitive Functions, Neurogenesis, and the Composition of the Intestinal Microbiota in Mice authors: - Aleksandra Szewczyk - Marta Andres-Mach - Mirosław Zagaja - Agnieszka Kaczmarczyk-Ziemba - Maciej Maj - Joanna Szala-Rycaj journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10047150 doi: 10.3390/cimb45030168 license: CC BY 4.0 --- # The Effect of a Diet Enriched with Jerusalem artichoke, Inulin, and Fluoxetine on Cognitive Functions, Neurogenesis, and the Composition of the Intestinal Microbiota in Mice ## Abstract The aim of the study was to assess the effect of long-term administration of natural prebiotics: Jerusalem artichoke (topinambur, TPB) and inulin (INU) as well as one of the most popular antidepressants, fluoxetine (FLU), on the proliferation of neural stem cells, learning and memory functions, and the composition of the intestinal microbiota in mice. Cognitive functions were assessed using the Morris Water Maze (MWM)Test. Cells were counted using a confocal microscope and ImageJ software. We performed 16S rRNA sequencing to assess changes in the gut microbiome of the mice. The obtained results showed that the 10-week supplementation with TPB (250 mg/kg) and INU (66 mg/kg) stimulates the growth of probiotic bacteria, does not affect the learning and memory process, and does not disturb the proliferation of neural stem cells in the tested animals. Based on this data, we can assume that both TPB and INU seem to be safe for the proper course of neurogenesis. However, 2-week administration of FLU confirmed an inhibitory impact on Lactobacillus growth and negatively affected behavioral function and neurogenesis in healthy animals. The above studies suggest that the natural prebiotics TPB and INU, as natural supplements, may have the potential to enrich the diversity of intestinal microbiota, which may be beneficial for the BGM axis, cognitive functions, and neurogenesis. ## 1. Introduction According to the most recent scientific estimates, there are about 3.8 × 1013 bacteria in the adult human body, which is 1.3 times more than human cells [1]. The totality of microorganisms in a specific environment is called the microbiota, and the collective genomes of all microorganisms are called the microbiome [2]. The human microbiome includes not only bacteria but also other microorganisms such as fungi, archaea, viruses, and protozoa [3]. The composition of the intestinal microbiota is an outcome of an interplay between numerous variables, such as a diet, an environment, host genetics, an exposure to infections, and an antibiotic usage [4]. Most of the human intestinal bacteria belong to four phyla: Firmicutes, Bacteroidetes, Actinobacteria, and Proteobacteria, with the Firmicutes and Bacteroidetes predominating. The equilibrium of the gut’s diverse microbial population is essential for preserving the host’s state of health [5]. In general, gut microbiome is beneficial for the human host. These microorganisms are key players in regulating gut metabolism, and are necessary to understand metabolism dysfunctions. For instance, there are connections between gut bacteria and the synthesis of vitamins B and K, short-chain fatty acid (SCFA) production, pathogen growth inhibition, preservation of intestinal barrier integrity and mucosal immune homeostasis, and involvement in the xenobiotic metabolism system [6,7]. The gut, the gut microbiota, and the brain are involved in a constant two-way communication that is referred to as the brain–gut–microbiome (BGM) axis [8,9]. Research on the BGM axis primarily includes the studies of pre- or probiotics as well as antibiotics in animal models [10,11,12,13], studies in germ-free (GF) animals [14], and fecal transplants [15]. This allows the identification of pathways that regulate BGM signaling between the digestive tract and the brain, including neural, endocrine, and immune pathways. It has been shown that the BGM axis plays an important role in the formation and maintenance of cognitive functions [16,17,18,19]. Moreover, the BGM axis was proved to be significant in the regulation of neurogenesis, one of the most essential processes of proliferation, migration, and differentiation of new neural cells [20,21,22,23]. Undoubtedly, one of the key factors negatively affecting the BGM and the neurogenesis is stress [24,25,26,27]. BGM is crucial in maintaining a proper homeostasis, and several psychiatric and nonpsychiatric illnesses have been proved to be at least in part responsible its dysfunction [28,29,30]. It should also be noted that, in addition to diseases, the composition and diversity of the gut microbiota can be affected by drugs used to treat the disorder, especially antibiotics. However, recent studies have shown that non-antibiotic drugs, such as the antipsychotics and antidepressants, also affect the intestinal microbiota [31,32]. Few studieshave been done so far to determine how the antidepressants affect the microbiota in the gut. One of them concerns research on fluoxetine hydrochloride (FLU), an antidepressant drug belonging to the group of selective serotonin reuptake inhibitors (SSRIs) used in depression, panic attacks, anxiety, or obsessive–compulsive symptoms [33,34]. The in vitro studies carried out so far have provided information that FLU has antimicrobial activity [35], whereas in vivo studies have shown its negative effect on the composition of the intestinal microbiome in both rats [35] and mice [36]. For years, we utilized probiotic bacteria in addition to food to maintain a healthy microbiome or to rebalance the system [37]. Probiotics are live bacteria and yeasts that have a beneficial effect onhuman health. The impact of probiotics may also be favored by prebiotics, which can be used as an alternative to probiotics or as their additional support [38]. Prebiotics are non-digestible food components that specifically promote the development of probiotic bacteria in the gut, including lactobacilli and bifidobacteria [39]. However, various prebiotics will promote the growth of different native gut bacteria [38]. The presence of prebiotics in the diet hasmany beneficial effects on the gut, the immune system, and brain function, particularly, brain-derived neurotrophic factor (BDNF) expression and N-methyl-D-aspartate (NMDA) receptor signaling. Substances classified as prebiotics are oligosaccharides, including galactooligosaccharides(GOS), transgalactooligosaccharides(TOS), xylooligosaccharides(XOS), fructooligosaccharides(FOS), isomaltooligosaccharides(IMO), as well as polysaccharides such as inulin (INU), cellulose, pectin, hemicellulose, or reflux starch [38]. INU is composed of fructose residues connected by β-[2,1] glycosidic linkages, and it plays the role of spare material in plants. INU is abundantly present in a variety of root vegetables including topinambur(TPB), banana, chicory, leek, and onion. TPB, a tuberous perennial plant of the Asteraceae family commonly known as wild sunflower or Jerusalem artichoke, has 160–200 g of INU per kilogram of fresh weight [40,41]. According to the study done thus far, TPB has several health-promoting qualities, including lowering blood glucose, triglycerides, LDL cholesterol, and total cholesterol [42]. In addition, TPB supplementary diet has an additive impact on the probiotic bacteria in the gut of rats [43] and mice [41]. Taking into account the above data, the aim of this study was to determine the impact of the long-term supplementation with prebiotics, TPB, and INU, in the form of natural compounds and 2-week administration of the antidepressant drug FLU, on the development of probiotic bacteria necessary for the proper functioning of the BGM axis, and thus their influence on cognitive functions and neurogenesis in healthy mice. ## 2.1. Animals and Experimental Conditions All tests were carried out on 6-week-old male C57BL/6J mice (20–25 g). Animals were housed under standard laboratory conditions (natural light–dark cycle, 55 ± $5\%$ humidity, and a temperature of 21 ± 1 °C) and allowed food (a complete feed for mice and rats; AGROPOL, Marynin, Poland) and water ad libitum. After a 7-day acclimatization period the animals were randomly assigned to four experimental groups (TPB, INU, FLU, and control) consisting of seven mice each. All experimental procedures were approved by the Local Ethics Committee at the University of Life Science in Lublin (No $\frac{73}{2020}$). ## 2.2. Drugs The following substances and drugs were used in the presented study: TPB (Organic, Sieniawa, Poland), fluoxetine hydrochloride (FLU; Sigma Aldrich, St. Louis, MO, USA), INU isolated from chicory roots (FORMEDS, Poznań, Poland), 5-bromo-2′-deoxyuridine BrdU(Sigma Aldrich, St. Louis, MO, USA), medetomidine hydrochloride (Tocris Bioscience, Bristol, UK), isoflurane (Baxter, Warszawa, Poland), methylscopolamine(Sigma Aldrich, St. Louis, MO, USA). All substances were suspended in water for injections (Baxter, Poland). TPB, INU, and FLU administrated via gastric gavages, whereas BrdU and medetomidine hydrochloride were administrated via intraperitoneal (i.p.) injections. All substances were administrated with 1 mL syringes in a volume of 10 mL/kg. ## 2.3. Drugs Administration Animals were divided into fourgroups (sevenmice per group, $$n = 7$$):TPBINUFLUcontrol group (water for injection) The supplementation lasted for 10 weeks (Figure 1). The animals were given a freshly prepared suspension of powdered TPB, INU, and FLU via an oral administration. According to the manufacturer’s recommendation, daily dose forhuman intake of the TPB is 10–15 g per day and INU is 4 g per day. The substances were calculated into the mouse body weight, dissolved in water for injections, and administered once a day (assumed average adult weight of 60 kg, the target dose per kg of body weight was 250 mg-TBP and 66 mg-INU). FLU (12 mg/kg) was administered to the FLU group in the last two weeks of the experiments. The dose of FLU was selected based on the latest literature data [44]. Control animals were given water for injection orally, throughout the duration of the diet. To measure changes in body weight following the 10-week diet, all animals were weighed at the 1st, 3rd, 5th, 7th, and 10th weeks. Additionally, on the 9th week of the diet, the animals were given an i.p. injection of BrdU(a cell proliferation marker) once daily at a concentration of 50 mg/kg for 5 days. Fecal samples were taken to analyze the microbiome after the diet, and the animals underwent the behavioral Morris Water Maze Test (MWM), which measures spatial learning and memory abilities. After the behavioral studies, animals were perfused and their brains were extracted for quantitative analysis of neurogenesis. ## 2.4. Morris Water Maze (MWM) Test Animals were subjected to behavioral studies10 weeks after the TPB, INU, and FLU diet. Animals underwent a behavioral MWM test 24 h after the last administration of TPB, INU, FLU, and water for injections (for the control group), according to the methods described earlier [45,46,47]. The MWM is the one of the most commonly used behavioral tests to assess learning and memory processes. In brief, to perform the MWM test, a mouse placed in a circular tank filled with water had to find a platform located above or just below the surface of the water, where it could safely rest. During 5 days of the training test recorded with a TSE (one daily session consisting of four 60-s trials), the animals learned and memorized how to find the platform by means of special signs placed on the walls of the room. On 6th day, the final test was performed. The course of the video tracking system VideoMot2 (TSE Systems, Berlin, Germany) and three parameters were measured: escape latency (the average time needed to find the platform), distance (the average distance traveled in order to find the platform), and time spent in the W-channel. ## 2.5. Fecal Collection, DNA Extraction, and NGS Sequencing Fresh fecal samples from fivemice in each group ($$n = 5$$) were collected into sterile Eppendorf tubes and frozenat −80 °C until needed for DNA extraction. Genomic DNA samples were extracted from about 15 mg of fecal samples using the GeneMATRIX Stool DNA Purification Kit (EurX, Gdańsk, Poland) and following the manufacturer’s protocol. Prior the extraction, all samples were homogenized, and the homogenization step was performed by MP FastPrep-24 Classic instrument (MP Biomedicals, USA) for 1 min in 6.5 m/s. DNA concentrations and the A260/A280 absorption ratios were assessed using an Eppendorf BioPhotometer D30 spectrophotometer (Eppendorf, Hamburg, Germany). After extraction, the DNA was stored at −20 °C until further use. The genetic material extracted for 19samples was sent to the Genoplast Laboratory (Poland) for library preparation and 16S sequencing using Illumina MiSeq platform. The V3-V4 hypervariable regions of the bacterial 16S rRNA gene were amplified using primers 341F/785R [48]. Details were congruent with methodology described in Kaczmarczyk-Ziemba et al. [ 49]. Raw NGS data are deposited and fully available in the Sequence Read Archive (accession number PRJNA912402). Demultiplexed paired-end reads were imported into QIIME2 (2019.1 release) [50]. The DADA2 algorithm was applied to filter out noise and correct errors in marginal sequences, remove chimeric sequences, merge paired-end reads, and summarize amplicon sequence variants (ASVs) [51]. Taxonomy assignment was performed with a pre-trained SILVA 132 $99\%$ OTUs based Naïve-Bayes classifier [52]. ASVs matching with chloroplast and mitochondrial sequences were removed from the dataset for downstream analyses. ## Biographical Analysis of Microorganisms Alpha diversity reflects the variation in microbial composition within a single sample. The following indices were used to estimate alpha diversity: Chao1 and Shannon. The Chao1 index (the so-called wealth index) refers to the abundance of individual samples, whereas the Shannon index summarizes the diversity of the population. Analyses were performed for four groups of samples (control, FLU, INU, and TPB). Alpha diversity measures (Chao1 richness index and Shannon diversity index) were calculated using the MicrobiomeAnalyst platform [53,54]. The non-parametric Kruskal–Wallis test was used to compare differences in alpha diversity between different groups. A fundamental property of microbiomes is beta diversity. It is a measure of the similarity or dissimilarity of samples and quantifies differences in overall taxonomic composition. ## 2.6. Transcardial Perfusion and Brain Slice Preparation Threeweeks after the last injection of BrdU, the animals were transcardiallyprefused. Immediately prior to perfusion, the mice were i.p. injected with the analgesic medetomidine at a dose of 0.25 mg/kg to relieve pain during surgery. Subsequently, each animal was anesthetized with $2\%$ isoflurane. After opening the mouse’s chest, a cannula was put into theleft ventricle of the heart. Next, saline and $4\%$ paraformaldehyde were injected under equal pressure to fix the brain tissues needed for further research. After perfusion, brains were dissected and cut into 50 µm sections with a Leica vibratome and stored at 4 °C in the cryoprotectant solution. Then, the sections prepared in this way were used for BrdU/NeuN/GFAP staining to determine the effects of TPB, INU, and FLU on the neurogenesis process. ## 2.7. Immunohistochemical Staining-Neurogenesis For determining the influence of TPB, INU, and FLU on the process of neurogenesis, 50 μm free-floating sections (stored at 4 °C) were immunohistochemically stained to visualize BrdU/NeuN and BrdU/GFAP positive cells according to the methods described previously [45,46,47,55,56]. ## 2.8. Confocal Microscopy and Cell Counting To further establish the phenotype of BrdU labeled cells, we performed a double/triple confocal immunofluorescence staining with BrdU, neuronal nuclei (NeuN), and glial acid fibrillar protein (GFAP) in the granular cell layer (GCL) and subgranular zone (SGZ) of dentate gyrus (DG) using the methods described earlier [45,46,47,55,56] five mice in each group ($$n = 5$$).The exact area of DG of the hippocampus designated for the quantitative analysis of neurogenesis is presented in Figure 2. Confocal imaging was performed using a Nikon A1R confocal microscope (Tokyo, Japan). ## 2.9.1. Statistical Analysis of MWM and Neurogenesis Results One-way ANOVA was used to analyze the data using the Windows version of the commercial program GraphPad Prism 8.0. ( GraphPad Software, San Diego, CA, USA). Bonferroni’s test was then applied for multiple comparisons. Every piece of data is expressed as a mean with standard errors. ## 2.9.2. Principal Coordinate Analysis (PCoA) PCoA was used to show species diversity between samples. Primer 7 [57] software was used to calculate the two-dimensional PCoA analysis and to generate the PCoA graph. ## 2.9.3. ANOSIM and PERMANOVA Analysis Analysis of similarities (ANOSIM) statistically determines whether the difference between the groups was greater than the difference within the groups. Primer7 software [57] was used to calculate it. In addition, it was used to calculate permutation-based multivariate analysis of variance (PERMANOVA) and to determine if there are significant differences between predefined sample groups. ## 2.9.4. SIMPER Analysis The SIMPER analysis was conducted to calculate the overall mean differences between the profiles of the compared groups and to identify the types of bacteria that primarily shape the differences between the microbiomes of the study groups. Past 4.0 [58] was used to perform the SIMPER analysis and pairwise calculation of the overall average differences between the profiles, as well as to identify the bacterial taxa associated primarily with the observed differences between the studied microbiomes. ## 2.9.5. LEfSe Analysis To identify bacterial taxa responsible for differences in beta diversity, LEfSe(linear discriminant analysis [LDA] effect size) analysis was performed [59]. The analysis was performed using the MicrobiomeAnalyst platform. Relative taxonomic abundances were used as input to the LEfSe pipeline. The metabolic potential of the gut microbiota was predicted by the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) analysis using the KEGG Pathway database [60]. ## 3.1. Effect of TPB, INU, and FLU Administration on the Body Weight of Healthy Mice Body weight gain analysis of mice receivingTPB, INU, and FLU at the 1st, 3rd, 5th, 7th, and 10th week of the diet showed no significant changes for TPB and INU throughout the duration of the experiments. Two weeks of FLU administration in the 9th and 10th week of the experiment indicated a slight decrease in weight for the FLU mice (Figure 3). ## 3.2. Effect of TPB, INU, and FLU Administration on Mouse Spatial Learning and Memory In order to assess potential memory and learning disorders as a result of TPB, INU, and FLU administration, the animals were subjected to the Morris Water Maze Test. Three parameters were analyzed: [1] the average time needed to find the platform, [2] the average distance traveled in order to find the platform, and [3] the mean percentageof time spent in the W-channel. No statistically significant disturbances in the average time and distance needed to find the platform in all study groups were observed compared to the control group;however, it should be emphasized that FLU mice showed the most deviated time and distance parameters in relation to the other groups (Figure 4A,B). Moreover, the time spent in the W channel for FLU mice was significantly shorter compared to the INU group (25.80 ± 2.930 and 46.78 ± 4.821; * $p \leq 0.05$, $$n = 7$$; Figure 4C). The analysis of W-channel data recorded with the use of Video Mot2 System software made it possible to visualize the directional flow paths for individual groups. Figure 5 shows a record of the route taken in the W-channel for a selected animal from each study group starting from the 3rd quadrant. ## 3.3. The Impact of Long-Term Treatment with TPB, INU and FLU on the Neurogenesis in the SGZ and GCL in Mice To determine whether TPB, INU, and FLU administration may affect the number of progenitor cells in the SGZ and GCL, BrdU, (a thymidine analog that is incorporated into DNA during cell division) was used to label dividing cells. Quantification of BrdU-labelled cells showed that 2-week administration of FLU at a dose of 12 mg/kg statistically significantly reduced the number of BrdU-positive cells compared to the control group, TPB, and INU (1242 ± 102.2 vs. 2022 ± 66.24, 2106 ± 33.73, 1801 ± 33.79, respectively, $p \leq 0.0001$, $p \leq 0.0001$, $p \leq 0.001$, $$n = 5$$, Figure 6A). Interestingly, 10-week INU administration significantly decreased the total number of BrdU-positive cells compared to TPB (1801 ± 33.79 vs. 2106 ± 33.73, respectively, $p \leq 0.05$, $$n = 5$$; Figure 6A). In order to assess possible changes in the proliferation, migration, and differentiation of newly formed cells labeled with the BrdU marker into neurons (NeuN) and astrocytes (GFAP), immunohistochemical staining of colocalization of BrdU/NeuN and BrdU/GFAP-positive cells was performed. The analysis of the obtained results showed that the number of cells labeled with BrdU/NeuN was significantly reduced in mice receiving FLU compared to the control group, TPB and INU (737.6 ± 60.63 vs. 1250 ± 40.95, 1252 ± 19.95, 1106 ± 20.72 respectively, $p \leq 0.0001$, $$n = 5$$, Figure 6B). The labeling of a specific GFAP protein co-localizing with BrdU allowed visualization of the differences in the level of newly formed astrocytes in the hippocampus. A statistically significant reduction in the number of GFAP+ cells was observed in the FLU group compared to the control, TPB, and INU groups (122.8 ± 10.03 vs. 194.2 ± 6.351, 221.0 ± 3.479, 165 ± 3.169, respectively, $p \leq 0.0001$, $p \leq 0.0001$, $p \leq 0.01$, $$n = 5$$, Figure 6C). Interestingly, INU mice showed a significantly lower level of astrocytes compared to control and TPB mice (165 ± 3.169 vs. 194.2 ± 6.351, 221.0 ± 3.479, respectively, $p \leq 0.05$, $p \leq 0.0001$; $$n = 5$$, Figure 6C). Representative images of immunohistochemical changes for each of the study groups (control, TPB, INU, FLU) were included in the Supplementary File (Figures S1–S4). ## 3.4. Effect of Long-Term Administration of TPB, INU, and FLU on the Composition of the Intestinal Microbiota in Mice To evaluate whether the gut microbiome was changed by TPB, INU, and FLU administration in mice, we conducted an in-depth analysis of 16S rRNA sequencing. The alpha diversity analysis revealed differences within the composition of the microbial community of each group (Figure 7A). Compared to the control group, the Chao1 index for INU and TPB groups decreased, whereas, for FLU mice, an increase was observed ($H = 8.0849$, $$p \leq 0.04$$). In turn, Shannon diversity indices were more similar and did not show significant differences. Across all tested samples, four phyla were the most abundant: Bacteroidetes, Firmicutes, Epsilonbacteraeota and Proteobacteria (Figure 8). Moreover, the abundance of Actinobacteria was higher in the FLU group than in other sample sets. At the genus level, seventeen identified genera were observed with anabundance higher than $1\%$ in at least two tested groups (Figure 9). Their abundances differed among tested microbiota profiles (Supplementary Table S1). For example, Dubosiella was observed only in control and FLU groups, and Blautia was more abundant in those two groups than in INU and TPB ($3.22\%$ in control, $2.39\%$ in FLU, $0.09\%$ in INU, and $0.12\%$ in TPB). In turn, the genus Bacteroides was observed in all but one profilewith a relative abundance above $8\%$. Only in the FLU profile was its abundance lower ($1.30\%$). It was also observed that the abundance of Lactobacillus was the highest in the TPB group, whereas it was significantly reduced in the FLU group compared to the control group. Results of ANOSIM analysis confirmed lower distances among intragroup samples than among intergroup samples ($R = 0.342$, $$p \leq 0.001$$). The PCoA analysis based on the abundances of all identified microbial taxa showed a high similarity of the tested microbiota profilegroups according to the treatment except FLU mice, where the subtle separation was observed (Figure 10). Moreover, results from pairwise PERMANOVA analysis supported also FLU mice separation (Table 1). The results of the SIMPER analysis showed differences between the microbiome of the FLU groups and the other three profiles tested. Calculated overall average dissimilarities between profiles of compared groups were as follows: $48.65\%$ for FLU-control, $50.95\%$ for FLU-INU, and $49.40\%$ for FLU-TPB. In turn, in all three comparisons (i.e., FLU-control, FLU-INU, and FLU-TPB), nine bacterial genera were identified as those shaping differences between groups’ microbiota (Helicobacter, Bacteroides, Lactobacillus, Dubosiella, Blautia, Lachnospiraceae NK4A136 group, Alloprevotella, Bifidobacterium, and Anaerostipes, respectively). In the case of FLU-INU comparison, an additional four genera were linked with observed differences (Alistipes, Prevotellaceae UCG-001, Lachnospiraceae UCG-001, and Prevotellaceae NK3B31). In turn, differences between FLU and TPB groups were linked to the abundance of unidentified representatives of the Gastranaerophilales order and Desulfovibrionaceae family. Based on the LEfSe analysis performed, six taxa of bacteria were identified to explain the difference in gut microbiota between the different treatment groups (Figure 11). The results showed that the biomarker taxa in the control group at the genus level were Bacteroides and Ruminococcaceae UCG-013. The biomarker genera in the FLU groups were Dubosiella, Bifidobacterium, and Faecalibacterium. In turn, for INU group Prevotellaceae UCG-001 was identified as a marker genus. Only for the TPB group did LEfSe not determine the biomarker taxa. Histogram of the LDA scores reveals the most differentially abundant taxa among tested groups. ## 4. Discussion Over the past few years, there has been an increasing interest around the interactions between the gut microbiota and the effects of supplementation, including prebiotics as well as drugs in various therapeutic areas. In this study, we assessed thepotential impact of orally administered natural prebiotics—TPB and INU, and the antidepressant drug—FLU on learning and memory, neurogenesis, and the composition of the intestinal microbiota in healthy mice. The research results proved that the 10-week supplementation with TPB and INU stimulates the growth of probiotic bacteria, with no negative effect on the cognitive functions and the proliferation of neural stem cells in the tested animals. A 2-week administration of FLU confirmed its negative impact onbehavioral functions and neurogenesis. Moreover, we have also demonstrated an inhibitory effect of FLU on the growth of Lactobacillus. Our research indicated no significant negative effect of TPB, INU, and FLU on the weight of the tested animals, which is consistent with the results obtained by Koch and coworkers [61] using supplementation with INU/FOS in C57BL/6 mice. Similarly, research by Petersen et al. [ 62] in BALB/c mice provided information that a 3-week supplementation of $10\%$ of the diet with INU, FOS, XOS, GOS, apple pectin, polydextrose, or beta-glucan didn’t disturb the body weight of the tested animals. In turn, results from clinical trials showed, that FLU given chronically at a dose of 60 mg/day might have a modest effect on weight loss compared to a placebo in adults with overweight or obesity, although it may also induce many adverse events [63]. The MWM test was performed to evaluate the cognitive functions of healthy mice after INU, TPB, and FLU administration. The results of our experiment showed that mice receiving FLU covered a longer route and needed much more time to locate the platform than TPB, INU, and control mice, which indicated memory and learning impairments in the FLU group. In addition, the FLU mice spent definitely less time in the W-channel compared to the other groups, suggesting an impairment in spatial recognition. The effect of treatment with SSRIs such as FLU on behavior has been the subject of numerous studies in recent years. In vivo studies indicated that chronic therapy with FLU may have a very different effect on the behavior of treated rats or mice, depending on disease/dysfunction/health condition. The results we obtained are consistent with the studies of Majlesii and Naghdi [64] showing that FLU at doses of 8 and 16 mg/kg and citalopram at doses of 4 and 8 mg/kg significantly impaired the performance of the rats in the MWM test compared to the control animals. Interestingly, results obtained by Golub et al. [ 65] from the study on male juvenile rhesus monkeys indicated that 2 years of treatment with the drug impaired a sustained attention. Although response accuracy has not been affected, FLU monkeys had more missed trial initiations and choices during testing than control animals. On the contrary, FLU (5 mg/kg) treatment used in rat model of Alzheimer’s disease (AD) has been shown to increase learning and memory processes when compared with control AD rats. It should be emphasized, that FLU used in the treatment of neurological diseases responsible for cognitive dysfunctions has a beneficial effect, and therefore can improve the ability to learn and remember. Taking into account the TPB and INU groups, the obtained results for all measured parameters were similar to the control mice, which confirms no impact of supplementation on learning and memory functions. Interestingly, the studies by Messaoudi et al. [ 66] have provided information that INU may significantly improve cognitive functions. They observed that a 14-day administration of a $5\%$ or $10\%$ mixture of INU and oligofructose inmale Wistar rats had a beneficial effect on the behavior of the tested animals. In turn, enriching the diet with INU enhanced the effect of the probiotic bacteria (Enterococcus faecium) and improved learning and memory in Sprague-Dawley rats [67]. Moreover, preliminary results from human studies conducted on young and middle-aged volunteers confirmed that the consumption of prebiotics such as FOS and INU, in doses of 5–10 g per day for 4–12 weeks, may be beneficial for brain function improvement (i.e., learning and working memory) and behavior (i.e., anxiety and mood) [68,69,70]. Taking into account the fact that TPB contains a significant amount of INU, the results of our research prove the beneficial effect of TPB oncognitive functions. In the next part of the study, we evaluated, for the first time, the potent impact of TPB, INU, and FLU administration on the process of proliferation, migration, and differentiation ofthe neural stem cells in the mouse brain. The results of our study showed that a 10-week diet with TPB and INU did not affect the neurogenesis process in mice compared to the control group. At the moment there are no scientific reports concerning the effect of prebiotic administration on the course of the neurogenesis process. Interestingly a 2-week FLU administration significantly disturbed the process of neurogenesis. Similar to our current data, results from the study by Klomp et al. [ 71] showed an inhibitory effect of a 3-week administration of FLU (5 mg/kg) on the process of neurogenesis in rats. Moreover, in several other studies conducted on rodents, the lack of a stimulating effect of antidepressants such as FLU on the neurogenesis process was observed [72,73]. In turn, contrary results were presented by Marcussen et al. [ 74], where a 28-day treatment of FLU (10 mg/kg)was shown to stimulate neurogenesis in healthy rats. Similarly, the obtained results by Hovorka et al. [ 75] showed that a 14-day administration of FLU (4 mg/kg) induces the formation of new cells in growing rats. In addition, Hodes and colleagues [76] showed that 26-day treatment of female mice with FLU at doses of 5 and 10 mg/kg increased cell proliferation, whereas a low dose of FLU (2.5 mg/kg) did not affect the neurogenesis process in the tested animals [76]. As the final stage of the research, 16S rRNA sequencing was performed to assess changes in the gut microbiome of mice. We observed that the intestinal microbiota of mice receiving FLU was significantly different from the other study groups. The following bacterial taxa affecting differences in the gut microbiota were identified: for the FLU group Dubosiella, Bifidobacterium, and Faecalibacterium; for INU mice Prevotellaceae UCG-001, and for the control group Bacteroides and Ruminococcaceae UCG-01. Surprisingly, the biomarker taxa could not be determined for the TPB group. Additionally, four types of bacteria were most abundant in all tested groups: Bacteroidetes, Firmicutes, Epsilonbacteraeota,and Proteobacteria. Interestingly, the abundance of Lactobacillus was noticeably higher in the TPB group compared to the control group. Similar results were obtained in our previous initial screening study in male Albino Swiss mice [41], where 4-week supplementation with TPB (250 mg/kg) stimulated the growth of one of the most common probiotic bacteria, Lactobacillus gasseri, as well as Enterobacteriaceae (Escherichia coli, Enterobacter asburiae, Kliebsiellaoxytoca). The beneficial impact of TPB on the microbiota can be explained by the high content of INU in TPB tubers and as has been shown so far thatINU stimulates the growth of probiotic bacteria of the genus Lactobacillus and Bifidobacterium [77]. The results of our study are also in line with an earlier experiment by Samal et al. [ 43], evaluating the effect of a 12-week supplementation with TPB tuber meal (0, 2, 4, $6\%$) on the growth of probiotic bacteria in the intestines of rats. They showed that the enrichment of the diet in TPB increased the Lactobacillus spp. population and Bifidobacterium spp. in the cecum, colon, and anus. Bioinformatic analysesfrom our experiment allowed us to study the effect of 2 weeks of FLU administration on the composition and number of the intestinal microbiota. One of the more important effects of FLU was the inhibition of the growth of Lactobacillus. Similar data, although from in vitro studies by Cussotto et al. [ 35], indicated that FLU at doses of 400 and 600 μg/mL completely stopped the growth of L. rhamnosus. In addition, at doses of 100, 400, and 600 μg/mL, it inhibited the multiplication of Escherichia coli. Further in vitro studies indicated that FLU and other SSRIs, that is, sertraline and paroxetine, exhibited antimicrobial activity against some Gram-positive bacteria, including Staphylococcus and Enterococcus [78,79]. Interestingly, Lyte et al. [ 36] noted a decrease in the number of Lactobacilli after 29-day treatment with FLU (20 mg/kg) in mice, although only some of the Lactobacillus strains were reduced, which may indicate a differential effect of FLU on different Lactobacillus strains. However, it should be mentioned that some of the results presented by other researchers differ from our reports. We observed a decrease in the number of Lachnospiraceae NK4A136 in the FLU group compared to the control mice, whereas research by Lyte et al. [ 36] showed that several OTUs belonging to the family Lachnospiraceae (OTUs 32, 38, 86, 93) were significantly more numerous in FLU-treated mice. Another bacterial taxon distinguishing the FLU group in our study was Faecalibacterium. It is well known that this bacterium is involved in the production of butyric acid, the main SCFA produced by the intestinal flora [80]. Moreover, in the studies involving patients with major depressive disorder (MDD), this bacterium was identified as a potential MDD biomarker [81]. Similarly, Zhou et al. [ 82] showed that both Faecalibacterium and Butyricicoccus may be important in diagnosing and treating patients suffering from postpartum depression. Our findings revealed that the dominant type of bacteria in the mouse gut microbiota, Bacteroides, was most abundant in the control, INU, and TPB groups. The lowest amount was observed in the FLU group. Studies conducted so far have shown that Bacteroides metabolize polysaccharides and oligosaccharides, providing food and vitamins to the host and other intestinal bacteria. It is also known that these bacteria maintain a complex and generally beneficial relationship with the host when they remain in the intestine, but if they change the localization they can become pathogenic [83] Analyzing the microbiome results obtained for the INU group, we observed an increase in the number of Prevotellaceae UCG 001. These results are in line with the study by Song et al. [ 84] where 4 weeks of INU (10 g/kg/day) supplementation increased the abundance of the family Prevotellaceae in C57BL/6J ob/ob mice, which are deficient in the leptin gene. In the same study, it was noted that the administration of INU also increased the amount of Prevotellaceae UCG 001 in mice. Prevotella, as a succinate-producing bacterium, may participate in the decomposition of INU [85]. In addition, it has enzymes that are responsible for the degradation of cellulose and xylan [86]. On this basis, we can assume that enriching the diet with INU has a beneficial effect on the increase in the number of Prevotella UCG-001. The analysis of the results obtained from our research, namely the effect of prebiotic supplementation on the microbiota, the BGM axis, and thus on cognitive functions and neurogenesis in mice, undoubtedly confirms the dependence of healthy brain function on proper homeostasis and gut-brain communication. Metabolites produced by the intestinal microbiota, such as SCFAs, amino acids, modified peptides, or oligosaccharides, may have an impact on the improvement of cognitive functions, which was indicated in animal models of neurodevelopmental and neurodegenerative diseases [87,88]. In turn, results from the study by Ogbonnay et al. [ 20] clearly proved the importance of microbiota for a proper neurogenesis. They indicated the altered levels of hippocampal neurogenesis in GF mice. The results of our research showed that the bacterial taxa involved in the production of these secondary metabolites are primarily Faecalibacterium and Bacteroides. Research by Bravo et al. [ 11] showed that Lactobacillus regulates emotional behavior and central GABA receptor expression in mice via the vagus nerve. Moreover, Liang et al. [ 89] reported that Lactobacillus helveticus NS8 improved behavioral and cognitive impairments in rats. In this study, we observed a reduction in the abundance of Lactobacillus in fecal samples of the FLU group, which may be a potential contributor to cognitive impairment. At the moment, our knowledge of the metabolic properties of many strains of bacteria is still limited. Therefore, it is extremely important to continue and implement advanced research in order to understand the significance of supplementation in the proper functioning of the microbiota and the BGM axis, especially in the context of exposure to various environmental factors that can induce dysbiosis. ## 5. Conclusions In conclusion, the presented research showed that a long-term diet enriched with TPB or INU stimulates the growth of probiotic bacteria, does not affect the learning and memory process, and it does not disturb the proliferation of neural stem cells in the tested animals. Based on this data we can assume that both TPB and INU seem to be safe for the proper course of neurogenesis. On the contrary, 2-week administration of FLU confirmed an inhibitory impact on Lactobacillus growth and negatively affected behavioral function and neurogenesis in healthy animals. The above studies suggest that the natural prebiotics TPB and INU may have the potential to become natural supplements that enrich the diversity of intestinal microbiota while having a beneficial effect on the BGM axis, cognitive functions, and neurogenesis, which is certainly worth further advanced preclinical research. ## References 1. Sender R., Fuchs S., Milo R.. Revised Estimates for the Number of Human and Bacteria Cells in the Body. PLoS Biol. (2016) 14. DOI: 10.1371/journal.pbio.1002533 2. Cryan J.F., Dinan T.G.. Mind-altering microorganisms: The impact of the gut microbiota on brain and behaviour. Nat. Rev. Neurosci. (2012) 13 701-712. DOI: 10.1038/nrn3346 3. Sekirov I., Russell S.L., Antunes L.C.M., Finlay B.B., Galla S., Chakraborty S., Cheng X., Yeo J., Mell B., Zhang H.. Gut Microbiota in Health and Disease. Physiol. Rev. (2010) 90 859-904. DOI: 10.1152/physrev.00045.2009 4. Murphy E.F., Cotter P.D., Healy S., Marques T.M., O’Sullivan O., Fouhy F., Clarke S.F., O’Toole P.W., Quigley E.M., Stanton C.. Composition and energy harvesting capacity of the gut microbiota: Relationship to diet, obesity and time in mouse models. Gut (2010) 59 1635-1642. DOI: 10.1136/gut.2010.215665 5. Cong X., Henderson W.A., Graf J., McGrath J.M.. Early Life Experience and Gut Microbiome: The Brain-Gut-Microbiota Signaling System. Adv. Neonatal Care (2015) 15 314-323. DOI: 10.1097/ANC.0000000000000191 6. Rowland I., Gibson G., Heinken A., Scott K., Swann J., Thiele I., Tuohy K.. Gut microbiota functions: Metabolism of nutrients and other food components. Eur. J. Nutr. (2018) 57 1-24. DOI: 10.1007/s00394-017-1445-8 7. Patterson E., Cryan J.F., Fitzgerald G.F., Ross R.P., Dinan T.G., Stanton C.. Gut microbiota, the pharmabiotics they produce and host health. Proc. Nutr. Soc. (2014) 73 477-489. DOI: 10.1017/S0029665114001426 8. Rhee S.H., Pothoulakis C., Mayer E.A.. Principles and clinical implications of the brain–Gut–Enteric microbiota axis. Nat. Rev. Gastroenterol. Hepatol. (2009) 6 306-314. DOI: 10.1038/nrgastro.2009.35 9. Wood J.D., Alpers D.H., Andrews P.L.R.. Fundamentals of neurogastroenterology. Gut (1999) 45 II6-II16. DOI: 10.1136/gut.45.2008.ii6 10. Bercik P., Park A.J., Sinclair D., Khoshdel A., Lu J., Huang X., Deng Y., Blennerhassett P.A., Fahnestock M., Moine D.. The Anxiolytic effect of bifidobacterium longum NCC3001 involves vagal pathways for gut-brain communication. Neurogastroenterol. Motil. (2011) 23 1132-1139. DOI: 10.1111/j.1365-2982.2011.01796.x 11. Bravo J.A., Forsythe P., Chew M.V., Escaravage E., Savignac H.M., Dinan T.G., Bienenstock J., Cryan J.F.. Ingestion of Lactobacillus strain regulates emotional behavior and central GABA receptor expression in a mouse via the vagus nerve. Proc. Natl. Acad. Sci. USA (2011) 108 16050-16055. DOI: 10.1073/pnas.1102999108 12. Chen D., Yang X., Yang J., Lai G., Yong T., Tang X., Shuai O., Zhou G., Xie Y., Wu Q.. Prebiotic Effect of Fructooligosaccharides from Morinda officinalis on Alzheimer’s Disease in Rodent Models by Targeting the Microbiota-Gut-Brain Axis. Front. Aging Neurosci. (2017) 9 403. DOI: 10.3389/fnagi.2017.00403 13. Bercik P., Collins S.M., Lyte M., Cryan J.F.. The effects of inflammation, infection and antibiotics on the microbiota-gut-brain axis. Micro-Bial Endocrinology: The Microbiota-Gut-Brain Axis in Health and Disease (2014) 279-289 14. Sudo N., Chida Y., Aiba Y., Sonoda J., Oyama N., Yu X.-N., Kubo C., Koga Y.. Postnatal microbial colonization programs the hypothalamic-pituitary-adrenal system for stress response in mice. J. Physiol. (2004) 558 263-275. DOI: 10.1113/jphysiol.2004.063388 15. Dinan T.G., Cryan J.F.. Brain–Gut–microbiota axis—Mood, metabolism and behaviour. Nat. Rev. Gastroenterol. Hepatol. (2017) 14 69-70. DOI: 10.1038/nrgastro.2016.200 16. Gareau M.G.. Microbiota-Gut-Brain Axis and Cognitive Function. Adv. Exp. Med. Biol. (2014) 817 357-371. DOI: 10.1007/978-1-4939-0897-4_16 17. Chen Y., Xu J., Chen Y.. Regulation of Neurotransmitters by the Gut Microbiota and Effects on Cognition in Neurological Disorders. Nutrients (2021) 13. DOI: 10.3390/nu13062099 18. Wang X.-Q., Li H., Li X.-N., Yuan C.-H., Zhao H.. Gut-Brain Axis: Possible Role of Gut Microbiota in Perioperative Neurocognitive Disorders. Front. Aging Neurosci. (2021) 13 745774. DOI: 10.3389/fnagi.2021.745774 19. Escobar Y.-N.H., O’Piela D., Wold L.E., Mackos A.R.. Influence of the Microbiota-Gut-Brain Axis on Cognition in Alzheimer’s Disease. J. Alzheimer’s Dis. (2022) 87 17-31. DOI: 10.3233/JAD-215290 20. Ogbonnaya E.S., Clarke G., Shanahan F., Dinan T.G., Cryan J.F., O’Leary O.F.. Adult Hippocampal Neurogenesis Is Regulated by the Microbiome. Biol. Psychiatry (2015) 78 e7-e9. DOI: 10.1016/j.biopsych.2014.12.023 21. Kundu P., Lee H.U., Garcia-Perez I., Tay E.X.Y., Kim H., Faylon L.E., Martin K.A., Purbojati R., Drautz-Moses D.I., Ghosh S.. Neurogenesis and prolongevity signaling in young germ-free mice transplanted with the gut microbiota of old mice. Sci. Transl. Med. (2019) 11 4760. DOI: 10.1126/scitranslmed.aau4760 22. Scott G.A., Terstege D.J., Vu A.P., Law S., Evans A., Epp J.R.. Disrupted Neurogenesis in Germ-Free Mice: Effects of Age and Sex. Front. Cell Dev. Biol. (2020) 8 407. DOI: 10.3389/fcell.2020.00407 23. Celorrio M., Abellanas M.A., Rhodes J., Goodwin V., Moritz J., Vadivelu S., Wang L., Rodgers R., Xiao S., Anabayan I.. Gut microbial dysbiosis after traumatic brain injury modulates the immune response and impairs neurogenesis. Acta Neuropathol. Commun. (2021) 9 40. DOI: 10.1186/s40478-021-01137-2 24. Cameron H.A., Glover L.R.. Adult Neurogenesis: Beyond Learning and Memory. Annu. Rev. Psychol. (2015) 66 53-81. DOI: 10.1146/annurev-psych-010814-015006 25. Kelly J.R., Keane V.O., Cryan J.F., Clarke G., Dinan T.G.. Mood and Microbes. Gastroenterol. Clin. N. Am. (2019) 48 389-405. DOI: 10.1016/j.gtc.2019.04.006 26. Pusceddu M.M., Barboza M., Keogh C.E., Schneider M., Stokes P., Sladek J.A., Kim H.J.D., Torres-Fuentes C., Goldfild L.R., Gillis S.E.. Nod-like receptors are critical for gut–brain axis signalling in mice. J. Physiol. (2019) 597 5777-5797. DOI: 10.1113/JP278640 27. Ribeiro M.F., Santos A.A., Afonso M.B., Rodrigues P.M., Sá Santos S., Castro R.E., Rodrigues C.M.P., Solá S.. Di-et-dependent gut microbiota impacts on adult neurogenesis through mitochondrial stress modulation. Brain Commun. (2020) 2 fcaa165. DOI: 10.1093/braincomms/fcaa165 28. Cryan J.F., O’Mahony S.M.. The microbiome-gut-brain axis: From bowel to behavior. Neurogastroenterol. Motil. (2011) 23 187-192. DOI: 10.1111/j.1365-2982.2010.01664.x 29. Grenham S., Clarke G., Cryan J.F., Dinan T.G.. Brain-Gut-Microbe Communication in Health and Disease. Front. Physiol. (2011) 2 94. DOI: 10.3389/fphys.2011.00094 30. O’Mahony S.M., Hyland N.P., Dinan T.G., Cryan J.F.. Maternal separation as a model of brain-gut axis dysfunction. Psychopharmacology (2011) 214 71-88. DOI: 10.1007/s00213-010-2010-9 31. Vich Vila A., Collij V., Sanna S., Sinha T., Imhann F., Bourgonje A., Mujagic Z., Jonkers D.M.A.E., Masclee A.A.M., Fu J.. Impact of commonly used drugs on the composition and metabolic function of the gut microbiota. Nat. Commun. (2020) 11 362. DOI: 10.1038/s41467-019-14177-z 32. Zhang W., Qu W., Wang H., Yan H.. Antidepressants fluoxetine and amitriptyline induce alterations in intestinal microbiota and gut microbiome function in rats exposed to chronic unpredictable mild stress. Transl. Psychiatry (2021) 11 131. DOI: 10.1038/s41398-021-01254-5 33. Sommi R.W., Crismon M.L., Bowden C.L.. Fluoxetine: A Serotonin-specific, Second-generation Antidepressant. Pharmacother. J. Hum. Pharmacol. Drug Ther. (1987) 7 3496. DOI: 10.1002/j.1875-9114.1987.tb03496.x 34. Silva L.J., Martins M.C., Pereira A.M., Meisel L.M., Gonzalez-Rey M., Bebianno M.J., Lino C.M., Pena A.. Uptake, accumulation and metabolization of the antidepressant fluoxetine by Mytilus galloprovincialis. Environ. Pollut. (2016) 213 432-437. DOI: 10.1016/j.envpol.2016.02.022 35. Cussotto S., Strain C.R., Fouhy F., Strain R.G., Peterson V.L., Clarke G., Stanton C., Dinan T.G., Cryan J.F.. Differential effects of psychotropic drugs on microbiome composition and gastrointestinal function. Psychopharmacology (2018) 236 1671-1685. DOI: 10.1007/s00213-018-5006-5 36. Lyte M., Daniels K.M., Schmitz-Esser S.. Fluoxetine-induced alteration of murine gut microbial community structure: Evidence for a microbial endocrinology-based mechanism of action responsible for fluoxetine-induced side effects. PeerJ (2019) 7 e6199. DOI: 10.7717/peerj.6199 37. Quigley E.M.. Prebiotics and Probiotics in Digestive Health. Clin. Gastroenterol. Hepatol. (2018) 17 333-344. DOI: 10.1016/j.cgh.2018.09.028 38. Markowiak P., Śliżewska K.. Effects of Probiotics, Prebiotics, and Synbiotics on Human Health. Nutrients (2017) 9. DOI: 10.3390/nu9091021 39. Saulnier D.M., Ringel Y., Heyman M.B., Foster J.A., Bercik P., Shulman R.J., Versalovic J., Verdu E.F., Dinan T.G., Hecht G.. The intestinal microbiome, probiotics and prebiotics in neurogastroenterology. Gut Microbes (2013) 4 17-27. DOI: 10.4161/gmic.22973 40. Moshfegh A.J., Friday J.E., Goldman J.P., Ahuja J.K.C.. Presence of Inulin and Oligofructose in the Diets of Americans. J. Nutr. (1999) 129 S1407-S1411. DOI: 10.1093/jn/129.7.1407S 41. Szewczyk A., Zagaja M., Bryda J., Kosikowska U., Stępień-Pyśniak D., Winiarczyk S., Andres-Mach M.. Topinambur–new possibilities for use in a supplementation diet. Ann. Agric. Environ. Med. (2019) 26 24-28. DOI: 10.26444/aaem/102767 42. Zaky E.A.. Physiological Response to Diets Fortified with Jerusalem Artichoke Tubers (. Environ. Sci. (2009) 5 682-688 43. Samal L., Chaturvedi V., Pattanaik A.. Effects of dietary supplementation with Jerusalem artichoke (. J. Anim. Feed. Sci. (2017) 26 50-58. DOI: 10.22358/jafs/68779/2017 44. Sun L., Zhang H., Cao Y., Wang C., Zhao C., Wang H., Cui G., Wang M., Pan Y., Shi Y.. Fluoxetine ameliorates dysbiosis in a depression model induced by chronic unpredicted mild stress in mice. Int. J. Med. Sci. (2019) 16 1260-1270. DOI: 10.7150/ijms.37322 45. Andres-Mach M., Szewczyk A., Zagaja M., Luszczki J., Maj M., Rola R., Abram M., Kaminski K.. Evaluation of the impact of compound C11 a new anticonvulsant candidate on cognitive functions and hippocampal neurogenesis in mouse brain. Neuropharmacology (2020) 163 107849. DOI: 10.1016/j.neuropharm.2019.107849 46. Andres-Mach M., Szewczyk A., Zagaja M., Szala-Rycaj J., Lemieszek M.K., Maj M., Abram M., Kaminski K.. Preclinical Assessment of a New Hybrid Compound C11 Efficacy on Neurogenesis and Cognitive Functions after Pilocarpine Induced Status Epilepticus in Mice. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22063240 47. Szewczyk A., Zagaja M., Szala-Rycaj J., Maj M., Andres-Mach M.. Effect of Lacosamide and Ethosuximide Chronic Treatment on Neural Precursor Cells and Cognitive Functions after Pilocarpine Induced Status Epilepticus in Mice. Brain Sci. (2021) 11. DOI: 10.3390/brainsci11081014 48. Klindworth A., Pruesse E., Schweer T., Peplies J., Quast C., Horn M., Glöckner F.O.. Evaluation of General 16S Ribosomal RNA Gene PCR Primers for Classical and Next-Generation Sequencing-Based Diversity Studies. Nucleic Acids Res. (2013) 41 e1. DOI: 10.1093/nar/gks808 49. Kaczmarczyk-Ziemba A., Zagaja M., Wagner G.K., Pietrykowska-Tudruj E., Staniec B.. First Insight into Microbiome Profiles of Myrmecophilous Beetles and Their Host, Red Wood Ant Formica polyctena (Hymenoptera: Formicidae)—A Case Study. Insects (2020) 11. DOI: 10.3390/insects11020134 50. Bolyen E., Rideout J.R., Dillon M.R., Bokulich N.A., Abnet C.C., Al-Ghalith G.A., Alexander H., Alm E.J., Arumugam M., Asnicar F.. Reproducible, Interactive, Scalable and Extensible Microbiome Data Science using QIIME 2. Nat. Biotechnol. (2019) 37 852-857. DOI: 10.1038/s41587-019-0209-9 51. Callahan B.J., Mcmurdie P.J., Rosen M.J., Han A.W., Johnson A.J.A., Holmes S.P.. DADA. Nat. Methods (2016) 13 581-583. DOI: 10.1038/nmeth.3869 52. Quast C., Pruesse E., Yilmaz P., Gerken J., Schweer T., Yarza P., Peplies J., Glöckner F.O.. The SILVA Ribosomal RNA Gene Database Project: Improved Data Processing and Web-Based Tools. Nucleic Acids Res. (2013) 41 D590-D596. DOI: 10.1093/nar/gks1219 53. Chong J., Liu P., Zhou G., Xia J.. Using MicrobiomeAnalyst for comprehensive statistical, functional, and meta-analysis of microbiome data. Nat. Protoc. (2020) 15 799-821. DOI: 10.1038/s41596-019-0264-1 54. Dhariwal A., Chong J., Habib S., King I.L., Agellon L.B., Xia J.. MicrobiomeAnalyst: A web-based tool for comprehensive statistical, visual and meta-analysis of microbiome data. Nucleic Acids Res. (2017) 45 W180-W188. DOI: 10.1093/nar/gkx295 55. Andres-Mach M., Haratym-Maj A., Zagaja M., Rola R., Maj M., Chrościńska-Krawczyk M., Luszczki J.J.. ACEA (a highly selective cannabinoid CB1 receptor agonist) stimulates hippocampal neurogenesis in mice treated with antiepileptic drugs. Brain Res. (2015) 1624 86-94. DOI: 10.1016/j.brainres.2015.07.028 56. Andres-Mach M., Zagaja M., Haratym-Maj A., Rola R., Maj M., Haratym J., Dudra-Jastrzębska M., Łuszczki J.J.. A Long-Term Treatment with Arachidonyl-2′-Chloroethylamide Combined with Valproate Increases Neurogenesis in a Mouse Pilocarpine Model of Epilepsy. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18050900 57. Clarke K.R., Gorley R.N.. PRIMER v7: User Manual/Tutorial (2015) Volume 29 1060-1065 58. Hammer Ø., Harper D.A.T., Ryan P.D.. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontol. Electronica. (2001) 4 9 59. Segata N., Izard J., Waldron L., Gevers D., Miropolsky L., Garrett W.S., Huttenhower C.. Metagenomic biomarker discovery and explanation. Genome Biol. (2011) 12 R60. DOI: 10.1186/gb-2011-12-6-r60 60. Douglas G.M., Maffei V.J., Zaneveld J.R., Yurgel S.N., Brown J.R., Taylor C.M., Huttenhower C., Langille M.G.I.. PICRUSt2 for prediction of metagenome functions. Nat. Biotechnol. (2020) 38 685-688. DOI: 10.1038/s41587-020-0548-6 61. Koch F., Derno M., Langhammer M., Tuchscherer A., Hammon H.M., Mielenz M., Metges C.C., Kuhla B.. A high-protein diet containing inulin/oligofructose supports body weight gain associated with lower energy expenditure and carbohydrate oxidation, and alters faecal microbiota in C57BL/6 mice. J. Nutr. Sci. (2021) 10 42. DOI: 10.1017/jns.2021.42 62. Petersen A., Heegaard P.M., Pedersen A.L., Andersen J.B., Sørensen R.B., Frøkiær H., Lahtinen S.J., Ouwehand A.C., Poulsen M., Licht T.R.. Some putative prebiotics increase the severity of Salmonella entericaserovar Typhimurium infection in mice. BMC Microbiol. (2009) 9 245. DOI: 10.1186/1471-2180-9-245 63. Serralde-Zuñiga A.E., González-Garay A.G., Rodríguez-Carmona Y., Meléndez-Mier G.. Use of Fluoxetine to Reduce Weight in Adults with Overweight or Obesity: Abridged Republication of the Cochrane Systematic Review. Obes. Facts (2022) 15 473-486. DOI: 10.1159/000524995 64. Majlessi N., Naghdi N.. Impaired spatial learning in the Morris water maze induced by serotonin reuptake inhibitors in rats. Behav. Pharmacol. (2002) 13 237-242. DOI: 10.1097/00008877-200205000-00007 65. Golub M.S., Hackett E.P., Hogrefe C.E., Leranth C., Elsworth J.D., Roth R.H.. Cognitive performance of juvenile monkeys after chronic fluoxetine treatment. Dev. Cogn. Neurosci. (2017) 26 52-61. DOI: 10.1016/j.dcn.2017.04.008 66. Messaoudi M., Rozan P., Nejdi A., Hidalgo S., Desor D.. Behavioural and cognitive effects of oligofructose-enriched inulin in rats. Br. J. Nutr. (2005) 93 S27-S30. DOI: 10.1079/BJN20041348 67. Romo-Araiza A., Gutiérrez-Salmeán G., Galván E.J., Hernández-Frausto M., Herrera-López G., Romo-Parra H., García-Contreras V., Fernández-Presas A.M., Jasso-Chávez R., Borlongan C.V.. Probiotics and Prebiotics as a Therapeutic Strategy to Improve Memory in a Model of Middle-Aged Rats. Front. Aging Neurosci. (2018) 10 416. DOI: 10.3389/fnagi.2018.00416 68. Schmidt K., Cowen P., Harmer C., Tzortzis G., Errington S., Burnet P.W.J.. Prebiotic intake reduces the waking cortisol response and alters emotional bias in healthy volunteers. Psychopharmacology (2014) 232 1793-1801. DOI: 10.1007/s00213-014-3810-0 69. Smith A.P.. The concept of well-being: Relevance to nutrition research. Br. J. Nutr. (2005) 93 S1-S5. DOI: 10.1079/BJN20041351 70. Talbott S., Talbott J.. Effect of BETA 1, 3/1, 6 GLUCAN on Upper Respiratory Tract Infection Symptoms and Mood State in Marathon Athletes. J. Sports Sci. Med. (2009) 8 509-515. PMID: 24149590 71. Klomp A., Václavů L., Meerhoff G.F., Reneman L., Lucassen P.J.. Effects of Chronic Fluoxetine Treatment on Neurogenesis and Tryptophan Hydroxylase Expression in Adolescent and Adult Rats. PLoS ONE (2014) 9. DOI: 10.1371/journal.pone.0097603 72. Cowen D.S., Takase L.F., Fornal C.A., Jacobs B.L.. Age-dependent decline in hippocampal neurogenesis is not altered by chronic treatment with fluoxetine. Brain Res. (2008) 1228 14-19. DOI: 10.1016/j.brainres.2008.06.059 73. Navailles S., Hof P.R., Schmauss C.. Antidepressant drug-induced stimulation of mouse hippocampal neurogenesis is age-dependent and altered by early life stress. J. Comp. Neurol. (2008) 509 372-381. DOI: 10.1002/cne.21775 74. Marcussen A.B., Flagstad P., Kristjansen P.E.G., Johansen F.F., Englund U.. Increase in neurogenesis and behavioural benefit after chronic fluoxetine treatment in Wistar rats. Acta Neurol. Scand. (2007) 117 94-100. DOI: 10.1111/j.1600-0404.2007.00910.x 75. Hovorka M., Ewing D., Middlemas D.S.. Chronic SSRI Treatment, but Not Norepinephrine Reuptake Inhibitor Treatment, Increases Neurogenesis in Juvenile Rats. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23136919 76. Hodes G.E., Hill-Smith T.E., Lucki I.. Fluoxetine treatment induces dose dependent alterations in depression associated behavior and neural plasticity in female mice. Neurosci. Lett. (2010) 484 12-16. DOI: 10.1016/j.neulet.2010.07.084 77. Umu Ö.C.O., Rudi K., Diep D.B.. Modulation of the gut microbiota by prebiotic fibres and bacteriocins. Microb. Ecol. Health Dis. (2017) 28 1348886. DOI: 10.1080/16512235.2017.1348886 78. Ayaz M., Subhan F., Ahmed J., Khan A.-U., Ullah F., Ullah I., Ali G., Syed N.-I., Hussain S.. Sertraline enhances the activity of antimicrobial agents against pathogens of clinical relevance. J. Biol. Res. (2015) 22 4. DOI: 10.1186/s40709-015-0028-1 79. Coban A.Y., Cayci Y.T., Uludağ S.K., Durupinar B.. Investigation of antibacterial activity of sertraline. Mikrobiyoloji Bul. (2009) 43 651-656 80. Sun X., Wang D., Wei L., Ding L., Guo Y., Wang Z., Kong Y., Yang J., Sun L., Sun L.. Gut Microbiota and SCFAs Play Key Roles in QingFei Yin Recipe Anti-Streptococcal Pneumonia Effects. Front. Cell. Infect. Microbiol. (2021) 11 791466. DOI: 10.3389/fcimb.2021.791466 81. Zhong Q., Chen J.-J., Wang Y., Shao W.-H., Zhou C.-J., Xie P.. Differential Gut Microbiota Compositions Related With the Severity of Major Depressive Disorder. Front. Cell. Infect. Microbiol. (2022) 12 907239. DOI: 10.3389/fcimb.2022.907239 82. Zhou Y., Chen C., Yu H., Yang Z.. Fecal Microbiota Changes in Patients With Postpartum Depressive Disorder. Front. Cell. Infect. Microbiol. (2020) 10 567268. DOI: 10.3389/fcimb.2020.567268 83. Zafar H., Saier M.H.. Gut. Gut Microbes (2021) 13 8158. DOI: 10.1080/19490976.2020.1848158 84. Song X., Zhong L., Lyu N., Liu F., Li B., Hao Y., Xue Y., Li J., Feng Y., Ma Y.. Inulin Can Alleviate Me-tabolism Disorders in ob/ob Mice by Partially Restoring Leptin-related Pathways Mediated by Gut Microbiota. Genom. Proteom. Bioinform. (2019) 17 64-75. DOI: 10.1016/j.gpb.2019.03.001 85. Nakayama J., Yamamoto A., Palermo-Conde L.A., Higashi K., Sonomoto K., Tan J., Lee Y.-K.. Impact of Westernized Diet on Gut Microbiota in Children on Leyte Island. Front. Microbiol. (2017) 8 197. DOI: 10.3389/fmicb.2017.00197 86. De Filippo C., Cavalieri D., Di Paola M., Ramazzotti M., Poullet J.B., Massart S., Collini S., Pieraccini G., Lionetti P.. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc. Natl. Acad. Sci. USA (2010) 107 14691-14696. DOI: 10.1073/pnas.1005963107 87. Stilling R.M., Dinan T.G., Cryan J.F.. Microbial genes, brain &behaviour—Epigenetic regulation of the gut-brain axis. Genes Brain Behav. (2014) 13 69-86. DOI: 10.1111/gbb.12109 88. Donia M.S., Fischbach M.A.. Small molecules from the human microbiota. Science (2015) 349 1254766. DOI: 10.1126/science.1254766 89. Liang S., Wang T., Hu X., Luo J., Li W., Wu X., Duan Y., Jin F.. Administration of Lactobacillus helveticus NS8 improves behavioral, cognitive, and biochemical aberrations caused by chronic restraint stress. Neuroscience (2015) 310 561-577. DOI: 10.1016/j.neuroscience.2015.09.033
--- title: Inhibition of Sphingosine Kinase 2 Results in PARK2-Mediated Mitophagy and Induces Apoptosis in Multiple Myeloma authors: - Jian Wu - Shengjun Fan - Daniel Feinberg - Xiaobei Wang - Shaima Jabbar - Yubin Kang journal: Current Oncology year: 2023 pmcid: PMC10047154 doi: 10.3390/curroncol30030231 license: CC BY 4.0 --- # Inhibition of Sphingosine Kinase 2 Results in PARK2-Mediated Mitophagy and Induces Apoptosis in Multiple Myeloma ## Abstract Mitophagy plays an important role in maintaining mitochondrial homeostasis by clearing damaged mitochondria. Sphingosine kinase 2 (SK2), a type of sphingosine kinase, is an important metabolic enzyme involved in generating sphingosine-1-phosphate. Its expression level is elevated in many cancers and is associated with poor clinical outcomes. However, the relationship between SK2 and mitochondrial dysfunction remains unclear. We found that the genetic downregulation of SK2 or treatment with ABC294640, a specific inhibitor of SK2, induced mitophagy and apoptosis in multiple myeloma cell lines. We showed that mitophagy correlates with apoptosis induction and likely occurs through the SET/PP2AC/PARK2 pathway, where inhibiting PP2AC activity may rescue this process. Furthermore, we found that PP2AC and PARK2 form a complex, suggesting that they might regulate mitophagy through protein–protein interactions. Our study demonstrates the important role of SK2 in regulating mitophagy and provides new insights into the mechanism of mitophagy in multiple myeloma. ## 1. Introduction Multiple myeloma (MM) is a common plasma cell malignancy accounting for more than $17\%$ of hematological malignancies and $1.8\%$ of all cancers in the United States [1]. MM remains an incurable disease, and nearly all myeloma patients eventually relapse from conventional therapies [2]. Therefore, a better understanding of the cellular and molecular mechanisms underlying the pathogenesis of MM is essential for developing effective strategies to treat this devastating disorder and prevent its progression and relapse. Mitochondria are the main site of adenosine triphosphate (ATP) synthesis in mammalian cells and are critical for most biochemical and physiological processes such as cell growth, survival, and migration [3]. In the past decade, several studies have shown that mitochondria play crucial roles in regulating metabolism, calcium homeostasis, cellular aging, and neurocognitive functions [4,5,6]. Mitochondrial dysfunction is associated with many diseases, including diabetes, coronary artery disease, aging, and neurodegeneration. More recently, much attention has shifted to the importance of mitochondria in cancer development and progression, as well as to the association between mitophagy and tumor apoptosis. Mitochondria are directly involved in regulating cell death, including apoptosis [7,8]. B-cell lymphoma-2 (Bcl2) family member proteins interact with mitochondria by binding to voltage-dependent anion channels (VDACs). This binding accelerates channel opening and the release of cytochrome c [9]. Additionally, it regulates cancer progression and therapeutic resistance [10]. Myeloid leukemia cell differentiation protein-1 [MCL-1] [11] and Bcl-xl inhibit apoptosis by antagonizing pro-apoptotic members of the Bcl-2 family located at the outer mitochondrial membrane [12]. Moreover, MCL-1 and Bcl-xl regulate mitochondrial homeostasis and bioenergetics by preserving the integrity of the inner mitochondrial membrane and promoting the assembly of ATP-synthase oligomers in the electron transport chain [7]. Mitophagy, or mitochondrial autophagy, clears damaged mitochondria and determines mitochondrial quality and homeostasis [13,14]. However, the role of mitophagy in tumorigenesis remains unclear. Mitophagy is activated in transformed cells and is beneficial for tumor maintenance and progression [15,16]. However, excessive autophagy can act as a tumor-suppressive mechanism, possibly by initiating cell death [17,18]. The most likely explanation for these divergent findings is that mitophagy plays different roles in cancer pathogenesis depending on the stage of the disease, cell types, oncogenic drivers, and activation signal intensity [19,20,21]. Despite the important role of mitophagy in development and disease, the molecular mechanisms of mitophagy are not well understood. To date, many studies have identified various molecules that are involved in regulating mitophagy. PINK1 (PTEN-induced putative kinase 1)-PARK2 (parkin RBR E3 ubiquitin protein ligase)-dependent mitophagy is the most well-characterized pathway [22]. PINK1 is stabilized and accumulates in the outer membrane, where it binds and recruits PARK2 [23,24]. PARK2 then ubiquitinates and promotes the degradation of several outer mitochondrial membrane proteins, including the mitochondrial fusion proteins MFN1 (mitofusin 1) and MFN 2 [25,26]. Finally, phagophores target the mitochondria via specific receptors such as LC3B. Sequestered mitochondria are then degraded by fusing with lysosomes [27]. Sphingosine kinase 2 (SK2), an enzyme that catalyzes the formation of bioactive lipid sphingosine 1-phosphate (S1P), has recently been identified as a viable target for therapeutic intervention in MM [28,29]. High SK2 expression is involved in various biological processes, including cell growth, survival, and disease pathogenesis [30,31]. Moreover, SK2 expression levels were increased in bone marrow CD138+ myeloma cells from patients [32,33,34]. ABC294640, an SK2-selective inhibitor, induces caspase 3-mediated apoptosis and inhibits proliferation in MM cells [34]. The role and molecular mechanisms of SK2 in regulating mitophagy in MM are unknown. In this study, we determined the role of SK2 in the mitophagy of MM. Additionally, the underlying mechanisms were investigated using both pharmacological and genetic approaches. ## 2.1. Cell Lines The MM cell lines used in this study included MM1.R, MM1.S, NCIH929, and U266. The MM1.R (ATCC CRL-2975) and MM1.S (ATCC CRL-2974) cells were purchased from ATCC (Manassas, VA, USA). The NCIH929 (540-CRL-9068) and U266 (TIB-196) cells were purchased from the Duke Cell Culture Facility (CCF). All cell lines were cultured at 37 °C under $5\%$ CO2 in a RPMI1640 medium supplemented with $1\%$ (v/v) penicillin and $10\%$ FBS (Mediatech, Herndon, VA, USA). HEK293 cells were maintained in DMEM supplemented with $10\%$ FBS and a 1:100 antibiotic–antimycotic solution. ## 2.2. Antibodies and Reagents SK2 antibodies were obtained from Santa Cruz Biotechnology (Cat#: SC-398394). PARK2 antibody (#Proteintech, 14060-1-AP), PP2AC antibody (#Cell Signaling, 2038S), AKT antibody (#Cell Signaling, 9272S), SET antibody (#Abcam, ab97596), SET beta antibody, PTEN antibody (#Cell Signaling, 9552S), p-PTEN antibody (#Cell Signaling, 9557S), c-Myc antibody (#Cell Signaling, 9402), caspase 3 antibody, various caspase 9 antibody, LC3B antibody (#Cell Signaling, 2775S), BCL-2 antibody (#BD Bioscience, 610538), and MCL-1 antibody (#Abcam, ab32087) were purchased from commercial sources as indicated. ABC294640 (an SK2-specific inhibitor) was synthesized by Apogee Biotechnology Corp. The mitophagy inhibitor, bafilomycin, was obtained from Sigma-Aldrich (St. Louis, MO, USA). Okadaic acid (OA), a PP2A inhibitor, was purchased from Calbiochem (Gibbstown, NJ, USA). ## 2.3. Cell Proliferation Assay For the thiazolyl blue tetrazolium bromide (MTT) cell proliferation assay, myeloma cells were plated in triplicate in 96-well plates at a final volume of 100 μL containing 5 × 104 cells/well and concentrations of ABC294640. The cells were cultured at 37 °C in a $5\%$ CO2 incubator for various durations as indicated. At the time-points indicated, 20 μL of the combined MTS–PMS solution (5 mg/mL MTT) was added onto each well of the 96-well assay plate and incubated for 3–4 h at 37 °C in a $5\%$ CO2 incubator. The absorbance was measured using a ELISA plate reader at 490 nm. ## 2.4. Annexin V–Dye Apoptosis Assay The annexin V apoptosis assay was performed according to the manufacturer’s instructions (G-Biosciences, Cat#786-1548). Briefly, the cells were washed with 1× annexin V–PBS binding buffer and resuspended in 100 µL of annexin V–dye conjugate/propidium iodide staining solution (1 × 105 cells) for 15 min at room temperature in the dark. Then, 400 µL of 1× annexin V–PBS binding buffer was added to the cell suspension. Cells were analyzed on a flow cytometer using 488 nm excitation and 525 nm (FL1 channel) emission for annexin–FITC and 730 nm (FL2 channel) emission for propidium iodide. ## 2.5. Western Blot Analysis The MM cells were harvested, washed with phosphate buffer saline (PBS), and re-suspended in a RIPA lysis buffer (#Thermo Scientic, 89900, Rockford, IL, USA) with a cocktail protein inhibitor (#Thermo Scientific, 1862209) on ice for 15 min. The lysates were centrifuged at >10,000× g for 15 min to remove cell debris. Total protein was quantified using a Dc protein estimation kit (Bio-Rad) with bovine serum albumin (BSA) as a standard curve. Approximately 20 μg of protein was loaded and subjected to SDS-PAGE. The proteins were transferred onto polyvinylidene fluoride membranes. The membranes were blocked with $5\%$ BSA in Tris-buffered saline containing $0.05\%$ Tween 20. Primary antibodies were incubated with $5\%$ BSA in a buffer overnight at 4 °C with gentle rocking. The membranes were then probed with an HRP-conjugated secondary antibody and developed using the Pierce ECL substrate. ## 2.6. Lentivirus Production and Gene Transduction The production and gene transduction of lentiviruses encoding LC3B, SK2, and control vectors were performed as previously described [27,34]. PINK-specific shRNA, SK2-specific shRNA, and SK2-overexpressing plasmids were obtained from Addgene. PARK2-specific shRNAs were designed using the Block-iT RNAi Designer tool (Invitrogen) (accession number NC_000006.12) with target sequence CTTAGACTGTTTCCACTTATA. The lentiviruses were produced after the transient transfection of HEK 293T cells with an individual lentiviral vector along with packaging plasmids (VSV-G and psPax2) according to the manufacturer’s instructions. Supernatants containing viral particles were collected 48 h after transfection. The cells were transduced with lentiviruses by co-centrifugation at 3000× g for 3 h at 37 °C in the presence of 8 µg/mL polybrene. ## 2.7. Mitochondrial Membrane Potential (Δψm) Analysis Mitochondrial membrane potential (Δψm) was determined using a JC-1 fluorescent probe kit (Molecular Probe, Eugene, OR, USA) as previously described [15]. Briefly, the MM cells were suspended in 1 mL of a warm medium at a density of approximately 1 × 106 cells/mL. Ten μL of 200 µM JC-1 was added to the cell suspension, and the cells were incubated at 37 °C in $5\%$ CO2 for 15–30 min. The cells were then washed with 2 mL warm PBS, resuspended in 500 µL PBS, and analyzed on a flow cytometer with 488 nm excitation using an emission filter for the Alexa Fluor 488 dye. ## 2.8. Co-Immunoprecipitation (co-IP) Assay Co-immunoprecipitation of endogenous PP2AC and PARK2 was performed using the Pierce Immunoprecipitation Crosslink Magnetic IP/Co-IP kit (Thermo Scientific, #88805) according to the manufacturer’s instructions. Briefly, cell lysates were prepared in a lysis/wash buffer supplemented with 1× Complete Protease Inhibitor Cocktail (Roche, Mannheim, Germany) and 1× PhosSTOP Phosphatase Inhibitor Cocktail (Roche). Protein concentration was determined using Pierce BCA Protein Assay (Thermo Scientific). Four µg of the anti-PP2AC antibody (Cell Signaling, #2038S) or of IgG (serving as a control) was cross-linked to Protein A/G magnetic beads, incubated with 1 mg of protein lysate, and incubated overnight at 4 °C to prepare the immune complex. Then, 10 μL of the antigen sample/antibody mixture was removed as the input for subsequent western blotting assays, and the remaining sample was mixed with magnetic beads and incubated at room temperature for 2 h. The magnetic beads were collected, washed, and eluted in an elution buffer. Lastly, the supernatant was used for western blotting [35]. ## 2.9. PP2A Activity Assay The cells (1 × 106 cells) were treated with ABC294640 (30 µM) or transfected with lenti-SK2-specific shRNA (shSK2) or an SK2 overexpression plasmid for 48 h, and then lysed using a RIPA lysis buffer. PP2A activity was measured using a PP2A Immunoprecipitation Phosphatase Assay kit (17-313, Millipore, Temecula, CA, USA). This kit measured the activity of the C subunit of PP2A. Briefly, the protein lysates were incubated with the PP2A antibody at 4 °C with continuous rotation for 2 h. Following the addition of assay buffers and a malachite green solution, the plate was read at an absorbance of 650 nm using a microplate reader (BioTek Instruments, Winooski, VT, USA). Phosphatase activity was determined using a standard curve. The experiments were repeated at least in triplicate, and phosphatase activity was reported by picomoles of phosphate per minute as the mean ± standard error. ## 2.10. Immunofluorescence Confocal Microscopy The U266 and MM1.R cells were attached to a glass slide coated with 10 mg/mL fibronectin (Sigma, catalog 341635) for 1 h at 37 °C. The cells were subsequently fixed with $4\%$ formaldehyde in PBS for 15 min at room temperature. After fixation, the slides were blocked with $10\%$ FBS in a cell culture medium and subsequently incubated overnight at 4 °C with PP2AC and PARK2 antibodies. The slides were subsequently washed three times with PBS and stained with an Alexa Fluor 594 goat anti-rabbit antibody (Thermo Fisher Scientific, R37117) for 1 h. After washing them three times with PBS, the slides were stained with DAPI (Cell Signaling, 4083) for 5 min and mounted with an antifade mounting medium (Vector, H-1000). Images were acquired using a confocal laser-scanning microscope (Leica SP5 inverted-confocal microscope). Sequential scanning of different channels was performed at a resolution of 1024 × 1024 pixels. The system was equipped with 63 × 1.1HC PLAPO CS2. ## 2.11. Seahorse Assay MM1.S cells were treated with ABC294640 for 24 h, harvested, and seeded into an XF cell culture microplate (XF24 Flux Pak; Seahorse Biosciences, North Billerica, MA, USA) at a density of 40,000 cells per well. The cells were treated sequentially with oligomycin (1 µM), FCCP (3 µM), and antimycin (2.5 µM) [36]. The oxygen consumption rate was then recorded using an XF24 extracellular flux analyzer (Seahorse Biosciences). ## 2.12. Transmission Electron Microscopy (TEM) Assay The MM cells receiving different treatments were resuspended, washed with a HBSS buffer three times at room temperature, and fixed with a TEM fixative (10 mL $20\%$ formaldehyde, 4 mL $25\%$ glutaraldehyde, 5 mL 10× PBS, $0.01\%$ malachite green, and 31 mL distilled water) for at least 2 h at room temperature or overnight at 37 °C. The fixative was removed, and the samples were washed with PBS, post-fixed with OSO4 for 1 h, blocked with $1\%$ uranyl acetate, dehydrated in ethanol, and flat-embedded in Araldite 502 (Electron Microscopy Sciences, Hatfield, PA, USA). A 60 nm En face section was cut and stained with uranyl acetate and lead citrate using standard methods. Stained grids were examined using a Philips CM-12 electron microscope (EFI, Hillsboro, OR, USA) [27]. ## 2.13. Statistical Analysis Each experiment was performed in triplicates, and values were presented as mean ± standard error of the mean (SEM). Data were analyzed using a Student’s t-test. p values were designated as follows: * $p \leq 0.05$; $p \leq 0.01$; * $p \leq 0.001$; NS means not statistically significant. ## 3.1. SK2 Expression Is Upregulated in Abnormal Plasma Cells of Patients with MM, and SK2 Overexpression Is Associated with Poor Survival Given the accumulating evidence relating high SK2 expression to oncogenesis in other types of cancers, we first investigated the expression level of SK2 in patients with MM and the correlation between SK2 expression and clinical outcomes. Using Gene Expression Omnibus (GEO) datasets and the Genomic Scape database (http://www.genomicscape.com/, accessed on 1 June 2020), we observed a significant increase in the expression of SK2 in MM. This increase was observed in dataset GSE6477 when comparing relapsed MM or newly diagnosed MM to normal donors (Figure 1A). We also analyzed SK2 expression from the GSE13591 dataset and found that SK2 was overexpressed in MM (GSE13591, MM vs. ND $$p \leq 0.0384$$) (Figure 1B). These data were consistent with our previous data showing the upregulation of SK2, but not of SK1, in primary myeloma cells [34]. To investigate the prognostic significance of SK2 overexpression in MM development and progression, we evaluated SK2 gene expression in the APEX trial GEO microarray database (GSE9782) and correlated it with clinical outcomes such as progression-free survival (PFS) and overall survival (OS). High SK2 expression correlated with significantly shorter PFS (106 days vs. 140 days, $$p \leq 0.028$$) and OS (312 days vs. 628 days, $$p \leq 0.0011$$). Consistent with our previous findings, these data strongly suggest that SK2 plays a critical role in MM pathogenesis. ## 3.2. SK2 Inhibition Induces Mitophagy in MM Cells SK has two isoforms, SK1 and SK2. SK1 is mainly cytosolic, whereas SK2 is predominantly localized in the mitochondria, endoplasmic reticulum, and nucleus [37,38,39]. Recent studies from several laboratories, including ours, have demonstrated that PINK1-PARK2-dependent mitophagy plays an important role in the carcinogenesis of MM [27] and pancreatic cancer [40]. Therefore, we aimed to determine whether SK2 regulates mitophagy in MM cells. A loss of mitochondrial membrane potential (MMP) is a hallmark of mitophagy, representing an early event in mitochondrial damage coinciding with caspase activation. First, we determined the effect of SK2 inhibition on mitochondrial membrane depolarization. We treated the U266 and MM1.R cells with ABC294640 or genetically knocked down SK2 using shRNA and then measured mitochondrial membrane depolarization using JC-1 MitoProbe. With intact mitochondrial membrane potential, JC-1 remained a monomer (measured in the right upper quadrant, Figure 2A). However, upon mitochondrial membrane depolarization, JC-1 forms dimers and accumulates as aggregates (measured in the right lower quadrant, Figure 2A). Treatment with ABC294640 or genetically inhibiting SK2 induced mitochondrial membrane depolarization, as evidenced by the increased percentage of JC-1 aggregates in the right lower quadrant of the flow cytometer (Figure 2A and Supplementary Figure S2). Next, we used the Seahorse XF assay to measure and quantify the rate of ATP production and the mitochondrial system in the MM cells treated with ABC294640. ABC294640 decreased the oxygen consumption rate and extracellular acidification rate in a dose-dependent manner, and significantly reduced ATP production and spare respiratory capacity (Figure 2B), consistent with the mitophagy induction. Mitophagy is characterized by the fusion of mitochondria and lysosomes. Thus, the MM cells were treated with ABC294640, and transmission electron microscopy was performed to look for mitochondrial and lysosomal fusion. The ABC294640 treatment induced mitophagy, as shown by the large increase in the fusion of the mitochondria with the lysosomes (Figure 2C). Additionally, we monitored the interaction between the mitochondria and the lysosomes using Mito Tracker Red with Lysotracker Green staining, which tracks the colocalization of mitochondria and lysosomes in cells. After ABC294640 treatment, the lysosomal staining intensity and the number of lysosomes colocalized with mitochondria were strongly increased in the U266 and MM1.R cells (Figure 2D). Additionally, the protein expression levels of key molecules in mitophagy (PINK1, PARK2, and LC3B) were analyzed using a western blot (Figure 2E). PINK1, PARK2, and LC3B expression was upregulated, whereas SK2 levels were decreased in ABC294640-treated or lenti-shSK2 transduced MM cells. These data indicate that inhibiting SK2 expression induced MMP loss, mitochondria–lysosome fusion, and mitophagy. ## 3.3. Mitophagy Plays a Crucial Role in Mediating ABC294640 Induced Apoptosis in MM Cells We previously found that ABC294640 induces apoptosis and inhibits myeloma cell growth both in vitro and in vivo. To determine if mitophagy activation mediated the apoptotic effects of ABC294640, we used bafilomycin, a mitophagy inhibitor, to block mitophagy. Bafilomycin (0.5 nM) was added to the U266 and MM1.R cells treated with ABC294640 for 48 h (Figure 3A). Treatment with bafilomycin reversed the inhibitory effect of cell viability mediated by SK2 inhibition. Moreover, the MM cells exhibited a marked decrease in apoptosis when ABC294640 was combined with bafilomycin (Figure 3B). Apoptosis markers caspase 3, caspase 9, Bcl-2, Mcl-1, and c-Myc were also measured in ABC294640-treated U266 and MM1.R cells with or without bafilomycin (Figure 3C). We observed that SK2 inhibition via ABC294640 or lenti-shSK2 increased the expression of cleaved caspase 3 and caspase 9, and downregulated the expression of Bcl-2, Mcl-1, and c-Myc; however, SK2 inhibition was reversed by bafilomycin co-treatment. Taken together, these data suggest that blocking mitophagy through bafilomycin reverses the apoptotic cell death induced by SK2 inhibition. This shows that mitophagy induction mediates the apoptotic pathways caused by SK2 inhibition. ## 3.4. SK2 Inhibition Promotes Crosstalk between PP2AC and PARK2 Mediated by AKT/c-Myc/SET Signaling Next, we determined the molecular pathways of SK2-regulated mitophagy. The serine-threonine protein kinase AKT1 is an oncogene that selectively regulates PINK1-dependent mitophagy [41,42]. PTEN, a tumor suppressor [43] and dual functional phosphatase [44], is considered an important negative regulator of the PI3K/Akt pathway [45,46]. PTEN inhibits PINK1-PARK2-mediated mitophagy [47,48]. To determine the effects of SK2 on the expression levels of Akt and PTEN, we treated the MM cells with ABC294640 or genetically knocked down SK2 using shRNA (Figure 4A and Supplementary Figure S4). Additionally, SK2 was overexpressed in the MM cells (Figure 4B and Supplementary Figure S4). The pharmacological inhibition of SK2 or the genetic downregulation of SK2 resulted in a decrease in Akt expression and an increase in PTEN levels. In contrast, SK2 overexpression increased Akt expression and decreased PTEN expression. We then determined how SK2 regulates PP2A expression. The SET oncogene is a physiological inhibitor of PP2A, and there are two alternatively spliced SET isoforms: SET alpha and SET beta [49]. SET oncoproteins participate in cancer progression by affecting multiple cellular processes, including the control of the cell cycle, gene transcription, apoptosis, cell migration, and epigenetic regulation. SET contributes to tumorigenesis by forming an inhibitory protein complex with PP2A [50]. SET expression further exacerbates the effects of uncontrolled signaling by inhibiting the endogenous regulators of these pathways. We found that ABC294640 treatment or SK2 shRNA knockdown downregulated SET and SET beta expression (Figure 4A). Our study focused on protein phosphatase 2A (PP2A). PP2A is an important regulator of signal transduction pathways and is a tumor suppressor gene. PP2A negatively regulates multiple pro-growth/pro-survival signaling pathways associated with cancer progression, such as Akt and c-Myc [51,52]; SK2 was previously found to regulate PP2A activity. PP2A can deactivate Akt. The PP2A core enzyme comprises a 65kD scaffold subunit (known as the A or PR65 subunit) and a 36kD catalytic subunit (or C subunit) [53]. PP2AC phosphorylation leads to PP2A inactivation [54,55,56]. The pharmacologic inhibition of SK2 or the genetic downregulation of SK2 increased PP2A expression and the activation of PP2A (Figure 4A). In contrast, SK2 overexpression downregulated PP2A expression and inhibited PP2A (Figure 4B). We also measured PP2A phosphatase activity and found that SK2 inhibition increased PP2A phosphatase activity (Figure 4C). These data demonstrate the important role of SK2 in regulating PP2A activity. We determined how PP2A regulates mitophagy. PARK2 plays a key role in mitophagy. We performed a co-immunoprecipitation of PP2A and PARK2, and found that the PP2A and PARK2 formed a complex and interacted with each other (Figure 4D). We further fractionated the mitochondria and cytosol and found that the PP2A interacted with the PARK2 in the mitochondria (Figure 4D). Double immunofluorescence labeling was performed to confirm this finding (Figure 4E). Treatment with ABC294640 enhanced the interaction between the PP2AC and PARK2. ## 3.5. Inhibition of PP2AC Expression Blocks the Effect of ABC294640 We next sought to determine the role of PP2A in ABC294640-mediated effects. Okadaic acid (OA), a specific inhibitor of PP2A phosphatase activity at concentrations lower than 50 nM, was utilized in an amount of 20 nM [57]. The U266 and MM1.R cells were treated with ABC294640 for 48 h with or without OA, and cell viability was assessed. An MTS assay demonstrated that the co-treatment with OA markedly reversed the inhibitory effect of ABC294640 on cell viability (Figure 5A). To assess the combined effects of the ABC294640 and OA, the combination index (CI) value for each dose was calculated using the CompySyn software based on the Chou–Talalay method. The CI value was found to be greater than one, consistent with the antagonistic effect. The co-treatment with OA reduced the cytotoxic effect of ABC294640 at certain concentrations. ( Supplementary Figure S5). Furthermore, this reversal of the inhibitory effect on cell proliferation by OA was associated with a decrease in apoptosis. Blocking PP2AC decreased the number of apoptotic cells induced by the treatment with ABC294640 in the U266 and MM1.R cells (Figure 5B). Lastly, OA reduced the mitophagy induced by the ABC294640 treatment. Confocal immunofluorescence microscopy images of the U266 and MM1.R cells treated with the ABC294640 and/or OA were analyzed after staining the mitochondria with the MitoTracker dye (deep red) (Figure 5C). The treatment with ABC294640 increased the population of damaged-mitochondria-containing cells. The ABC294640-induced mitochondrial damage was then blocked by OA, as evidenced by the increase in Mito Tracker signals in the combination group. Similarly, the JC-1 results showed a similar trend: PP2AC inhibition by OA decreased mitochondrial membrane depolarization, in contrast to what happened when the MM cells were exposed to ABC294640 (Figure 5D). These data suggest that inhibiting PP2AC plays an important role in SK2-inhibition-induced PARK2-mediated mitophagy. ## 3.6. Knockdown of PINK1 or PARK2 Expression Attenuates ABC294640-Mediated Mitophagy To further confirm the role of the PINK1-PARK2 pathway in the ABC294640-mediated mitophagy, we knocked down PINK1 or PARK2 using specific shRNA in myeloma cell lines. The U266 and MM1.R cells lines were treated with 30 µM ABC294640 for 48 h with or without the knockdown of PINK1 or PARK2. The knockdown of PINK1 decreased the level of LC3B and PARK induced by the ABC294640 (Figure 6A, left panel). Similarly, the knockdown of PARK2 alleviated the upregulation of PINK1 and LC3B induced by the ABC294640 treatment (Figure 6A, right panel). Moreover, the JC-1 assay was performed to evaluate the mitochondrial membrane potential. PINK1-specific shRNA knockdown or PARK2-specific shRNA knockdown restored the change that occurred on the mitochondrial membrane, and attenuated the mitochondrial membrane depolarization induced by the ABC294640 (Figure 6B,C). These data demonstrated the important role of the PINK1-PARK2 pathway in ABC294640-induced mitophagy. ## 4. Discussion Our current study interrogated existing datasets and demonstrated that high SK2 expression occurred in patients with MM and that high SK2 levels were associated with poor clinical outcomes. Our findings are consistent with those of studies on solid tumors, in which the upregulation of SK2 were associated with tumor aggressiveness and poor outcomes [58,59]. Furthermore, the knockdown of SK2 with shSK2 inhibited MM cell proliferation and induced cell death. These data demonstrate the important role of SK2 in MM pathogenesis. Using the SK2-specific inhibitor, ABC294640, we showed that SK2 is a potential therapeutic target of treating MM. Additionally, we investigated the potential mechanism of ABC294640 in MM cell apoptosis. Our previous study showed that ABC294640 induces apoptosis in primary human CD138+ cells and MM cell lines [32]. Herein, we further demonstrate that ABC294640 activates mitophagy through the crosstalk between PP2AC and PARK2 which induces the apoptosis of MM cells. A PINK1- and PARK2-specific knockdown assay confirmed the role of this pathway in ABC294640-induced mitophagy. The inhibition of PP2AC by OA blocked the effects of the ABC294640. Our study provides direct evidence that SK2-mediated mitophagy plays a critical role in regulating myeloma apoptosis, and further elucidates the molecular mechanisms of SK2-mediated mitophagy. We also show that SK2 levels are a potential prognostic biomarker of MM. Taken together, these findings significantly advance our understanding of MM. ABC294640 is currently undergoing phase I/II clinical trials for myeloma (see trial NCT01410981) [32,60]. ABC294640 induces proteasome degradation and the downregulation of Mcl-1 and c-Myc but has no effect on Bcl-2 expression [32]. The combination of the ABC294640 and a Bcl-2 inhibitor (ABT-199) had a synergistic cytotoxic effect on the MM cells both in vitro and in vivo. Furthermore, ABC294640 showed good pharmacokinetics, oral bioavailability, and bioavailability [30]. Plasma concentrations can reach >200 µM without hematologic or major organ toxicity, approximately six- to seven-fold higher than the IC50 found in MM cell lines [61]. ABC294640 also had anticancer effects on various solid tumors [62]. ABC294640 causes cell cycle arrest in the S phase and increases the apoptosis rate in epithelial ovarian cancer cells [63]. An in vivo assay also showed the inhibitory effect of ABC294640 on tumor growth [64]. Our current study provides additional justification for testing ABC294640 in clinical settings with relapsed MM patients. The induction of apoptosis in tumor cells is an important goal in cancer chemotherapy [65]. Mechanistically, we found that ABC294640 induced apoptosis by stimulating mitophagy through the crosstalk between PP2AC and PARK2. Interestingly, the expression levels of phosphatase PP2AC have been reported as having an influence on the interplay between p38α and mTOR signaling [66]. In addition, ABC294640 downregulates pS6 expression [34,58]. This finding suggests that SK2 may also play a role in the mTOR signaling pathway and affect cell translation. Moreover, our previous study demonstrated that c-Myc and Mcl-1 expression levels were reduced by ABC294640 treatment via the proteasome degradation pathway in MM cell lines [32]. These data suggest that c-Myc and Mcl-1 may play major roles in mediating ABC294640-induced apoptosis. To further support this rationale, we performed the JC-1 assay to detect MMP in the MM cells and found that ABC294640 could induce mitochondrial membrane damage and cause early apoptosis. These data indicate that ABC294640 acts via the mitochondria-mediated apoptotic pathway. The relative roles of SK1 and SK2 in tumor biology have been of great interest to many investigators, and this has been a central issue in the selection of ABC294640 for our study. SK1 is a cytosolic protein that translocates to the cell membrane upon activation and is necessary for tumor progression [67]. While SK1 and SK2 share kinase homology and are $80\%$ similar, SK2 contains distinct, yet unidentified, localization and export signals, which differentiate its cellular localization and biological functions from those of SK1 [68,69]. SK2 also contains a pro-apoptotic BH3 domain that promotes apoptosis when overexpressed [70,71]. However, the exact role of SK2 in the pathogenesis of MM remains unclear. It remains to be determined whether the overexpression of SK2 in MM cells serves as a driving event to initiate the development of MM, or if it merely reflects phenotypic changes due to other oncogene aberrations. We are currently studying SK2-knockout mice to determine the incidence and severity of myeloma. These studies will help us to further define the role of SK2 in MM development. PINK1-PARK2-mediated mitophagy can result in either the recruitment and activation of the E3 ubiquitin ligase PARK2 and the downstream autophagy receptor SQSTM1, or in the PARK2-independent recruitment and activation of autophagy receptors like OPTN and CALCOCO2 [72]. PARK2 then localizes to the mitochondrial membrane and is phosphorylated to activate mitophagy [73]. Our data demonstrate the important role of PP2AC in ABC294640-mediated mitophagy and the interaction between PP2AC and PARK2. We found that the inhibition of PP2AC using okadaic acid rescued the mitophagy induced by the ABC294640. Our co-immunoprecipitation procedure clearly demonstrated that the PP2AC interacted with the PARK2 in the mitochondria (Figure 4D). The confocal microscope was less definitive, likely due to the high levels of PP2AC and PARK2 in other subcellular compartments. Despite this, our confocal microscope imaging showed the merge/colocalization of PP2AC and PARK2 (Figure 4E), further evidencing the interaction between PP2AC and PARK2. It is possible that the PP2A regulated the PARK2 phosphorylation level. However, the exact mechanism may be rather complicated, and an investigation of such mechanisms is beyond the scope of this study. In future work, we plan to conduct additional experiments to explore this mechanism in greater depth. ## 5. Conclusions In summary, our findings show that SK2 is aberrantly upregulated in MM cells and that the inhibition of SK2 through ABC294640 suppresses MM cells. Importantly, the inhibition of SK2 activates mitophagy through the crosstalk between PP2AC and PARK2 to induce apoptosis. Decreased PP2AC expression blocked the effects of the ABC294640. Our study provides important insights into the mechanisms of action of ABC294640 and could lead to the development of novel therapeutic interventions. ## References 1. Yang Y., Sun M., Yu Z., Liu J., Yan W., Liu Z., Wei M., Wang H.. Designing high affinity target-binding peptides to HLA-E: A key membrane antigen of multiple myeloma. Aging (2020) 12 20457-20470. DOI: 10.18632/aging.103858 2. Gimsing P., Hjertner O., Abildgaard N., Andersen N.F., Dahl T.G., Gregersen H., Klausen T.W., Mellqvist U.H., Linder O., Lindas R.. Salvage bortezomib-dexamethasone and high-dose melphalan (HDM) and autologous stem cell support (ASCT) in myeloma patients at first relapse after HDM with ASCT. A phase-2 trial. Bone Marrow Transplant. (2015) 50 1306-1311. DOI: 10.1038/bmt.2015.125 3. Sifuentes-Franco S., Pacheco-Moises F.P., Rodriguez-Carrizalez A.D., Miranda-Diaz A.G.. The Role of Oxidative Stress, Mitochondrial Function, and Autophagy in Diabetic Polyneuropathy. J. Diabetes Res. (2017) 2017 1673081. DOI: 10.1155/2017/1673081 4. Makrecka-Kuka M., Volska K., Antone U., Vilskersts R., Grinberga S., Bandere D., Liepinsh E., Dambrova M.. Trimethylamine N-oxide impairs pyruvate and fatty acid oxidation in cardiac mitochondria. Toxicol. Lett. (2017) 267 32-38. DOI: 10.1016/j.toxlet.2016.12.017 5. Spincemaille P., Cammue B.P., Thevissen K.. Sphingolipids and mitochondrial function, lessons learned from yeast. Microb. Cell (2014) 1 210-224. DOI: 10.15698/mic2014.07.156 6. de Moura M.B., dos Santos L.S., Van Houten B.. Mitochondrial dysfunction in neurodegenerative diseases and cancer. Environ. Mol. Mutagen. (2010) 51 391-405. DOI: 10.1002/em.20575 7. Grasso D., Zampieri L.X., Capeloa T., Van de Velde J.A., Sonveaux P.. Mitochondria in cancer. Cell Stress (2020) 4 114-146. DOI: 10.15698/cst2020.06.221 8. Arakawa S., Nakanomyo I., Kudo-Sakamoto Y., Akazawa H., Komuro I., Shimizu S.. Identification of a novel compound that inhibits both mitochondria-mediated necrosis and apoptosis. Biochem. Biophys. Res. Commun. (2015) 467 1006-1011. DOI: 10.1016/j.bbrc.2015.10.022 9. Shimizu S., Narita M., Tsujimoto Y.. Bcl-2 family proteins regulate the release of apoptogenic cytochrome c by the mitochondrial channel VDAC. Nature (1999) 399 483-487. DOI: 10.1038/20959 10. Perciavalle R.M., Stewart D.P., Koss B., Lynch J., Milasta S., Bathina M., Temirov J., Cleland M.M., Pelletier S., Schuetz J.D.. Anti-apoptotic MCL-1 localizes to the mitochondrial matrix and couples mitochondrial fusion to respiration. Nat. Cell Biol. (2012) 14 575-583. DOI: 10.1038/ncb2488 11. Wuilleme-Toumi S., Robillard N., Gomez P., Moreau P., Le Gouill S., Avet-Loiseau H., Harousseau J.L., Amiot M., Bataille R.. Mcl-1 is overexpressed in multiple myeloma and associated with relapse and shorter survival. Leukemia (2005) 19 1248-1252. DOI: 10.1038/sj.leu.2403784 12. Andersen J.L., Kornbluth S.. Mcl-1 rescues a glitch in the matrix. Nat. Cell Biol. (2012) 14 563-565. DOI: 10.1038/ncb2511 13. Kuang Y., Ma K., Zhou C., Ding P., Zhu Y., Chen Q., Xia B.. Structural basis for the phosphorylation of FUNDC1 LIR as a molecular switch of mitophagy. Autophagy (2016) 12 2363-2373. DOI: 10.1080/15548627.2016.1238552 14. Comita S., Femmino S., Thairi C., Alloatti G., Boengler K., Pagliaro P., Penna C.. Regulation of STAT3 and its role in cardioprotection by conditioning: Focus on non-genomic roles targeting mitochondrial function. Basic Res. Cardiol. (2021) 116 56. DOI: 10.1007/s00395-021-00898-0 15. Zheng Z., Fan S., Zheng J., Huang W., Gasparetto C., Chao N.J., Hu J., Kang Y.. Inhibition of thioredoxin activates mitophagy and overcomes adaptive bortezomib resistance in multiple myeloma. J. Hematol. Oncol. (2018) 11 29. DOI: 10.1186/s13045-018-0575-7 16. Feng X., Zhang Y., Zhang C., Lai X., Zhang Y., Wu J., Hu C., Shao L.. Nanomaterial-mediated autophagy: Coexisting hazard and health benefits in biomedicine. Part. Fibre Toxicol. (2020) 17 53. DOI: 10.1186/s12989-020-00372-0 17. Dong Q., Oh J.E., Yi J.K., Kim R.H., Shin K.H., Mitsuyasu R., Park N.H., Kang M.K.. Efavirenz induces autophagy and aberrant differentiation in normal human keratinocytes. Int. J. Mol. Med. (2013) 31 1305-1312. DOI: 10.3892/ijmm.2013.1327 18. Cordani M., Somoza A.. Targeting autophagy using metallic nanoparticles: A promising strategy for cancer treatment. Cell. Mol. Life Sci. (2019) 76 1215-1242. DOI: 10.1007/s00018-018-2973-y 19. White E.. Deconvoluting the context-dependent role for autophagy in cancer. Nat. Rev. Cancer (2012) 12 401-410. DOI: 10.1038/nrc3262 20. Song I.S., Kim H.K., Lee S.R., Jeong S.H., Kim N., Ko K.S., Rhee B.D., Han J.. Mitochondrial modulation decreases the bortezomib-resistance in multiple myeloma cells. Int. J. Cancer (2013) 133 1357-1367. DOI: 10.1002/ijc.28149 21. Jia L., Gopinathan G., Sukumar J.T., Gribben J.G.. Blocking autophagy prevents bortezomib-induced NF-kappaB activation by reducing I-kappaBalpha degradation in lymphoma cells. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0032584 22. Lin Q., Li S., Jiang N., Shao X., Zhang M., Jin H., Zhang Z., Shen J., Zhou Y., Zhou W.. PINK1-parkin pathway of mitophagy protects against contrast-induced acute kidney injury via decreasing mitochondrial ROS and NLRP3 inflammasome activation. Redox Biol. (2019) 26 101254. DOI: 10.1016/j.redox.2019.101254 23. Jin M., Liu X., Klionsky D.J.. SnapShot: Selective autophagy. Cell (2013) 152 368.e2. DOI: 10.1016/j.cell.2013.01.004 24. Patrick K.L., Bell S.L., Weindel C.G., Watson R.O.. Exploring the "Multiple-Hit Hypothesis" of Neurodegenerative Disease: Bacterial Infection Comes Up to Bat. Front. Cell. Infect. Microbiol. (2019) 9 138. DOI: 10.3389/fcimb.2019.00138 25. Williams J.A., Ding W.X.. Mechanisms, pathophysiological roles and methods for analyzing mitophagy—Recent insights. Biol. Chem. (2018) 399 147-178. DOI: 10.1515/hsz-2017-0228 26. Choubey V., Cagalinec M., Liiv J., Safiulina D., Hickey M.A., Kuum M., Liiv M., Anwar T., Eskelinen E.L., Kaasik A.. BECN1 is involved in the initiation of mitophagy: It facilitates PARK2 translocation to mitochondria. Autophagy (2014) 10 1105-1119. DOI: 10.4161/auto.28615 27. Fan S., Price T., Huang W., Plue M., Warren J., Sundaramoorthy P., Paul B., Feinberg D., MacIver N., Chao N.. PINK1-Dependent Mitophagy Regulates the Migration and Homing of Multiple Myeloma Cells via the MOB1B-Mediated Hippo-YAP/TAZ Pathway. Adv. Sci. (2020) 7 1900860. DOI: 10.1002/advs.201900860 28. Pyne N.J., Pyne S.. Sphingosine kinase 2 and multiple myeloma. Oncotarget (2017) 8 43596-43597. DOI: 10.18632/oncotarget.17420 29. Pyne N.J., McNaughton M., Boomkamp S., MacRitchie N., Evangelisti C., Martelli A.M., Jiang H.R., Ubhi S., Pyne S.. Role of sphingosine 1-phosphate receptors, sphingosine kinases and sphingosine in cancer and inflammation. Adv. Biol. Regul. (2016) 60 151-159. DOI: 10.1016/j.jbior.2015.09.001 30. Britten C.D., Garrett-Mayer E., Chin S.H., Shirai K., Ogretmen B., Bentz T.A., Brisendine A., Anderton K., Cusack S.L., Maines L.W.. A Phase I Study of ABC294640, a First-in-Class Sphingosine Kinase-2 Inhibitor, in Patients with Advanced Solid Tumors. Clin. Cancer Res. (2017) 23 4642-4650. DOI: 10.1158/1078-0432.CCR-16-2363 31. Gangoiti P., Granado M.H., Alonso A., Goni F.M., Gomez-Munoz A.. Implication of ceramide, ceramide 1-phosphate and sphingosine 1-phosphate in tumorigenesis. Transl. Oncogenomics (2008) 3 81-98. PMID: 21566746 32. Sundaramoorthy P., Gasparetto C., Kang Y.. The combination of a sphingosine kinase 2 inhibitor (ABC294640) and a Bcl-2 inhibitor (ABT-199) displays synergistic anti-myeloma effects in myeloma cells without a t(11;14) translocation. Cancer Med. (2018) 7 3257-3268. DOI: 10.1002/cam4.1543 33. Wallington-Beddoe C.T., Bennett M.K., Vandyke K., Davies L., Zebol J.R., Moretti P.A.B., Pitman M.R., Hewett D.R., Zannettino A.C.W., Pitson S.M.. Sphingosine kinase 2 inhibition synergises with bortezomib to target myeloma by enhancing endoplasmic reticulum stress. Oncotarget (2017) 8 43602-43616. DOI: 10.18632/oncotarget.17115 34. Venkata J.K., An N., Stuart R., Costa L.J., Cai H., Coker W., Song J.H., Gibbs K., Matson T., Garrett-Mayer E.. Inhibition of sphingosine kinase 2 downregulates the expression of c-Myc and Mcl-1 and induces apoptosis in multiple myeloma. Blood (2014) 124 1915-1925. DOI: 10.1182/blood-2014-03-559385 35. Seiler K., Humbert M., Minder P., Mashimo I., Schlafli A.M., Krauer D., Federzoni E.A., Vu B., Moresco J.J., Yates J.R.. Hexokinase 3 enhances myeloid cell survival via non-glycolytic functions. Cell Death Dis. (2022) 13 448. DOI: 10.1038/s41419-022-04891-w 36. Wang M., Sips P., Khin E., Rotival M., Sun X., Ahmed R., Widjaja A.A., Schafer S., Yusoff P., Choksi P.K.. Wars2 is a determinant of angiogenesis. Nat. Commun. (2016) 7 12061. DOI: 10.1038/ncomms12061 37. Kunkel G.T., Maceyka M., Milstien S., Spiegel S.. Targeting the sphingosine-1-phosphate axis in cancer, inflammation and beyond. Nat. Rev. Drug Discov. (2013) 12 688-702. DOI: 10.1038/nrd4099 38. Pyne N.J., Adams D.R., Pyne S.. Sphingosine Kinase 2 in Autoimmune/Inflammatory Disease and the Development of Sphingosine Kinase 2 Inhibitors. Trends Pharmacol. Sci. (2017) 38 581-591. DOI: 10.1016/j.tips.2017.04.003 39. Song D.D., Zhou J.H., Sheng R.. Regulation and function of sphingosine kinase 2 in diseases. Histol. Histopathol. (2018) 33 433-445. PMID: 29057430 40. Li C., Zhang Y., Cheng X., Yuan H., Zhu S., Liu J., Wen Q., Xie Y., Liu J., Kroemer G.. PINK1 and PARK2 Suppress Pancreatic Tumorigenesis through Control of Mitochondrial Iron-Mediated Immunometabolism. Dev. Cell (2018) 46 441-455.e8. DOI: 10.1016/j.devcel.2018.07.012 41. Soutar M.P.M., Kempthorne L., Miyakawa S., Annuario E., Melandri D., Harley J., O’Sullivan G.A., Wray S., Hancock D.C., Cookson M.R.. AKT signalling selectively regulates PINK1 mitophagy in SHSY5Y cells and human iPSC-derived neurons. Sci. Rep. (2018) 8 8855. DOI: 10.1038/s41598-018-26949-6 42. Larson-Casey J.L., Deshane J.S., Ryan A.J., Thannickal V.J., Carter A.B.. Macrophage Akt1 Kinase-Mediated Mitophagy Modulates Apoptosis Resistance and Pulmonary Fibrosis. Immunity (2016) 44 582-596. DOI: 10.1016/j.immuni.2016.01.001 43. Li J., Yen C., Liaw D., Podsypanina K., Bose S., Wang S.I., Puc J., Miliaresis C., Rodgers L., McCombie R.. PTEN, a putative protein tyrosine phosphatase gene mutated in human brain, breast, and prostate cancer. Science (1997) 275 1943-1947. DOI: 10.1126/science.275.5308.1943 44. Tamura M., Gu J., Matsumoto K., Aota S., Parsons R., Yamada K.M.. Inhibition of cell migration, spreading, and focal adhesions by tumor suppressor PTEN. Science (1998) 280 1614-1617. DOI: 10.1126/science.280.5369.1614 45. Cantley L.C., Neel B.G.. New insights into tumor suppression: PTEN suppresses tumor formation by restraining the phosphoinositide 3-kinase/AKT pathway. Proc. Natl. Acad. Sci. USA (1999) 96 4240-4245. DOI: 10.1073/pnas.96.8.4240 46. Hyun T., Yam A., Pece S., Xie X., Zhang J., Miki T., Gutkind J.S., Li W.. Loss of PTEN expression leading to high Akt activation in human multiple myelomas. Blood (2000) 96 3560-3568. DOI: 10.1182/blood.V96.10.3560 47. Wang L., Cho Y.L., Tang Y., Wang J., Park J.E., Wu Y., Wang C., Tong Y., Chawla R., Zhang J.. PTEN-L is a novel protein phosphatase for ubiquitin dephosphorylation to inhibit PINK1-Parkin-mediated mitophagy. Cell Res. (2018) 28 787-802. DOI: 10.1038/s41422-018-0056-0 48. Li G., Yang J., Yang C., Zhu M., Jin Y., McNutt M.A., Yin Y.. PTENα regulates mitophagy and maintains mitochondrial quality control. Autophagy (2018) 14 1742-1760. DOI: 10.1080/15548627.2018.1489477 49. Brander D.M., Friedman D.R., Volkheimer A.D., Christensen D.J., Rassenti L.Z., Kipps T.J., Guadalupe E., Chen Y., Zhang D., Wang X.. SET alpha and SET beta mRNA isoforms in chronic lymphocytic leukaemia. Br. J. Haematol. (2019) 184 605-615. DOI: 10.1111/bjh.15677 50. Li M., Guo H., Damuni Z.. Purification and characterization of two potent heat-stable protein inhibitors of protein phosphatase 2A from bovine kidney. Biochemistry (1995) 34 1988-1996. DOI: 10.1021/bi00006a020 51. Arnold H.K., Sears R.C.. A tumor suppressor role for PP2A-B56alpha through negative regulation of c-Myc and other key oncoproteins. Cancer Metastasis Rev. (2008) 27 147-158. DOI: 10.1007/s10555-008-9128-9 52. Resjo S., Goransson O., Harndahl L., Zolnierowicz S., Manganiello V., Degerman E.. Protein phosphatase 2A is the main phosphatase involved in the regulation of protein kinase B in rat adipocytes. Cell. Signal. (2002) 14 231-238. DOI: 10.1016/S0898-6568(01)00238-8 53. Virshup D.M., Shenolikar S.. From promiscuity to precision: Protein phosphatases get a makeover. Mol. Cell (2009) 33 537-545. DOI: 10.1016/j.molcel.2009.02.015 54. Chen J., Martin B.L., Brautigan D.L.. Regulation of protein serine-threonine phosphatase type-2A by tyrosine phosphorylation. Science (1992) 257 1261-1264. DOI: 10.1126/science.1325671 55. Chen J., Parsons S., Brautigan D.L.. Tyrosine phosphorylation of protein phosphatase 2A in response to growth stimulation and v-src transformation of fibroblasts. J. Biol. Chem. (1994) 269 7957-7962. DOI: 10.1016/S0021-9258(17)37144-2 56. Damuni Z., Xiong H., Li M.. Autophosphorylation-activated protein kinase inactivates the protein tyrosine phosphatase activity of protein phosphatase 2A. FEBS Lett. (1994) 352 311-314. DOI: 10.1016/0014-5793(94)00981-3 57. Xie L., Liu C., Wang L., Gunawardena H.P., Yu Y., Du R., Taxman D.J., Dai P., Yan Z., Yu J.. Protein phosphatase 2A catalytic subunit alpha plays a MyD88-dependent, central role in the gene-specific regulation of endotoxin tolerance. Cell Rep. (2013) 3 678-688. DOI: 10.1016/j.celrep.2013.01.029 58. Schrecengost R.S., Keller S.N., Schiewer M.J., Knudsen K.E., Smith C.D.. Downregulation of Critical Oncogenes by the Selective SK2 Inhibitor ABC294640 Hinders Prostate Cancer Progression. Mol. Cancer Res. (2015) 13 1591-1601. DOI: 10.1158/1541-7786.MCR-14-0626 59. Voelkel-Johnson C., Norris J.S., White-Gilbertson S.. Interdiction of Sphingolipid Metabolism Revisited: Focus on Prostate Cancer. Adv. Cancer Res. (2018) 140 265-293. PMID: 30060812 60. Kang Y., Sundaramoorthy P., Gasparetto C., Feinberg D., Fan S., Long G., Sellars E., Garrett A., Tuchman S.A., Reeves B.N.. Phase I study of opaganib, an oral sphingosine kinase 2-specific inhibitor, in relapsed and/or refractory multiple myeloma. Ann. Hematol. (2023) 102 369-383. DOI: 10.1007/s00277-022-05056-7 61. Lewis C.S., Voelkel-Johnson C., Smith C.D.. Targeting Sphingosine Kinases for the Treatment of Cancer. Adv. Cancer Res. (2018) 140 295-325. PMID: 30060814 62. LeBlanc F.R., Pearson J.M., Tan S.F., Cheon H., Xing J.C., Dunton W., Feith D.J., Loughran T.P.. Sphingosine kinase-2 is overexpressed in large granular lymphocyte leukaemia and promotes survival through Mcl-1. Br. J. Haematol. (2020) 190 405-417. DOI: 10.1111/bjh.16530 63. Song K., Dai L., Long X., Cui X., Liu Y., Di W.. Sphingosine kinase 2 inhibitor ABC294640 displays anti-epithelial ovarian cancer activities in vitro and in vivo. OncoTargets Ther. (2019) 12 4437-4449. DOI: 10.2147/OTT.S208519 64. Nema R., Vishwakarma S., Agarwal R., Panday R.K., Kumar A.. Emerging role of sphingosine-1-phosphate signaling in head and neck squamous cell carcinoma. OncoTargets Ther. (2016) 9 3269-3280 65. Elmore S.. Apoptosis: A review of programmed cell death. Toxicol. Pathol. (2007) 35 495-516. DOI: 10.1080/01926230701320337 66. Canovas B., Nebreda A.R.. Diversity and versatility of p38 kinase signalling in health and disease. Nat. Rev. Mol. Cell Biol. (2021) 22 346-366. DOI: 10.1038/s41580-020-00322-w 67. Carroll B., Donaldson J.C., Obeid L.. Sphingolipids in the DNA damage response. Adv. Biol. Regul. (2015) 58 38-52. DOI: 10.1016/j.jbior.2014.11.001 68. Truman J.P., Garcia-Barros M., Obeid L.M., Hannun Y.A.. Evolving concepts in cancer therapy through targeting sphingolipid metabolism. Biochim. Biophys. Acta (2014) 1841 1174-1188. DOI: 10.1016/j.bbalip.2013.12.013 69. Pitman M.R., Davies L.T., Pitson S.M.. An Improved Isoform-Selective Assay for Sphingosine Kinase 1 Activity. Methods Mol. Biol. (2018) 1697 9-20. PMID: 28540558 70. Gao P., Peterson Y.K., Smith R.A., Smith C.D.. Characterization of isoenzyme-selective inhibitors of human sphingosine kinases. PLoS ONE (2012) 7. DOI: 10.1371/journal.pone.0044543 71. Giussani P., Tringali C., Riboni L., Viani P., Venerando B.. Sphingolipids: Key regulators of apoptosis and pivotal players in cancer drug resistance. Int. J. Mol. Sci. (2014) 15 4356-4392. DOI: 10.3390/ijms15034356 72. Leermakers P.A., Kneppers A.E.M., Schols A., Kelders M., de Theije C.C., Verdijk L.B., van Loon L.J.C., Langen R.C.J., Gosker H.R.. Skeletal muscle unloading results in increased mitophagy and decreased mitochondrial biogenesis regulation. Muscle Nerve (2019) 60 769-778. DOI: 10.1002/mus.26702 73. Yang J., Sun M., Cheng R., Tan H., Liu C., Chen R., Zhang J., Yang Y., Gao X., Huang L.. Pitavastatin activates mitophagy to protect EPC proliferation through a calcium-dependent CAMK1-PINK1 pathway in atherosclerotic mice. Commun. Biol. (2022) 5 124. DOI: 10.1038/s42003-022-03081-w
--- title: Identification of a miRNA Panel with a Potential Determinant Role in Patients Suffering from Periodontitis authors: - Oana Baru - Lajos Raduly - Cecilia Bica - Paul Chiroi - Liviuta Budisan - Nikolay Mehterov - Cristina Ciocan - Laura Ancuta Pop - Smaranda Buduru - Cornelia Braicu - Mandra Badea - Ioana Berindan-Neagoe journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10047163 doi: 10.3390/cimb45030145 license: CC BY 4.0 --- # Identification of a miRNA Panel with a Potential Determinant Role in Patients Suffering from Periodontitis ## Abstract In recent years, the role of microRNA (miRNA) in post-transcriptional gene regulation has advanced and supports strong evidence related to their important role in the regulation of a wide range of fundamental biological processes. Our study focuses on identifying specific alterations of miRNA patterns in periodontitis compared with healthy subjects. In the present study, we mapped the major miRNAs altered in patients with periodontitis ($$n = 3$$) compared with healthy subjects ($$n = 5$$), using microarray technology followed by a validation step by qRT-PCR and Ingenuity Pathways Analysis. Compared to healthy subjects, 159 differentially expressed miRNAs were identified among periodontitis patients, of which 89 were downregulated, and 70 were upregulated, considering a fold change of ±1.5 as the cut-off value and p ≤ 0.05. Key angiogenic miRNAs (miR-191-3p, miR-221-3p, miR-224-5p, miR-1228-3p) were further validated on a separate cohort of patients with periodontitis versus healthy controls by qRT-PCR, confirming the microarray data. Our findings indicate a periodontitis-specific miRNA expression pattern representing an essential issue for testing new potential diagnostic or prognostic biomarkers for periodontal disease. The identified miRNA profile in periodontal gingival tissue was linked to angiogenesis, with an important molecular mechanism that orchestrates cell fate. ## 1. Introduction MicroRNAs (miRNAs) are evolutionarily conserved small non-coding RNA molecules that are 18–25 nucleotides long. They play a crucial role in normal cellular physiology, ensuring fine modulation of gene expression at a post-transcriptional level. Moreover, they coordinate tissue repair in adults [1,2,3]. In the first years after their discovery, the main focus of the miRNA was on their role in homeostasis; recent studies have observed that the abnormal expression of miRNAs leads to dysregulation of cellular responses involved in the adaptive immune reactions related to chronic inflammatory disease [4,5]. The fact that miRNAs are small molecules responsible for targeting multiple genes that belong to the same regulatory network makes them ideal tools for drug delivery and tissue regeneration [6]. miRNA function is essential for periodontal health, and immunity might affect periodontal homeostasis [4,7,8,9]. The process of gingival injury is followed by immediate tissue regeneration and repair [10]. At the cellular level, tissue regeneration and repair are triggered by cellular differentiation, dedifferentiation, transdifferentiation and reprogramming at the injured area. Moreover, injury-inducible coding genes and the signaling networks involved at the site of tissue interruption are controlled by miRNAs [2,4,9,11]. In addition, the significant mechanisms of angiogenesis, proliferation, migration, and morphogenesis of endothelial cells are controlled by specific miRNAs in an endothelial-specific manner. miRNAs known to regulate angiogenesis in vivo are called angiomiRs [7,11,12]. Lately, the expression of miRNAs in tissues affected by periodontitis has been explored [13]. Periodontitis and peri-implantitis are two diseases widely spread among the adult population, known to be triggered by a pathogenic bacterium that alters the host’s immune and inflammatory response [14]. From a clinical point of view, these alterations are translated into inflammatory lesions of the connective tissue around teeth or implants, with progressive bone loss and increased probing depths with an accelerating pattern [15]. The ongoing homeostasis between the bone resorption matrix and the new bone formation matrix is altered [16,17]. Differences in gene and miRNA expression in parodontids can affect specific cellular processes [18,19]. The literature reveals that the expression of specific miRNAs differs between the damaged periodontal tissue and the healthy periodontium. For example, miR-142-3p and miR-146a could be conclusive markers for disease activity [8]. Although the precise mechanism that leads to different changes in miRNA levels is not fully understood, studies have proven that the expression of certain miRNAs, such as miR-146a and miR-146b, was significantly higher. In contrast, the expression of miR-155 was significantly lower in inflamed tissues than in healthy tissues [9]. Periodontal disease is fast becoming the most frequent oral cavity disease, and implant therapy has already become a standard treatment for edentulous patients. To reduce the possible complications of implant therapy, such as peri-implantitis, our study aimed to identify the miRNAs as biomarkers in patients suffering from periodontitis, with subsequent consequences for the success of implant therapy and regeneration techniques. This study aimed to identify the altered miRNA signature and its implication in regulating key molecular mechanisms related to periodontal disease and then to discuss key challenges in translating this knowledge to the clinic and to propose novel biomarkers for validation. Combining the understanding of miRNA biology with cutting-edge technologies for gene expression analysis, such as microarray and validation of the obtained data by qRT-PCR, would help establish new miRNAs with potential biomarker properties for parodontids disease. As criteria for qRT-PCR validation, miRNAs were selected from the top 25 upregulated and downregulated and, simultaneously, those involved in angiogenesis. ## 2. Materials and Methods Patients cohorts. This study included 57 patients, 29 females ($50.87\%$) and 28 males ($49.13\%$). We had frozen gingival tissue for all the patients in the study, for microarray study are presented in Table 1 and for validation study in Table 2. The latest staging and classification in the field were used for the periodontitis diagnosis, according to the data published in 2018 [20]. The present study was approved by the institutional ethics committee of Iuliu Hatieganu University of Medicine and Pharmacy (UMPh), no. 81, from 11 March 2019. RNA extraction. Fresh frozen tissue was used for RNA extraction using the classical phenol–chloroform method. Mainly, the tissue was homogenized in 800 µL TripleXtractor (Grisp, Portugal), and then, the sample was used for RNA extraction. First, the sample was treated with chloroform (160 µL), mixed well by vortex, incubated at room temperature (RT) for 5 min, and centrifuged for 20 min at 13,000 rpm and 4 °C. The transparent phase was transferred to a new 1.5 mL tube, and RNA was precipitated with 500 µL of isopropanol, mixed by tube inversion, and incubated for 15 min at RT, then centrifuged at 13,000 rpm and 4 °C for 15 min. The supernatant was removed, and the pellet was washed with 1 mL of $75\%$ ethanol and centrifuged for 5 min at 10,000 rpm and 4 °C. After removal of the ethanol, the pellet was left to air-dry for 10–15 min and then dissolved in 25 µL nuclease-free water. The obtained RNA was quantified using NanoDrop (Thermo Fischer) spectrophotometer. Gingival tissue miRNA microarray evaluation. To evaluate the gingival miRNA pattern in both periodontitis and healthy tissues, 100 ng of total RNA from each sample was hybridized using a microRNA Spike-In kit and miRNA Complete Labeling and Hyb Kit (Agilent technologies). The microarray slides were hybridized for 20 h at a temperature of 55 °C, and washed and scanned using an Agilent Microarray Scanner. An additional purification step was performed using Micro Bio-Spin 6 (Biorad, Mississauga, ON, Canada) spin columns, followed by a desiccation step in a vacuum centrifuge and resuspension using 18 μL of RNase-free water. Microarray bioinformatics analysis. Data analysis for each file was performed using the Agilent GeneSpring GX software. The obtained images were further processed using the Feature Extraction program from Agilent to convert them to tabular structures containing numeric values referring to the specific expression for each miRNA. After normalization, differential expression analysis was conducted using the “Filter on Volcano Plot” module, with moderated t test, a fold change cut-off of 1.5 and a p value < 0.05. The lists of differentially expressed miRNAs were exported from the software for subsequent analysis. A series of comparisons were then performed using Venn diagrams to identify the miRNAs involved in specific signaling pathways such as angiogenesis and epithelial-to-mesenchymal transition. For this, we previously searched and downloaded from the NCBI (The National Center for Biotechnology Information) website the lists of genes involved in the pathways mentioned above by searching the Gene module for “Homo sapiens and angiogenesis” and “Homo sapiens and epithelial to mesenchymal transition”. miRNA-mRNA network analysis. The Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Redwood City, CA, USA) analyzed miRNA upstream regulators, networks, and associated pathways. All altered miRNAs were integrated into networks and were algorithmically generated based on their connectivity and scores. The score is displayed as a numerical value considering the relevance of a particular network to the original list of transcripts. qRT-PCR validation of selected miRNA. To evaluate the expression of miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p, the obtained RNA was reverse transcribed using the TaqMan MicroRNA Transcription kit (Applied Biosystems) and TaqMan microRNA primer assay (ThermoFisher Scientific) for the selected miRNAs. RNU6 and RNU48 were used for data normalization. One µL of total RNA was mixed with 0.75 µL of 10X RT Buffer, 0.1 µL of RNase inhibitor, 0.075 µL dNTP, 0.1825 µL of each of the 20X miRNA RT primers, 4.52 µL of nuclease-free water, and 0.5 µL of MultiScribed RT enzyme. The mixture was incubated at 16 °C for 30 min, 42 °C for 30 min, 85 °C for 5 min and held at 4 °C. The obtained cDNA was diluted six times with nuclease-free water and then used in real-time PCR reaction. We made a mixture containing 5.03 µL of ready-to-use TaqMan Fast Advanced Master Mix (Applied Biosystems), 0.47 µL of TaqMan microRNA primer and 5.2 µL of cDNA for each of the miRNAs analyzed. Then, 5 µL of the ready mix was loaded into two individual wells of the PCR plate. The PCR program run on the Viia7 instrument was as follows: 1 cycle for 2 min at 50 °C, one cycle for 20 s at 95 °C and 40 cycles at 95 °C for 1 s and 60 °C for 20 s in FastMode. The obtained CT values were analyzed using the ΔΔCT method, and the obtained results were imported into GraphPad Prism software for graphical presentation. GraphPad Prism software performed the ROC (receiver operating characteristic curve) analysis. In contrast, multiROC curve analysis was performed using Combiroc web software (http://combiroc.eu/, accessed on 7 January 2023) to emphasize the diagnostic power [21] of miRNAs as biomarkers in parodontids. ## 3. Results miRNA profiling in periodontitis and healthy gingival tissue. miRNA microarray analysis using Agilent technology was used to uncover the miRNA expression profiles in samples from patients with periodontal disease ($$n = 3$$) and healthy controls ($$n = 5$$). The miRNAs with significant changes in expression level (fold change ±1.5, and p values < 0.05) are presented as a heatmap in Figure 1. Furthermore, 159 differentially expressed miRNAs were identified among patients with periodontitis compared with healthy subjects. Among them, 89 were downregulated, and 70 were upregulated (Table 3). By overlapping the altered miRNA signature with the miRNA related to angiogenesis and EMT (epithelial to mesenchymal transition) (downloaded from NCBI), we identified a panel of 8 miRNAs related to angiogenesis and EMT, 2 miRNAs associated with EMT, and 17 miRNAs related to angiogenesis from the downregulated miRNAs list in periodontitis. Five were related to angiogenesis from the overexpressed miRNA list (Figure 2). Network analysis by IPA. IPA knowledge base was used to connect networks resulting in large merged networks. Predicted targets for the altered miRNAs list were mapped to the corresponding target in the Ingenuity knowledge base. The molecular and cellular functions identified based on altered miRNA signature are presented in Table 4. Figure 3 presents the miRNA-mRNA network; in the case of network N1 (Figure 3A) being related to “Neurological Disease, Organismal Injury and Abnormalities, Psychological Disorders”, the core element is VEGF, interconnected with AKT, MAP2K$\frac{1}{2}$, SMAD$\frac{2}{3}$, TGFB and RAS. N2 (Figure 3B) is related to “Gene Expression, Organismal Injury and Abnormalities, Reproductive System Disease”, and the core elements of these networks are ALOX5 and PAX3-FOXO1. N3 (Figure 3C) is related to “Glomerular Injury, Inflammatory Disease, Inflammatory Response”, the core element of the network being AGO2. N4 is related to “Cellular Development, Cellular Movement, Protein Synthesis”, with the core genes represented by IGF1B, TGFB1 and PTEN; miR-221 downregulated in parodontids is directly interconnected with PTEN, GAS5 and MTIF. As modulated by miRNAs, these genes could be key signaling molecules in the network. The network function would change as a consequence, considering the high number of downregulated miRNAs interconnected with these genes. qRT-PCR validation of microarray data. To validate the differentially expressed miRNAs from the microarray experiment, two downregulated and two overexpressed from the top 25 up/downregulated miRNAs were selected. The selection was based on their role in angiogenesis and the lack of data in the literature for their participation in periodontal disease. On the base of the above-applied, miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p were chosen for validation by qRT-PCR, selected from the top 25 up-regulated and downregulated genes list and in the same time being related with the angiogenesis mechanism. The performed validation experiment confirmed the upregulation of miR-191-3p and miR-1228-3p and the downregulation of miR-221-3p and miR-224-5p in gingival periodontal tissue compared to normal gingiva tissue, a scenario observed in the microarray study (Figure 4A–D). When analyzing the ROC (receiver operating characteristic) curves for the selected miRNAs, we observed that all tested had an AUC (area under the curve) greater than 0.7 (Figure 4E–H). The highest AUC value (0.076) was observed for Combo I (miR-191-3p + miR-1228-3p) (Figure 5). Regarding the ROC curves for the combination of the tested biomarkers, we obtained only ROC curves for three of the four biomarkers tested, which passed the quality control criteria for specificity and selectivity of the CombiROC software. Table 5 presents the statistical values using Pearson correlation between the expressions of the tested targets. The values in red represent statistically significant correlations. A direct statistically significant correlation was observed between miR-221-3p and miR-224-5p and between miR-224-5p and miR-1228-3p. Complex biological processes modulated by miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p and their target genes. The main physical processes and target genes for miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p were identified using the DIANA-miRPath v3.0 interface. A heatmap created directly from the DIANA-miRPath v3.0 interface reveals the main biological pathways that the selected miRNAs target (Figure 6A). Thus, we identified several overlapping target genes related to hippo signaling between the selected miRNAs, as follows: for miR-191-3p, three target genes; for miR-221-3p, twelve target genes; and for miR-224-5p, sixteen target genes were observed. Only miR-1228-3 had no target gene overlapping with the others. These data could emphasize YWHAZ as a common target for the first three transcripts (Figure 6B). Given the intricacy of the regulation of processes touched by these transcripts, predicting the dynamics of gene expression in parodontids is challenging. Additional functional studies will be required to evaluate the relative contribution of individual players. ## 4. Discussion In the present study, we conducted a miRNA microarray analysis of gingival tissue for patients with periodontitis versus healthy tissue and further validated it in independent validation sets. We identified a panel of potential novel biomarker candidates with application in periodontal disease [8,13], with implications in dental implantation for these patients. The periodontal region is a highly dynamic microenvironment that undergoes continuous remodeling due to frequent tissue fitting mechanical stress and inflammatory conditions. miRNAs are considered promising candidates based on their better features, including high abundance, stability, ease of sampling, and importance as global cellular regulators. The evaluation of the expression differences in miRNA and miRNA-mRNA interactions is only available in a few studies on periodontal disease. They can target several genes and influence multiple regulatory networks [22], also sustained by IPA analysis. Understanding the expression variation may help us deduce the occurrence and development of this complex disease (Figure 3). Previously, an IPA analysis revealed that the altered miRNA pattern in parodontids in the Japanese population is associated with inflammatory disease, organismal injury, abnormalities, urological disease, and cancer [23], a fact confirmed by the present data. The literature presents VEGF as a potential molecular target in periodontitis [24]. This is the core element of N1 (Figure 3A), directly interconnected by miR-21 and miR-1. MiR-21 is upregulated in patients with periodontitis and in mice induced with a periodontitis field [14,17], confirmed by the present microarray data. The literature data present that the FOXO1 signaling axis can regulate periodontal bacteria–epithelial interactions, immune-inflammatory response, bone remodeling, and wound healing [25]. In N2 from Figure 3B, it emphasizes the interconnection of the altered miRNAs with PAX3-FOXO1. Several therapeutic approaches targeting PAX3-FOXO1 were developed [26], the axis that should be further considered and exploited for periodontal disease. miRNAs act as post-transcriptional gene suppressors through their association with argonaute 2 (AGO2), a vital member of the RNA induced silencing complex (RISC) [27]. Several studies have demonstrated miRNA interaction with AGO2 [27]. AGO2 is crucial for the biogenesis of miRNAs and functions of multiple mechanisms, including angiogenesis [28,29]. AGO2 is considered a marker of differentiating periodontitis. AGO2 is the core of N3, interconnected with several downregulated miRNAs in parodontids. Our study reveals that periodontal diseases are more complex than previously assumed, emphasizing the inhibition of an important number of angiogenic-related miRNAs, as shown in Figure 2. We identified five angiogenesis-related miRNAs; in the top 25 upregulated, there were only four (miR-150, miR-188, miR-191 and miR-1228). Among these transcripts, there are limited data related to the implication in periodontitis; one study presented miR-150, miR-223 and miR-200b as overexpressed and miR-379, miR-199a-5p and miR-214 as underexpressed in inflamed gingival tissues in a Japanese population [23]. miR-188-3p was demonstrated to suppress human periodontal ligament stem cell osteogenesis through upregulating LEP [30]. No other information is related to the implication of miR-191 and miR-1228 in parodontids. Only 3 of the top 25 downregulated miRNAs (miR-221, miR-224 and miR-540) were related to angiogenesis and EMT, with none of these transcripts known to be related to periodontitis. Our findings clearly emphasized the clinical utility of miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p when analyzed individually or when considered as the signature in their ability to efficiently distinguish periodontitis patients from controls. Moreover, as we analyzed samples from periodontitis regions, we could speculate that miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p expression is disease-specific. These findings suggest that periodontitis-specific miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p modulation is a new molecular tool for disease diagnosis, which can be validated through further studies on different and larger patient populations. One of the miRNAs identified in our study, miR-221-3p, appeared to be induced in mechanical force-induced osteoblastic/cementoblasts differentiation of human periodontal ligament cells, thus providing a direct link of the above-mentioned process with the development of periodontitis [31]. In addition, Qiao et al. reported that miR-224-5p is highly expressed in dental periodontal ligament cells compared to dental pulp stem cells. Further functional studies performed on miR-221-3p-depleted dental pulp stem cells showed impaired cell viability followed by apoptosis through Rac family small GTPase 1 (Rac1) direct targeting [32]. miR-1228 was previously described as an actor of osteoblastic cell differentiation, targeting BMP-2K (bone morphogenetic protein–2 induced kinase) by inhibiting protein translation [33]. Aravindraja et al. proved that differentially expressed miRNAs that target different pathways in periodontal disease are connected to bacterial invasion and host response. They mention that miR-191 is involved in ischemic stroke [34], which leads to the idea that since periodontal disease is proven to be related to heart disease [35], miR-191 should also be investigated more closely in the oral field, as we stated above. Hippo signaling is an important pathway involved that affects mineralized tissue homeostasis and remodeling [36,37]. Our study emphasizes the diverse target genes related to Hippo signaling to the validated miRNAs. This should be further investigated for deciphering complex regulatory pathways related to miR-191-3p, miR-221-3p and miR-224-5p. miR-191-3p is not presented in the literature as related to parodontids disease. miR-221-3p modulates apoptosis in periodontal ligament cells [38]. miR-221-3p and miR-222-3p inhibited osteogenic differentiation of BMSCs via the IGF-1/ERK pathway [38]. Downregulation of miR-224-5p may promote dental pulp stem cell proliferation and migration [39]. As future perspectives, an additional investigation should be considered to better understand the altered mechanisms, which will lead to better therapeutic strategies for parodontids, particularly when considering the limitation of our study related to the reduced number of cases used for the microarray and for the validation cohort by qRT-PCR. Our research is focused on miRNA analysis as a potential biomarker for periodontal disease, with subsequent benefits in clinical protocols, and it involves highly trained specialists and financial resources. We consider it useful to identify patients at risk by using chairside point-of-care diagnostic technologies (PoCT) such as the aMMP-8 (active matrix metalloproteinase-8) oral fluid test as a first step in selecting more proper patients who require deeper and more complex investigations before implant placement [40]. Future research is required to validate the mechanism of these transcripts to potentially benefit from the angiogenic-related miRNAs’ roles in implant regeneration therapy. ## 5. Conclusions Our findings indicate an altered pattern of miRNA in gingival tissue, which should be considered an essential issue in generating new prognostic or diagnostic biomarkers for periodontal diseases, such as miR-191-3p, miR-221-3p, miR-224-5p and miR-1228-3p. Understanding the functional roles of miRNAs in the pathogenesis of periodontitis is very important due to their strong potential as therapeutic targets in alveolar bone regeneration. miRNAs display a specific periodontal gingival tissue, indicating that alterations in angiogenesis and critical cellular signaling networks have important implications in periodontal homeostasis and disease. Therefore, it is crucial to elucidate further how these altered transcripts might play a role in periodontal homeostasis and illness, considering the complex molecular mechanisms regulated by the altered transcripts. ## References 1. Fang F., Zhang K., Chen Z., Wu B.. Noncoding RNAs: New insights into the odontogenic differentiation of dental tissue-derived mesenchymal stem cells. Stem Cell Res. Ther. (2019) 10 297. DOI: 10.1186/s13287-019-1411-x 2. Sen C.K., Ghatak S.. miRNA control of tissue repair and regeneration. Am. J. Pathol. (2015) 185 2629-2640. DOI: 10.1016/j.ajpath.2015.04.001 3. Sehic A., Tulek A., Khuu C., Nirvani M., Sand L.P., Utheim T.P.. Regulatory roles of microRNAs in human dental tissues. Gene (2017) 596 9-18. DOI: 10.1016/j.gene.2016.10.009 4. Aghiorghiesei O., Zanoaga O., Raduly L., Aghiorghiesei A.I., Chiroi P., Trif A., Buiga R., Budisan L., Lucaciu O., Pop L.A.. Dysregulation of miR-21-5p, miR-93-5p, miR-200c-3p and miR-205-5p in Oral Squamous Cell Carcinoma: A Potential Biomarkers Panel?. Curr. Issues Mol. Biol. (2022) 44 1754-1767. DOI: 10.3390/cimb44040121 5. Aghiorghiesei O., Zanoaga O., Nutu A., Braicu C., Campian R.S., Lucaciu O., Berindan Neagoe I.. The World of Oral Cancer and Its Risk Factors Viewed from the Aspect of MicroRNA Expression Patterns. Genes (2022) 13. DOI: 10.3390/genes13040594 6. Luan X., Zhou X., Trombetta-eSilva J., Francis M., Gaharwar A.K., Atsawasuwan P., Diekwisch T.G.H.. MicroRNAs and Periodontal Homeostasis. J. Dent. Res. (2017) 96 491-500. DOI: 10.1177/0022034516685711 7. Luan X., Zhou X., Fallah P., Pandya M., Lyu H., Foyle D., Burch D., Diekwisch T.G.H.. MicroRNAs: Harbingers and shapers of periodontal inflammation. Semin. Cell Dev. Biol. (2022) 124 85-98. DOI: 10.1016/j.semcdb.2021.05.030 8. He W., Yang Y., Cai L., Lei Q., Wang Z., Che X.. MicroRNA expression profiles in peri-miniscrew implant crevicular fluid in orthodontics: A pilot study. BMC Oral Health (2021) 21. DOI: 10.1186/s12903-021-02009-w 9. Osan C., Chira S., Nutu A.M., Braicu C., Baciut M., Korban S.S., Berindan-Neagoe I.. The Connection between MicroRNAs and Oral Cancer Pathogenesis: Emerging Biomarkers in Oral Cancer Management. Genes (2021) 12. DOI: 10.3390/genes12121989 10. Fraser D., Caton J., Benoit D.S.W.. Periodontal Wound Healing and Regeneration: Insights for Engineering New Therapeutic Approaches. Front. Dent. Med. (2022) 3 815810. DOI: 10.3389/fdmed.2022.815810 11. Sen C.K.. MicroRNAs as new maestro conducting the expanding symphony orchestra of regenerative and reparative medicine. Physiol. Genom. (2011) 43 517-520. DOI: 10.1152/physiolgenomics.00037.2011 12. Irimie A.I., Sonea L., Jurj A., Mehterov N., Zimta A.A., Budisan L., Braicu C., Berindan-Neagoe I.. Future trends and emerging issues for nanodelivery systems in oral and oropharyngeal cancer. Int. J. Nanomed. (2017) 12 4593-4606. DOI: 10.2147/IJN.S133219 13. Asa’ad F., Garaicoa-Pazmiño C., Dahlin C., Larsson L.. Expression of MicroRNAs in Periodontal and Peri-Implant Diseases: A Systematic Review and Meta-Analysis. Int. J. Mol. Sci. (2020) 21. DOI: 10.3390/ijms21114147 14. Heikkilä P., Niskanen L., But A., Sorsa T., Haukka J.. Oral health associated with incident diabetes but not other chronic diseases: A register-based cohort study. Front. Oral Health (2022) 3 956072. DOI: 10.3389/froh.2022.956072 15. Leppilahti J., Majuri M.L., Sorsa T., Hirvonen A., Piirilä P.. Associations Between Glutathione-S-Transferase Genotypes and Bronchial Hyperreactivity Patients with Di-isocyanate Induced Asthma. A Follow-Up Study. Front. Med. (2019) 6 220. DOI: 10.3389/fmed.2019.00220 16. Kikuchi T., Hayashi J.I., Mitani A.. Next-Generation Examination, Diagnosis, and Personalized Medicine in Periodontal Disease. J. Pers. Med. (2022) 12. DOI: 10.3390/jpm12101743 17. Elgezawi M., Haridy R., Almas K., Abdalla M.A., Omar O., Abuohashish H., Elembaby A., Christine Wölfle U., Siddiqui Y., Kaisarly D.. Matrix Metalloproteinases in Dental and Periodontal Tissues and Their Current Inhibitors: Developmental, Degradational and Pathological Aspects. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23168929 18. Sharma S., Eghbali M.. Influence of sex differences on microRNA gene regulation in disease. Biol. Sex Differ. (2014) 5 3. DOI: 10.1186/2042-6410-5-3 19. Guo L., Zhang Q., Ma X., Wang J., Liang T.. miRNA and mRNA expression analysis reveals potential sex-biased miRNA expression. Sci. Rep. (2017) 7 39812. DOI: 10.1038/srep39812 20. Tonetti M.S., Greenwell H., Kornman K.S.. Staging and grading of periodontitis: Framework and proposal of a new classification and case definition. J. Periodontol. (2018) 89 S159-S172. DOI: 10.1002/JPER.18-0006 21. Hsu M.-J., Chang Y.-C.I., Hsueh H.-M.. Biomarker selection for medical diagnosis using the partial area under the ROC curve. BMC Res. Notes (2014) 7. DOI: 10.1186/1756-0500-7-25 22. Santonocito S., Polizzi A., Palazzo G., Isola G.. The Emerging Role of microRNA in Periodontitis: Pathophysiology, Clinical Potential and Future Molecular Perspectives. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms22115456 23. Ogata Y., Matsui S., Kato A., Zhou L., Nakayama Y., Takai H.. MicroRNA expression in inflamed and noninflamed gingival tissues from Japanese patients. J. Oral Sci. (2014) 56 253-260. DOI: 10.2334/josnusd.56.253 24. Ren B., Feng Q., He S., Li Y., Fan J., Chai G., Liu L., Liu H., Yang C., Wang Y.. VEGF as a potential molecular target in periodontitis: A meta-analysis and microarray data validation. J. Inflamm. (2021) 18 18. DOI: 10.1186/s12950-021-00281-9 25. Ren L., Yang J., Wang J., Zhou X., Liu C.. The Roles of FOXO1 in Periodontal Homeostasis and Disease. J. Immunol. Res. (2021) 2021 5557095. DOI: 10.1155/2021/5557095 26. Nguyen T.H., Barr F.G.. Therapeutic Approaches Targeting PAX3-FOXO1 and Its Regulatory and Transcriptional Pathways in Rhabdomyosarcoma. Molecules (2018) 23. DOI: 10.3390/molecules23112798 27. Chu-Tan J.A., Cioanca A.V., Feng Z.-P., Wooff Y., Schumann U., Aggio-Bruce R., Patel H., Rutar M., Hannan K., Panov K.. Functional microRNA targetome undergoes degeneration-induced shift in the retina. Mol. Neurodegener. (2021) 16 60. DOI: 10.1186/s13024-021-00478-9 28. Annese T., Tamma R., De Giorgis M., Ribatti D.. microRNAs Biogenesis, Functions and Role in Tumor Angiogenesis. Front. Oncol. (2020) 10 581007. DOI: 10.3389/fonc.2020.581007 29. Landskroner-Eiger S., Moneke I., Sessa W.C.. miRNAs as modulators of angiogenesis. Cold Spring Harb. Perspect. Med. (2013) 3 a006643. DOI: 10.1101/cshperspect.a006643 30. Yang Y., Li Y., Yan S., Song Y., Hu J., Li W., Wang X., Zhang J.. Long noncoding RNA LINC00673 sponges miR-188-3p to suppresses osteogenesis of human periodontal ligament stem cells through upregulating LEP. Res. Sq. (2022). DOI: 10.21203/rs.3.rs-2296285/v1 31. Wu Y., Ou Y., Liao C., Liang S., Wang Y.. High-throughput sequencing analysis of the expression profile of microRNAs and target genes in mechanical force-induced osteoblastic/cementoblastic differentiation of human periodontal ligament cells. Am. J. Transl. Res. (2019) 11 3398-3411. PMID: 31312353 32. Qiao W., Li D., Shi Q., Wang H., Wang H., Guo J.. miR-224-5p protects dental pulp stem cells from apoptosis by targeting Rac1. Exp. Ther. Med. (2020) 19 9-18. DOI: 10.3892/etm.2019.8213 33. Lisse T.S., Adams J.S., Hewison M.. Vitamin D and microRNAs in bone. Crit. Rev. Eukaryot. Gene Expr. (2013) 23 195-214. DOI: 10.1615/CritRevEukaryotGeneExpr.2013007147 34. Aravindraja C., Vekariya K.M., Botello-Escalante R., Rahaman S.O., Chan E.K.L., Kesavalu L.. Specific microRNA Signature Kinetics in Porphyromonas gingivalis-Induced Periodontitis. Int. J. Mol. Sci. (2023) 24. DOI: 10.3390/ijms24032327 35. Humphrey L.L., Fu R., Buckley D.I., Freeman M., Helfand M.. Periodontal disease and coronary heart disease incidence: A systematic review and meta-analysis. J. Gen. Intern. Med. (2008) 23 2079-2086. DOI: 10.1007/s11606-008-0787-6 36. Pandya M., Gopinathan G., Tillberg C., Wang J., Luan X., Diekwisch T.G.H.. The Hippo Pathway Effectors YAP/TAZ Are Essential for Mineralized Tissue Homeostasis in the Alveolar Bone/Periodontal Complex. J. Dev. Biol. (2022) 10. DOI: 10.3390/jdb10010014 37. Chen J., Cheng J., Zhao C., Zhao B., Mi J., Li W.. The Hippo pathway: A renewed insight in the craniofacial diseases and hard tissue remodeling. Int. J. Biol. Sci. (2021) 17 4060-4072. DOI: 10.7150/ijbs.63305 38. Gan K., Dong G.H., Wang N., Zhu J.F.. miR-221-3p and miR-222-3p downregulation promoted osteogenic differentiation of bone marrow mesenchyme stem cells through IGF-1/ERK pathway under high glucose condition. Diabetes Res. Clin. Pract. (2020) 167 108121. DOI: 10.1016/j.diabres.2020.108121 39. Ke Z., Qiu Z., Xiao T., Zeng J., Zou L., Lin X., Hu X., Lin S., Lv H.. Downregulation of miR-224-5p Promotes Migration and Proliferation in Human Dental Pulp Stem Cells. Biomed. Res. Int. (2019) 2019 4759060. DOI: 10.1155/2019/4759060 40. Sorsa T., Nwhator S.O., Sakellari D., Grigoriadis A., Umeizudike K.A., Brandt E., Keskin M., Tervahartiala T., Pärnänen P., Gupta S.. aMMP-8 Oral Fluid PoC Test in Relation to Oral and Systemic Diseases. Front. Oral Health (2022) 3 897115. DOI: 10.3389/froh.2022.897115
--- title: 'Why We Should Look at Dinner Plates: Diet Changes in Cancer Patients' authors: - Katja Döring - Lara Wiechers - Jens Büntzel - Judith Büntzel journal: Current Oncology year: 2023 pmcid: PMC10047165 doi: 10.3390/curroncol30030205 license: CC BY 4.0 --- # Why We Should Look at Dinner Plates: Diet Changes in Cancer Patients ## Abstract Objective: Malnutrition is often underestimated in the context of cancer therapy: the dietary trends initiated by patients after diagnosis are usually neither known to nor evaluated by the medical staff. Here, we propose a combined screening instrument evaluating malnutrition and dietary trends. Methods: The validated screening tool NRS-2002 was combined with a four-item questionnaire assessing whether [1] patients preferred certain foods, [2] avoided certain foods, [3] used dietary supplements or followed a special diet since the time of cancer diagnosis. The screening tool was routinely used by cancer patients in the daily practice of three oncological departments. The presented analysis was performed retrospectively and anonymized. Results: Overall, 102 cancer patients undergoing systemic therapy (CP), 97 undergoing radiation therapy (RP), and 36 head–neck cancer patients (HNP) were screened. The CP cohort showed a higher rate of malnutrition ($50.00\%$) than the HNP ($28.13\%$) or RP ($26.80\%$) cohort. Overall, diet changes were observed in $33.63\%$ of all patients. Avoiding meat, stimulants, or hard and edgy food was often mentioned in free text answers, while patients reported a preference for fruit and vegetables. Nutritional supplements were used by $28.76\%$ of the patients. While dietary changes were common, only $6.64\%$ of the patients mentioned adhering to a specific cancer diet. Conclusion: Malnutrition is still underestimated nowadays. Diet trends, especially avoiding certain foods, are common in cancer patients, while adhering to a specific cancer diet is an exception. Diet trends should be assessed and addressed to avoid or aggravate malnutrition. ## 1. Introduction Malnutrition is a common problem in cancer patients, often based on unintentional weight loss due to inadequate nutrient intake or uptake [1]. Nutritional status of patients differs between cancer entities and is also influenced by side-effects of oncological therapy [2]. Approximately 15–$40\%$ suffer from malnutrition at the disease’s onset, and the prevalence further increases up to 40–$80\%$ in patients undergoing oncological therapy [3]. Malnutrition affects several aspects of cancer treatment and outcome. Malnourished patients show an increase in treatment toxicity and a worse overall survival compared to well-nourished patients undergoing the same treatment [2]. Nutritional problems leading (in the worst case) up to cancer cachexia should be viewed as a continuum, starting from initial signs and symptoms of anorexia to cachexia or even refractory cachexia [4]. It is well known that the efficacy and impact of nutritional interventions are related to the timing of support, with the best results obtained with early intervention or prehabilitation [5,6]. Accordingly, it is important to regularly screen cancer patients for malnutrition during the various phases of treatment and the disease. Nutritional status is fluid and changes overtime. Not only cancer entity, as mentioned above, but also tumor stage, treatment type and setting, and concomitant diseases influence the patient’s nutritional needs. This argues for the necessity to continuously assess nutritional status [2]. Several standardized tools have been established for malnutrition screening. The European Society for Clinical Nutrition and Metabolism suggests to use the Malnutrition Screening Tool (MUST), the Nutritional Risk Screening (NRS-2002), or the Mini Nutritional Assessment [7]. The scored Patient-Generated Subjective Global Assessment (PG SGA) tool offers a time-intensive instrument, which was validated for tumor patients [8]. Nevertheless, it is not been established in the clinical practice. So, the current ESPEN practical guidelines commonly endorse to regularly evaluate nutritional intake, weight changes, and BMI at the time of cancer diagnosis and argue for subsequent reevaluations during cancer treatment [9]. While cancer and oncological treatment may lead to malnutrition [2], we should also consider the patients themselves. A small cross-sectional study in Germany revealed that up to $70\%$ of the patients surveyed had or planned to change their diet. These dietary changes themselves may have also an effect on (mal)nutrition [10]. This argues to not only assess for malnutrition but also dietary changes. We recently proposed a short questionnaire to detect cancer diets [11]. We propose to combine the latter with the Appendix A NRS-2002 [12] and present here a cohort of 235 cancer patients assessed for both malnutrition and dietary changes. ## 2. Materials and Methods Patients suffering from a solid or hemato-oncology cancer were enrolled in a non-interventional, anonymous, cross-sectional, retrospective study. Patients were routinely screened at the time of admission to inpatient care as a part of the clinical (admission) routine. The following departments participated: Institute of Radiation Oncology and Radiation Therapy of the University Medical Center Göttingen, Department for Hematology and Medical Oncology of the University Medical Center Göttingen, and the Department of Otorhinolarnygology of the Südharz Hospital in Nordhausen between September 2021 and June 2022. The retrospective analysis of data was approved by the local ethics committee of the University Medical Center (approval number: $\frac{22}{6}$/22). All included patients were screened with the NRS2002 and a four-item questionnaire proposed to identify dietary changes in cancer patients [11,12]. The NRS-2002 is a validated tool which offers a pre-screen, evaluating general low nutritional intake, weight loss, low BMI, or disease severity. In case of a positive pre-screen, the actual screening combines a more detailed assessment of weight loss/nutritional intake and disease severity and age [10]. The four-item questionnaire includes the following questions: [1] “Do you dispense or avoid specific food?”, [ 2] “Do you prefer specific food?”, [ 3] “Do you take additional supplements?”, and [4] “Do you follow a specific diet strategy?”. Patients are asked to answer if changes in nutritional behavior occurred after getting the diagnosis of cancer [11]. “ Dietary changes” were defined as changes (avoidance/preference) in consuming specific foods, while a “specific cancer diet” was defined as the conscious decision to follow a specific dietary regimen. If patients answered with “yes” concerning questions of avoiding or preferring certain foods, such as additional supplements or specific diet strategies, free text answers were written down. The free text answers of the patients concerning nutritional changes, cancer diets, or nutritional supplements were retrieved from questionnaires and translated into English. Answers were further summarized in categories (e.g., “red meat” and “beef” were summarized as “meat”). Each answer was considered equally. Instead of pie diagrams, we chose to draw word clouds to depict recurring main topics of patients. Word clouds were drawn using the free online software https://www.wortwolken.com/ (accessed on 9 January 2023). The larger the words are presented in each word cloud, the more often this specific answer was given by patients. Additional data on patients’ gender, cancer entity, and age were evaluated. When available, albumin and C-reactive protein (CRP) data were evaluated to calculate the modified Glasgow prognosis score (mGPS) [13]. Data were analyzed using an Excel Spreadsheet (Excel 2013) and GraphPad Prism (GraphPad Software, Version 8.0). Cohorts of patients were both analyzed separately by cohort. All cohorts were pooled for entity-specific subgroup analysis to reach a sufficient sample size for statistical analysis. If patient numbers of a single entity were not sufficient for subgroup analysis, single entities were summarized as an entity group: hematological malignant (lymphoma, acute myeloid leukemia, multiple myeloma, myeloproliferative neoplasia, chronic lymphatic leukemia, myelodysplastic syndrome, and chronic monocytic leukemia), hematology benign (anemia, idiopathic thrombocytopenic purpura, and others), lung cancer (non-small cell lung cancer, small cell lung cancer, and others), head–neck cancer (larynx carcinoma, oropharynx carcinoma, hypopharynx carcinoma, nasopharynx carcinoma, and others), other gynecological cancers (cervical and vulva carcinoma), uroonclogy (prostate carcinoma, urothelial carcinoma, and renal carcinoma), and upper gastrointestinal tract (esophagus carcinoma and others). Due to the sample size, Fisher’s Exact test instead of Chi-Square test was chosen for analyzing independence of dichotomous parameters. Pearson’s r was applied for correlation analysis. Patients with a positive pre-screen (NRS2002) were included for correlation analysis. We a priori planned to test for correlation between the NRS-2002 score (numerical values) and the albumin, C-reactive protein, or mGPS. A p-value < 0.05 was considered significant for the statistical tests applied. ## 3.1. Malnutrition Is Common amongst Cancer Patients Overall, we were able to pool data from three cohorts: the radiation cohort comprised 97 patients; the hematology/medical oncology cohort, 102 patients; and head–neck cohort, 36 patients. A total of 235 patients were included for data analysis. Of those, 144 identified as male and 91 as female. Median age was 65.64 years (range 29.43–88.35 years). We also stratified for cancer entities (Table 1). Overall, more than $50\%$ of the patients ($\frac{131}{235}$) included showed positive results in pre-screening and subsequently underwent the main screening of the NRS2002 tool. Here, 86 patients suffering from malnutrition were identified. Taken together, we observed malnutrition in $36.60\%$ ($\frac{86}{235}$) of all patients. Subgroup analysis (Table 2) was possible for the following five groups: breast cancer, hematology (malignant), head–neck cancer, lung cancer, and urooncology. Malnutrition rates were: $11.76\%$ ($\frac{2}{17}$, breast cancer), $48.44\%$ ($\frac{31}{64}$, hematology-malign), $31.82\%$ ($\frac{14}{44}$, head neck cancer), $43.60\%$ ($\frac{17}{39}$, lung cancer), and $47.06\%$ ($\frac{8}{17}$, urooncology). Compared to the whole cohort, patients suffering from malignant hematological neoplasia showed significantly higher malnutrition rates (Fisher’s exact test, $$p \leq 0.033$$, Table 2). We observed a trend towards lower albumin levels in these patients (Pearson’s r = −0.218, $$p \leq 0.071$$), while higher CRP levels were associated with a higher NRS2002 score (Pearson’s $r = 0.223$, $$p \leq 0.064$$). No significant correlation was found between NRS2002 and mGPS (Pearson’s $r = 0.108$, $$p \leq 0.376$$). ## 3.2. Diet Changes Are Common amongst Cancer Patients, Not Specific Cancer Diets Only a minority of our patients followed a specific (cancer) diet: $6.64\%$ ($\frac{15}{226}$ patients). Two items of our cancer diet questionnaire assessed whether patients avoided or preferred certain foods after cancer was diagnosed. A total of $33.63\%$ ($\frac{76}{226}$) of the patients changed their nutritional behavior by either avoiding ($27.43\%$, $\frac{62}{226}$ patients) or preferring ($20.80\%$, $\frac{47}{226}$ patients) certain foods. Stratification showed that overall diet changes were similar between entities. However, patients with malignant hematological neoplasia significantly more often preferred certain foods (Fisher’s exact test, $$p \leq 0.017$$) than other cancer patients, and head–neck cancer patients tended to avoid specific foods more often (Fisher’s exact test, $$p \leq 0.089$$). Overall, most patients avoided meat products ($38.46\%$, $\frac{30}{78}$ answers), stimulants ($16.77\%$, $\frac{13}{78}$ answers), edgy or hard foods ($11.54\%$, $\frac{11}{78}$ answers), and sugar/carbohydrates ($11.54\%$, $\frac{11}{78}$ answers). Only a minority avoided milk products ($6.41\%$, $\frac{5}{78}$ answers), spicy/sour foods ($5.13\%$, $\frac{4}{78}$ answers), or others ($10.26\%$, $\frac{8}{78}$ answers). Amongst preferred food were fruits ($25.00\%$, $\frac{23}{92}$ answers), vegetables ($25.00\%$, $\frac{23}{92}$ answers), and carbohydrates ($16.30\%$, $\frac{15}{92}$ answers). Less commonly preferred foods were milk products ($6.52\%$, $\frac{6}{92}$ answers) and fish/meat products ($6.52\%$, $\frac{6}{92}$ answers); A total of $20.65\%$ of foods ($\frac{19}{92}$ answers) was subsumed as “other”. Overall, we observed a subgroup of patients with food preferences showing an adaption towards mucositis or dysphagia ($17.39\%$, $\frac{16}{92}$ answers, categorized as “dysphagia nutrition”). Supplements were used by $28.76\%$ ($\frac{65}{226}$) of the patients. Here, we observed significantly higher user rates amongst breast cancer patients ($70.59\%$, $\frac{12}{17}$ patients; Fisher’s exact test, $$p \leq 0.002$$). In contrast, head–neck cancer patients had the lowest user rates within our cohort ($11.90\%$, $\frac{5}{42}$ patients; Fisher’s exact test, $$p \leq 0.001$$). For data concerning different entities and cohorts, refer to Table 3 and Table 4. Most commonly used supplements were vitamins ($50.35\%$, $\frac{72}{143}$ answers) and micronutrients ($32.17\%$, $\frac{46}{143}$ answers). Amino acids/proteins ($3.50\%$, $\frac{5}{143}$ answers) and fatty acids ($2.80\%$, $\frac{4}{143}$ answers) were less commonly used. As vitamins and micronutrients were the most commonly used supplements, we differentiated the answers. Overall, patients used the following vitamins: vitamin D ($23.01\%$, $\frac{26}{113}$ answers), vitamin B (including vitamin B12, vitamin B6, “vitamin B complex” preparations; $14.16\%$, $\frac{16}{113}$ answers), vitamin C ($8.85\%$, $\frac{10}{113}$ answers), or others ($14.16\%$, $\frac{16}{113}$ answers). Concerning micronutrients, patients used magnesium ($18.58\%$, $\frac{21}{113}$ answers), selenium ($7.08\%$, $\frac{8}{113}$ answers), iron ($6.19\%$, $\frac{7}{113}$ answers), zinc ($4.42\%$, $\frac{5}{113}$ answers), or others ($3.54\%$, $\frac{4}{113}$ answers). The foods standing out as the most often preferred were “fruits”, “potatoes”, “mashed potatoes”, “soup”, or “no hard foods” (Figure 1). The latter three may hint towards dysphagia and adaptation towards side-effects in patients. Patients avoided mostly “meat” or “sausages and cold cuts”, “sugar”, “alcohol”, “bread”, and “hard foods” (Figure 2). Here again, our patients’ answers may hint towards dysphagia (hard foods) and a reduced protein uptake (less meat). Answers on what diet changes happen after being diagnosed with cancer may offer insight on the causes of malnutrition. Amongst the supplements, vitamin D, vitamin B12, vitamin C, and magnesium are the most often used substances (Figure 3). ## 4. Discussion Since 2006, the worldwide Nutrition Day is an established benchmark program analyzing the role of nutrition in clinical practice of the participating health care systems. In 2021, $11\%$ of the participating German hospitals did not screen for malnutrition, another $8\%$ used visual assessment, $30\%$ used BMI or weighing, and only $32\%$ used the application of a screening tool for malnutrition [14]. Units treating cancer patients mainly considered nutritional treatment “when the patients asked” or at a weight loss of >$10\%$. Less than the half of all cancer care units ($46\%$) routinely recognized the need for nutritional treatment [14]. These data stand in contrast to the current ESPEN recommendations to routinely screen cancer patients’ nutritional intake, weight changes, and BMI at disease onset and to subsequently reevaluate nutritional status during cancer treatment [9]. The ESPEN suggests several different validated screening tools to detect malnutrition or patients at risk of developing malnutrition such as the PG-SGA, NRS-2002, MUST, or, in the elderly, the Mini Nutritional Assessment (MNA) [4]. The MUST was developed with the aim of community use, keeping in mind that here serious confounders of the effects of undernutrition are relatively rare. In contrast, the NRS-2002 focuses on detecting undernutrition or the risk of developing undernutrition in an in-patient setting and therefore assesses not only the nutritional components, such as the MUST, but also the disease severity [15]. The PG-SGA has been previously validated for cancer patients [4]. The high diagnostic performance of the PG-SGA in cancer patients [16] comes unfortunately at the cost of the time required for screening. This makes the PG-SGA difficult to integrate into daily clinical routine. Considering the latter and that the NRS-2002 is predominantly used in German hospitals [14], we decided to apply this screening tool in our cross-sectional study. The Global Leadership Initiative on Malnutrition (GLIM) established the “GLIM criteria” to ensure a well-defined, common definition of malnutrition in 2016. Amongst these are non-volitional weight loss, low BMI, reduced muscle mass, decreased food intake or assimilation, and inflammation or disease burden [17]. The NRS-2002 is valuable tool to assess for a majority of these criteria. However, neither NRS-2002 nor the GLIM criteria are able to give information on dietary changes of patients, which may also have an impact on weight-loss, food intake, or assimilation. Similar information will be received by using the PG SGA as a shortened screening instrument. Adding four items [11] to our nutritional screening, asking whether patients avoided or preferred certain foods from the time of cancer diagnosis, used micronutrients, or even followed a specific (cancer) diet, adds valuable information to the patients’ history. We know that malnutrition affects up to $75\%$ of cancer patients. The prevalence and resulting variation is determined by cancer-related (type, stage, and treatment), demographic (age), and social factors [2]. Overall, we used our modified screening tool to survey 235 patients. Half of all the patients showed a positive pre-screen and were referred to in-depth screening. In total we observed a rate of $36.60\%$, which is within the range of the published data [2,18]. Similar to previous studies [18,19,20,21], we observed different rates of malnutrition between cancer entities from $11.76\%$ in breast cancer patients to $48.40\%$ in patients with hematological neoplasia. Difference in malnutrition rates between entities—ranging in our cohort from $11.76\%$ in breast cancer patients to $48.44\%$ in patients with hematological neoplasia—is not an uncommon phenomenon. Malnutrition rates of lung ($43.60\%$) and head–neck cancer ($31.82\%$) in our study were similar to the previous data [20,21,22,23,24]. For patients with hematological neoplasia, the malnutrition rate in our study was higher than the data found in the literature [18,19,24,25]. Our higher rate might be explained by the facts that the three studies screening inpatients [18,19,24] did not use a screening tool but information on BMI and weight loss to assess for malnutrition and that [25] applied the PG-SGA on a cohort of (fitter) out-patients. Overall, our data shows the high risk and prevalence of malnutrition in specific groups of patients. This is not surprising as cancer therapy differs between entities and show different patterns of side effects, be it locoregional impairments (e.g., due to radiation or surgery) or side effects of systemic therapy (e.g., anorexia, oral discomfort) [26,27]; these may influence the development of malnutrition. Further, we should consider that malnutrition could also be influenced by tumor-induced activation of inflammatory pathways, which then may cause anorexia, altered metabolism, and involuntary loss of lean and fat mass, eventually leading to cachexia [28,29,30]. Very recently mass spectrometric analysis of blood sera even showed an entity of specific metabolomics profile in patients with upper gastrointestinal cancer [31]. Similar to these results, our study supports the previously published concept that each cancer entity is characterized by an entity-specific risk of accompanying malnutrition [2]. Oncologists should consider this entity-specific risk as malnutrition and cachexia are strong prognostic markers for unfavorable clinical outcomes [32]. Overall, this demonstrates the necessity to pay special attention to vulnerable subgroups as early as possible. Our data argue for paying increased attention towards the prognostic impact of malnutrition and low muscle mass on those undergoing treatment of hematological neoplasia, lung, or head–neck cancer and on patients suffering from urooncological neoplasia. While activation of inflammation pathways especially characterizes cachexia [28,29,30], we did not observe a significant association of C-reactive protein with malnutrition. Similarly, albumin status did not correlate with the results of the NRS-2002. This result fits with the analysis of a recent review that argues to consider both laboratory parameters as potential prognostic markers for overall survival but not for malnutrition [33]. Laboratory parameters as well anthropometric data do not offer insights in dietetic behavior of cancer patients. However, the latter may have a direct influence on (developing) malnutrition. The usefulness or also the harm caused by adhering to a specific cancer diet is a controversy discussed in the literature [34,35,36]. However, the community is aware that patients show interest in specific cancer diets that diverges from the official dietary guidelines of, for example, the American Cancer Society or the American Institute for Cancer Research/World Cancer Research Fund [35]. Despite the knowledge about patients’ interest in cancer diets, there is no tool to screen for adherence to a specific diet. Neither the NRS-2002 nor the MUST catch shifts in our patients’ dietary behavior. Both tools only register malnutrition, which might be a subsequent consequence of adhering to a specific diet. Therefore, our group proposed a short four-item questionnaire to assess for both specific cancer diets and dietary changes/intake of nutritional supplements [11]. Data on users of specific cancer diets are, to our knowledge, not available. Zick et al. proposed that the rate of patients using complementary and alternative medicine (CAM) might correspond with the rate of patients adhering to specific dietary regimen [35]. As 40–$90\%$ of cancer patients use CAM [37,38,39,40], one would expect high rates of patients following a specific diet. However, only $6.64\%$ of our cohort adhered to a specific diet. This low number might be explained by [1] the popularity of cancer diets is largely overestimated and [2] the rate might have been higher if our cohort had included more breast cancer patients, who are known to be more interested in CAM [39,41,42]. Summarizing, the low rate of patients using specific cancer diets argues for analyzing diet changes. Those are subtler, but may also have an impact on nutritional status (e.g., on protein intake if patients avoid meat without replacing proteins through alternative sources). We know that cancer patients often change or plan to adapt their dietary behavior [10,43]. Studies also showed that dietary changes of patients are not necessarily in line with the official dietary recommendations [35,43], and this may be, therefore, potentially an underlying cause of developing malnutrition. Dietary changes involve both avoiding and preferring certain foods. While another smaller German study described that up to $70\%$ of surveyed out-patients changed or planned to change their diet [10], we observed that only a third of the patients in our cohort reported dietary changes. By comparing the data between [10] and our study, we may however appreciate common, recurring topics amongst patients: e.g., a higher intake of fruits and vegetables, eating less meat, or avoiding sugar or carbohydrates. Our survey also offers a new, previously underrated insight—a small group of patients declared avoiding hard or edgy food. This insofar is interesting as this uncovers that dietary changes are not always due to the motivations of benefitting health or actively contributing to therapy [10] but are also an adaptation to current needs or impairments (e.g., not eating hard foods when suffering from oral discomfort). Our data shows that the presence of specific cancer diets is overestimated and the pitfalls of dietary changes are underestimated. Screening for malnutrition using established tools like the NRS-2002 should be complemented by taking patients’ nutritional history assessing patterns of preference and avoidance. Our short four-item questionnaire offers a fast option for screening and recognizing patients that may require further counseling or nutritional intervention. Additionally, our fourth question on usage of nutritional supplements offers additional information to the medical staff and enables them to counsel patients, whether supplement intake can have negative effects on cancer therapy (e.g., vitamin E and radiation therapy in head–neck cancer [44]). Overall, our results confirm that patients develop an awareness of their own diet and adapt their dietary behavior. A needs-based, individualized assessment of the nutritional status is necessary for individual counseling and intervention. Unfortunately, the actual attention given to nutritional status is far from the standard required in oncology management [45]. ## Limitations We present a retrospective cohort study with 235 patients only, undergoing either systemic (chemo-) therapy or radiation therapy. All cancer entities were included. Subgroup analysis was only possible for five entities, and even here, especially the numbers for patients with breast cancer and urooncological neoplasia were relatively small, which may have led to over- or underestimating rates of malnutrition and also dietary trends in these subgroups. Data analysis was only descriptive; therefore, we did not consider adjusting for multiple testing. Our screening tool is reductive and gives only an input for detailed counselling. Impact factors as surgery, irradiated fields, or GI diseases were not asked. The screening tool does not substitute any individual assessment and counselling. Furthermore, the questionnaire was used for patients under therapy. So, we have no sufficient information for longtime trends in cancer survivors. Nutritional trends and specific diets would be interesting terms in follow-up investigations too. Subgroup analysis of entities gives us first impressions on which patient groups are more prone to changing their diet or using supplements. Data of subgroup analysis however should be interpreted carefully due to a, sometimes, small case number. ## 5. Conclusions The spectrum and extent of malnutrition in oncological patients varies depending on the cancer entity. The NRS-2002 is a good tool to recognize malnourished patients; however, dietary trends and changes are not considered by established screening tools. Our four-item questionnaire is able to detect such nutritional trends, showing that the importance and prevalence of patients adhering to specific cancer diets is overestimated and the relevance of dietary trends and their potential influence on malnutrition are underrated. Therefore, we argue for adding a short screening of dietary trends to the standardized screening for malnutrition, which then offers a common ground for personalized counseling and intervention. ## References 1. Arends J., Baracos V., Bertz H., Bozzetti F., Calder P.C., Deutz N.E.P., Erickson N., Laviano A., Lisanti M.P., Lobo D.N.. ESPEN expert group recommendations for action against cancer-related malnutrition. Clin. Nutr. (2017.0) 36 1187-1196. DOI: 10.1016/j.clnu.2017.06.017 2. Bossi P., Delrio P., Mascheroni A., Zanetti M.. The Spectrum of Malnutrition/Cachexia/Sarcopenia in Oncology According to Different Cancer Types and Settings: A Narrative Review. Nutrients (2021.0) 13. DOI: 10.3390/nu13061980 3. Ravasco P.. Nutrition in Cancer Patients. J. Clin. Med. (2019.0) 8. DOI: 10.3390/jcm8081211 4. Muscaritoli M., Arends J., Aapro M.. From guidelines to clinical practice: A roadmap for oncologists for nutrition therapy for cancer patients. Ther. Adv. Med. Oncol. (2019.0) 11 1758835919880084. DOI: 10.1177/1758835919880084 5. Kubrak C., Martin L., Gramlich L., Scrimger R., Jha N., Debenham B., Chua N., Walker J., Baracos V.E.. Prevalence and prognostic significance of malnutrition in patients with cancers of the head and neck. Clin. Nutr. (2019.0) 39 901-909. DOI: 10.1016/j.clnu.2019.03.030 6. Gillis C., Davies S.J., Carli F., Wischmeyer P.E., Wootton S.A., Jackson A.A., Riedel B., Marino L.V., Levett D.Z.H., West M.A.. Current Landscape of Nutrition Within Prehabilitation Oncology Research: A Scoping Review. Front. Nutr. (2021.0) 8 644723. DOI: 10.3389/fnut.2021.644723 7. Miller J., Wells L., Nwulu U., Currow D., Johnson M.J., Skipworth R.J.E.. Validated screening tools for the assessment of cachexia, sarcopenia, and malnutrition: A systematic review. Am. J. Clin. Nutr. (2018.0) 108 1196-1208. DOI: 10.1093/ajcn/nqy244 8. Bauer J., Capra S., Ferguson M.. Use of the scored Patient-Generated Subjective Global Assessment (PG-SGA) as a nutrition assessment tool in patients with cancer. Eur. J. Clin. Nutr. (2002.0) 56 779-785. DOI: 10.1038/sj.ejcn.1601412 9. Muscaritoli M., Arends J., Bachmann P., Baracos V., Barthelemy N., Bertz H., Bozzetti F., Hütterer E., Isenring E., Kaasa S.. ESPEN practical guideline: Clinical Nutrition in cancer. Clin. Nutr. (2021.0) 40 2898-2913. DOI: 10.1016/j.clnu.2021.02.005 10. Schmid A., Mathies V., Buentzel J., Keinki C., Huebner J.. Diet Changes and Underlying Motives in Cancer Patients. Nutr. Cancer (2021.0) 74 2017-2028. DOI: 10.1080/01635581.2021.1986553 11. Büntzel J., Büntzel J.. Protocols Related to Nutritional Anamnesis in Head and Neck Cancer Patients. Basic Protocols in Foods and Nutrition (2022.0) 209-222 12. Kondrup J., Rasmussen H.H., Hamberg O., Stanga Z.. Nutritional risk screening (NRS 2002): A new method based on an analysis of controlled clinical trials. Clin. Nutr. (2003.0) 22 321-336. DOI: 10.1016/S0261-5614(02)00214-5 13. McMillan D.C., Crozier J.E.M., Canna K., Angerson W.J., McArdle C.S.. Evaluation of an inflammation-based prognostic score (GPS) in patients undergoing resection for colon and rectal cancer. Int. J. Color. Dis. (2007.0) 22 881-886. DOI: 10.1007/s00384-006-0259-6 14. Nutritionday Worldwide, Country Report Nutrition Day 2021 Germany 15. Kondrup J., Allison S.P., Elia M., Vellas B., Plauth M.. ESPEN guidelines for nutrition screening 2002. Clin. Nutr. (2003.0) 22 415-421. DOI: 10.1016/S0261-5614(03)00098-0 16. Ruan X., Nakyeyune R., Shao Y., Shen Y., Niu C., Zang Z., Miles T., Liu F.. Nutritional screening tools for adult cancer patients: A hierarchical Bayesian latent-class meta-analysis. Clin. Nutr. (2021.0) 40 1733-1743. DOI: 10.1016/j.clnu.2020.09.033 17. Cederholm T., Jensen G.L., Correia M.I.T.D., Gonzalez M.C., Fukushima R., Higashiguchi T., Baptista G., Barazzoni R., Blaauw R., Coats A.J.. GLIM criteria for the diagnosis of malnutrition—A consensus report from the global clinical nutrition community. J. Cachexia Sarcopenia Muscle (2019.0) 10 207-217. DOI: 10.1002/jcsm.12383 18. Li Z., Chen W., Li H.. Nutrition support in hospitalized cancer patients with malnutrition in China. Asia. Pac. J. Clin. Nutr. (2018.0) 27 1216-1224. DOI: 10.6133/apjcn.201811_27(6).0007 19. Pressoir M., Desné S., Berchery D., Rossignol G., Poiree B., Meslier M., Traversier S., Vittot M., Simon M., Gekiere J.P.. Prevalence, risk factors and clinical implications of malnutrition in French Comprehensive Cancer Centres. Br. J. Cancer (2010.0) 102 966-971. DOI: 10.1038/sj.bjc.6605578 20. Muscaritoli M., Lucia S., Farcomeni A., Lorusso V., Saracino V., Barone C., Plastino F., Gori S., Magarotto R., Carteni G.. Prevalence of malnutrition in patients at first medical oncology visit: The PreMiO study. Oncotarget (2017.0) 8 79884-79896. DOI: 10.18632/oncotarget.20168 21. Sanz E.Á., Siles M.G., Fernández L.R., Roldán R.V., Domínguez A.R., Abilés J.. Nutritional risk and malnutrition rates at diagnosis of cancer in patients treated in outpatient settings: Early intervention protocol. Nutrition (2019.0) 57 148-153. DOI: 10.1016/j.nut.2018.05.021 22. Bozzetti F., Mariani L., Vullo S.L., Amerio M.L., Biffi R., Caccialanza R., Capuano G., Correja I., Cozzaglio L.. The nutritional risk in oncology: A study of 1,453 cancer outpatients. Support. Care Cancer (2012.0) 20 1919-1928. DOI: 10.1007/s00520-012-1387-x 23. Büntzel J., Krauß T., Büntzel H., Küttner K., Fröhlich D., Oehler W., Glatzel M., Wackes M., Putziger J., Micke O.. Nutri-tional parameters for patients with head and neck cancer. Anticancer. Res. (2012.0) 32 2119-2123. PMID: 22593498 24. Hébuterne X., Lemarié E., Michallet M., De Montreuil C.B., Schneider S., Goldwasser F.. Prevalence of Malnutrition and Current Use of Nutrition Support in Patients With Cancer. J. Parenter. Enter. Nutr. (2014.0) 38 196-204. DOI: 10.1177/0148607113502674 25. Marshall K.M., Loeliger J., Nolte L., Kelaart A., Kiss N.K.. Prevalence of malnutrition and impact on clinical outcomes in cancer services: A comparison of two time points. Clin. Nutr. (2018.0) 38 644-651. DOI: 10.1016/j.clnu.2018.04.007 26. Büntzel J., Hübner J., Büntzel J.. Komplementärmedizinische Behandlungsansätze bei Inappetenz und Ösophagitis. Der Onkol. (2019.0) 25 1110-1117. DOI: 10.1007/s00761-019-00667-8 27. Büntzel J., Hübner J., Büntzel J.. Komplementärmedizinische Behandlungsansätze bei oraler Mukositis und Xerostomie. Der Onkol. (2019.0) 25 269-274. DOI: 10.1007/s00761-019-0521-5 28. Fearon K., Strasser F., Anker S.D., Bosaeus I., Bruera E., Fainsinger R.L., Jatoi A., Loprinzi C., MacDonald N., Mantovani G.. Definition and classification of cancer cachexia: An international consensus. Lancet Oncol. (2011.0) 12 489-495. DOI: 10.1016/S1470-2045(10)70218-7 29. Dev R.. Measuring cachexia—Diagnostic criteria. Ann. Palliat. Med. (2019.0) 8 24-32. DOI: 10.21037/apm.2018.08.07 30. Pin F., Barreto R., Couch M.E., Bonetto A., O’Connell T.M.. Cachexia induced by cancer and chemotherapy yield distinct perturbations to energy metabolism. J. Cachexia Sarcopenia Muscle (2019.0) 10 140-154. DOI: 10.1002/jcsm.12360 31. von Renesse J., von Bechtolsheim F., Jonas S., Seifert L., Alves T.C., Seifert A.M., Komorek F., Tritchkova G., Menschikowski M., Bork U.. Tumour catabolism independent of malnutrition and inflammation in upper GI cancer patients revealed by longitudinal metabolomics. J. Cachexia Sarcopenia Muscle (2022.0) 14 298-309. DOI: 10.1002/jcsm.13131 32. Le-Rademacher J., Lopez C., Wolfe E., Foster N.R., Mandrekar S.J., Wang X., Kumar R., Adjei A., Jatoi A.. Weight loss over time and survival: A landmark analysis of 1000+ prospectively treated and monitored lung cancer patients. J. Cachexia Sarcopenia Muscle (2020.0) 11 1501-1508. DOI: 10.1002/jcsm.12625 33. Bullock A.F., Greenley S.L., McKenzie G.A.G., Paton L.W., Johnson M.J.. Relationship between markers of malnutrition and clinical outcomes in older adults with cancer: Systematic review, narrative synthesis and meta-analysis. Eur. J. Clin. Nutr. (2020.0) 74 1519-1535. DOI: 10.1038/s41430-020-0629-0 34. Erickson N., Buchholz D., Hübner J.. Aktualisierte Stellungnahme ketogene und kohlenhydratarme Diäten veröffentlicht. Forum (2017.0) 32 429-430. DOI: 10.1007/s12312-017-0332-9 35. Zick S.M., Snyder D., Abrams D.I.. Pros and Cons of Dietary Strategies Popular Among Cancer Patients. Oncol. (Wil-List. Park) (2018.0) 32 542-547 36. Tajan M., Vousden K.H.. Dietary Approaches to Cancer Therapy. Cancer Cell (2020.0) 37 767-785. DOI: 10.1016/j.ccell.2020.04.005 37. Molassiotis A., Fernadez-Ortega P., Pud D., Ozden G., Scott J.A., Panteli V., Margulies A., Browall M., Magri M., Selvekerova S.. Use of complementary and alternative medicine in cancer patients: A European survey. Ann. Oncol. (2005.0) 16 655-663. DOI: 10.1093/annonc/mdi110 38. Micke O., Bruns F., Glatzel M., Schönekaes K., Micke P., Mücke R., Büntzel J.. Predictive factors for the use of complementary and alternative medicine (CAM) in radiation oncology. Eur. J. Integr. Med. (2009.0) 1 19-25. DOI: 10.1016/j.eujim.2009.02.001 39. Huebner J., Micke O., Muecke R., Buentzel J., Prott F.J., Kleeberg U., Senf B., Muenstedt K.. User rate of complementary and alternative medicine (CAM) of patients visiting a counseling facility for CAM of a German comprehensive cancer center. Anticancer. Res. (2014.0) 34 943-948. PMID: 24511037 40. Wortmann J.K., Bremer A., Eich H., Wortmann H.K., Schuster A., Fühner J., Büntzel J., Muecke R., Prott F., Huebner J.. Use of complementary and alternative medicine by patients with cancer: A cross-sectional study at different points of cancer care. Med. Oncol. (2016.0) 33 1-11. DOI: 10.1007/s12032-016-0790-4 41. Huebner J., Muenstedt K., Prott F.-J., Stoll C., Micke O., Buentzel J., Muecke R., Senf B.. Online Survey of Patients with Breast Cancer on Complementary and Alternative Medicine. Breast Care (2014.0) 9 60-63. DOI: 10.1159/000360381 42. Saghatchian M., Bihan C., Chenailler C., Mazouni C., Dauchy S., Delaloge S.. Exploring frontiers: Use of complementary and alternative medicine among patients with early-stage breast cancer. Breast (2014.0) 23 279-285. DOI: 10.1016/j.breast.2014.01.009 43. Tang J.W.C., Lam W.W.T., Ma A.S.Y., Law W.L., Wei R., Fielding R.. Dietary changes adopted by Chinese colorectal cancer patients: A qualitative study. Eur. J. Cancer Care (2019.0) 28 e13159. DOI: 10.1111/ecc.13159 44. Harvie M.. Nutritional Supplements and Cancer: Potential Benefits and Proven Harms. Am. Soc. Clin. Oncol. Educ. Book (2014.0) 34 e478-e486. DOI: 10.14694/EdBook_AM.2014.34.e478 45. Caccialanza R., Lobascio F., Cereda E., Aprile G., Farina G., Traclò F., Borioli V., Caraccia M., Turri A., De Lorenzo F.. Cancer-related malnutrition management: A survey among Italian Oncology Units and Patients’ Associations. Curr. Probl. Cancer (2020.0) 44 100554. DOI: 10.1016/j.currproblcancer.2020.100554
--- title: Comparison of MAFLD and NAFLD Characteristics in Children authors: - Yunfei Xing - Jiangao Fan - Hai-Jun Wang - Hui Wang journal: Children year: 2023 pmcid: PMC10047180 doi: 10.3390/children10030560 license: CC BY 4.0 --- # Comparison of MAFLD and NAFLD Characteristics in Children ## Abstract Background & aims: An international panel proposed a diagnostic framework for metabolic-associated fatty liver disease (MAFLD) in children. The aim was to compare the clinical features of MAFLD and nonalcoholic fatty liver disease (NAFLD) in children. Methods: The characteristic differences between NAFLD and MAFLD in children were compared with the National Health and Nutrition Examination Survey (NHANES) 2017–2018 in the U.S. and the Comprehensive Prevention Project for Overweight and Obese Adolescents (CPOOA) study in China. Results: In NHANES 2017–2018, regardless of which criteria were implemented, participants with hepatic steatosis were more likely to have higher BMI z-scores, a higher prevalence of hypertension or higher metabolic indices and higher non-invasive liver fibrosis scores (all $p \leq 0.05$). The cases diagnosed by those two definitions had a similarity of over $75\%$. More obese children were diagnosed with MAFLD than NAFLD ($p \leq 0.001$). However, approximately $19\%$ of children with NAFLD present with normal weight and fasting glucose levels and cannot be diagnosed with MAFLD. The CPOOA study excluded viral infected liver disease and certain kinds of congenital causes of liver steatosis patients, resulting in children with NAFLD being identical with MAFLD children. Conclusions: Most clinical features were similar between children with MAFLD and children with NAFLD, and more than $75\%$ of children with NAFLD can also be diagnosed with MAFLD. However, approximately $19\%$ of children with NAFLD cannot be categorized as MAFLD. Therefore, to gain greater benefits from renaming NAFLD to MAFLD in pediatrics, the prevalence of different causes of hepatic steatosis in children needs to be understood. ## 1. Introduction Nonalcoholic fatty liver disease (NAFLD) is a series of diseases characterized by excessive deposition of fat in hepatocytes, including simple fatty liver disease, nonalcoholic steatohepatitis and cirrhosis [1]. It is the most common cause of chronic liver disease in many countries and affects nearly $10\%$ of the general pediatric population [2,3]. The prevalence of NAFLD is closely associated with obesity. A meta-analysis of nine general population studies estimated that the prevalence of NAFLD in children aged 1–19 years was $2.3\%$ in normal-weight children, $12.5\%$ in overweight individuals and $36.1\%$ in obese individuals [4]. A recent meta-analysis conducted in Chinese children indicated that the prevalence of NAFLD was $6.3\%$ among the general pediatric population and $40.4\%$ in overweight/obese children [5]. However, the definition of NAFLD is problematic for children, as alcohol consumption is generally not an issue, since most countries have issued a prohibition for children under 18 years of age. Obesity and metabolic dysfunction are important clinical features of NAFLD; therefore, an international panel proposed the name ‘metabolic-associated fatty liver disease’ (MAFLD) to replace NAFLD in 2020 for adults [6]. Subsequently, an age-specific definition of pediatric MAFLD was released [7]. Compared to the definition of NAFLD, the criteria of MAFLD do not need to exclude other causes of steatosis, such as Wilson’s disease; rather, a requirement is the presence of abnormal metabolic features, including metabolic syndrome, prediabetes or overweight/obesity. Although some studies have investigated the prevalence of pediatric MAFLD [8,9], the differences in characteristics between MAFLD and NAFLD in children have not been thoroughly explored. Therefore, we used pediatric data from the 2017–2018 National Health and Nutrition Examination Survey (NHANES) of the U.S. and the Comprehensive Prevention Project for Overweight and Obese Adolescents (CPOOA) study of China to compare the clinical features between NAFLD and MAFLD. ## 2.1. Study Design and Population Selection The first study population was extracted from the NHANES 2017–2018, which was conducted by the National Center for Health Statistics to assess the health and nutritional status of adults and children in the United States. NHANES is a cross-sectional survey program combining interviews and physical examinations, and its complex, multistage, probability sampling design makes the population highly representative. More detailed information about NHANES can be found on the website (https://wwwn.cdc.gov/nchs/nhanes/, accessed on 31 July 2022). In total, 1051 participants aged 12–18 years were recruited in the NHANES 2017–2018. Participants without enough information to be categorized as NAFLD or MAFLD were removed. Specific exclusion criteria were lack of complete vibration-controlled transient elastography (VCTE) ($$n = 185$$) or complete absence of important laboratory covariates ($$n = 72$$): fasting plasma glucose (FPG), glycosylated hemoglobin (HbA1c), total cholesterol (TC), total triglyceride (TG) and high-density lipoprotein cholesterol (HDL-C); for children aged >15 years, C-reactive protein was also included. After exclusion, 794 participants were included in our study (Figure 1). The second study population was 1093 Chinese children aged 7–18 years who were extracted from the CPOOA study, as previously described [10]. The CPOOA study was carried out in 2 primary schools and 3 middle schools. Based on their medical history, participants with alcohol use or other liver-damaging factors besides NAFLD, including a history of diseases or drugs (including herbal medicines), infectious hepatic disease (hepatitis B virus [HBV] and hepatitis C virus [HCV]), autoimmune liver disease, Wilson’s disease and α1-antitrypsin (A1AT)-deficiency liver diseases, hepatic malignancies, biliary tract disease and any cardiovascular disease, were excluded from the CPOOA study [10]. Ultimately, ultrasound examination was performed on 1026 children with 1 trained doctor. The CPOOA study was approved by the ethics committee of the Peking University Health Science Center (IRB00001052-06084). All participants and their parents signed informed consent. ## 2.2.1. Definition of Hepatic Steatosis The diagnosis of hepatic steatosis in the NHANES 2017–2018 and the CPOOA was based on VCTE and abdominal ultrasonography, respectively. The criteria in NHANES 2017–2018 were based on two indicators of VCTE: median controlled attenuation parameter (CAP ≥ 248 dBm) or median liver stiffness measurement (≥7.4 kPa) [11]. In CPOOA, the diagnosis of hepatic steatosis requires at least two of the following conditions based on abdominal ultrasonography: diffusely increased echogenicity (‘bright’) liver with liver echogenicity greater than kidney or spleen, vascular blurring and deep attenuation of the ultrasound signal. ## 2.2.2. Definition of NAFLD Based on the presence of hepatic steatosis, the diagnosis of NAFLD in Chinese children excluded other liver-damaging factors identified in the medical history (self-report). In NHANES 2017–2018, alcohol consumption, congenital disorders or drug use cannot be excluded due to data limitations. In the present study, we only excluded participants who had positive hepatitis B/C infection or without hepatitis tests ($$n = 16$$, Figure 1). ## 2.2.3. Definition of MAFLD Children were categorized as having MAFLD based on the presence of hepatic steatosis and at least one of the following conditions: excess adiposity (overweight/obese or abdominal obesity), prediabetes or type 2 diabetes mellitus (T2DM), and metabolic abnormalities [7]. Overweight/obese were defined as having BMI z-scores >1 SD for children ≥ 5 and <10 years according to the WHO growth reference, and the diagnosis of abdominal obesity depended on waist circumference > 90th percentile [12]. Prediabetes was diagnosed by 5.6 ≤ FPG ≤ 6.9 mmol/L or HbA1c $5.7\%$ to $6.4\%$, while T2DM corresponded to FPG > 7.0 mmol/L or HbA1c > $6.5\%$. For children between 10 and 15 years old, metabolic abnormalities were defined as the presence of at least two of the following conditions: [1] systolic blood pressure >130 mm Hg or diastolic blood pressure > 85 mm Hg; [2] plasma TG > 150 mg/dL; [3] plasma HDL-C < 40 mg/dL; and [4] TG-to-HDL-C ratio > 2.25. For children > 15 years, the adult criteria of MAFLD were implemented [13]. ## 2.2.4. Definition of Non-Metabolic Dysfunction NAFLD (Non-MD-NAFLD) Non-MD-NAFLD referred to participants who met the definition of NAFLD but did not meet the definition of MAFLD. ## 2.2.5. Anthropometric and Biochemical Measurements In NHANES 2017–2018, anthropometric and biochemical measurements including waist circumference, BMI z-score, systolic blood pressure, diastolic blood pressure, platelet count, HbA1c, FPG, plasma HDL-C, blood urea nitrogen (BUN), serum creatinine, total bilirubin (TBIL), alanine transaminase (ALT), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), alkaline phosphatase (ALP), total protein, albumin, TC and TG were described in detail elsewhere [14]. Low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald formula [15]. In CPOOA, measurements of height, weight, TC, TG, HDL-C, LDL-C, FPG and ALT were performed according to standard protocols, details of which have been reported previously [16]. ## 2.3. Non-Invasive Scores Four non-invasive scores, including the AST-to-platelet ratio index (APRI) [17], NAFLD fibrosis score (NFS, based on age, BMI, diabetes, ALT, AST, platelet count and albumin) [18], Fibrosis-4 index (FIB-4, based on age, AST, ALT and platelet count) and BARD score (based on BMI, AST/ALT, diabetes score) [19] were used to assess liver fibrosis, with cutoff values of 1.5, −1.455, 1.3 and 2, respectively [20]. All of these calculations were performed only in the NHANES 2017–2018. ## 2.4. Statistical Analyses We performed all analyses using RStudio 2022.02.0. Given the multistage design of the NHANES study, we applied appropriate weights in all analyses based on NHANES data. Continuous variables were represented as weighted means ± SEs and categorical variables as weighted percentages. We used the design-adjusted Rao–Scott χ2 test and t-test to analyze the differences between categorical variables and continuous variables, respectively. For the analysis based on CPOOA data, continuous variables were represented as means ± SEs and categorical variables as percentages. Statistical significance was defined as a two-tailed p-value < 0.05. ## 3. Results Among 794 participants aged 12–18 years in NHANES 2017–2018, $51.2\%$ were males, with a mean age of 15.03 years and a mean BMI z-score of 1.15. The participants were mainly non-Hispanic white ($52.0\%$). T2DM was identified in $0.6\%$ of participants. The prevalence of MAFLD and NAFLD was $20.7\%$ and $26.4\%$, respectively. In comparison with children without NAFLD or MAFLD, participants with NAFLD or MAFLD had higher BMI z-scores, higher prevalence of hypertension, higher levels of platelets, higher metabolic indices (FPG, HOMA-IR, HOMA-β, HbA1c, unfavorable serum lipid levels and elevated liver enzymes) and higher non-invasive liver fibrosis scores (all $p \leq 0.05$) (Table 1). Additionally, most participants with NAFLD were categorized as MAFLD due to overweight/obesity and prediabetes or T2DM. In NHANES 2017–2018, the percentages of overweight/obesity in NAFLD and MAFLD children were $78.0\%$ and $98.0\%$, respectively (Figure 2). Prediabetes or T2DM is another dominant factor, accounting for $35.2\%$ and $30.7\%$ of children with MAFLD and NAFLD, respectively. Moreover, the proportions of MAFLD and NAFLD participants with at least one of the characteristics (overweight/obesity, prediabetes or T2DM, and metabolic dysregulation) were $100.0\%$ and $82.3\%$ respectively. Comparisons among NAFLD, MAFLD and non-MD-NAFLD are presented in Table 2. Participants with MAFLD had higher BMI z-scores and lower levels of ALT, TG and HDL-C than those with NAFLD ($p \leq 0.001$). The participants with non-MD-NAFLD had the lowest mean BMI z-scores (both $p \leq 0.001$), highest level of HDL-C (both $p \leq 0.001$), lowest levels of ALT, GGT, HOMA-IR and TG (all $p \leq 0.05$) and highest FIB-4 score (both $p \leq 0.05$) compared to the other two groups. In addition, 184 participants had both MAFLD and NAFLD (Figure 3). Forty-one participants met the definition of non-MD-NAFLD, accounting for $18.8\%$ of the participants with NAFLD. Three participants met the definition of MAFLD but not NAFLD, as they lacked information for the diagnosis of hepatitis B/C. In the CPOOA data, all NAFLD children fulfilled the definition of MAFLD. Therefore, children with NAFLD and children with MAFLD were identical populations (Table 3, Figure 4). All participants with NAFLD or MAFLD had at least one of the three characteristics: overweight/obese, prediabetes or T2DM, and metabolic dysregulation (Figure 2). ## 4. Discussion The present study is the first study to compare the characteristics of children with MAFLD and children with NAFLD in two different populations. Our study mainly found that MAFLD criteria can identify more obese children than NAFLD criteria. As we expected, in the population of overweight and obese children, MAFLD children were identical to NAFLD children when all infectious or congenital disorders related to liver steatosis were excluded, such as viral hepatitis and Wilson’s disease. Although the diagnosis of MAFLD differs from the criteria of NAFLD, the cases diagnosed by the two methods had a similarity over $75\%$. The reason for the high similarity is that both MAFLD and NAFLD are related to overweight/obesity. Previous meta-analyses showed that the prevalence of NAFLD was much higher in overweight/obese children, reaching $36.1\%$ globally and $40.4\%$ in Chinese children, respectively [4,5]. In the present study, the prevalence of MAFLD and NAFLD in overweight/obese children was $49.2\%$ and $49.9\%$ in NHANES 2017–2018, respectively, and both were $25.4\%$ in CPOOA, similar to the results of a contemporaneous study [21]. Notably, approximately $80\%$ of all adolescents with fatty liver disease were diagnosed with MAFLD, which is lower than the proportion reported in a recent study by Ciardullo and colleagues [22]. In that study, the percentage of MAFLD in a steatotic population was around $87.7\%$. A plausible reason might be that the subjects were collected from 2017 to 2020. Second, both MAFLD and NAFLD are related to hyperglycemia. Previous research indicated that hyperglycemia could drive the metabolic-related NAFLD and that genetic NAFLD could drive the metabolic-related hyperglycemia [23,24]. In a cross-sectional study across 12 centers in the U.S., $23.4\%$ of children with NAFLD had prediabetes, and participants with NAFLD who had T2DM were more likely to have metabolic steatohepatitis [25]. Our data indicated that participants with NAFLD or MAFLD had higher levels of FPG and more severe manifestations of insulin resistance, and 30–$40\%$ of children with NAFLD or MAFLD were hyperglycemic. Considering the higher prevalence of MAFLD and NAFLD in children with overweight/obesity and hyperglycemia, screening of liver function should be more widely promoted by the state and government in this population to prevent and control the development of fatty liver. A recent study [26] explored potential etiologies among 900 overweight/obese children with liver steatosis from two tertiary care centers in North America and found that 347 ($39\%$) children were biopsy-diagnosed with NAFLD and 19 ($2\%$) children had other causes, including thyroid disease, celiac disease, A1AT deficiency, hemochromatosis and Hodgkin’s lymphoma. No cases of viral hepatitis, Wilson’s disease or autoimmune hepatitis (AIH) were found [26]. Consistently, viral hepatitis was not found in children with MAFLD in the present study. Ioannou [27] found that the prevalence of HBV in the U.S. population aged 6 years and older and children aged 6–17 years in NHANES 1999– 2008 was $0.3\%$ and $0.03\%$. Intriguingly, the prevalence of viral infection with HBV was $0.3\%$ among U.S. adults, and none was detected in NHANES 2017–2018. Both hospital and population survey data indicated that only a few children were infected with HBV through vertical transmission, and the prevalence of hepatitis B is much lower in children than in adults, which might be related to widespread vaccination in children [27,28]. Additionally, Schwimmer et al. ’s research carried out in the western United States in 2013 revealed that AIH was the most common alternative diagnosis in non-MD-NAFLD children ($4\%$) [29]. In this context, a small portion of overweight/obese children with steatosis may have alternative etiologies, which vary by area or era. Understanding the constituents and proportion of hepatic steatosis among overweight/obese children and clinical features of different etiologies of hepatic steatosis is important for clinical doctors to find the appropriate treatment in a timely manner after lifestyle intervention if MAFLD is adopted in the future. However, we should bear in mind that MAFLD could also be concurrent with a clear cause of steatosis in children, for instance, HCV or congenital-disorder-induced hepatic steatosis. The diagnosis of NAFLD requires the exclusion of fatty liver caused by infections, dietary causes (alcohol use), medications or genetic/metabolic disorders. In contrast, the diagnosis of MAFLD does not require such exclusions. In the CPOOA study, all children with NAFLD met all the requirements of MAFLD in CPOOA as shown in Figure 4: participants with MAFLD not only covered the vast majority of participants with NAFLD but also included those with fatty liver caused by infections, diet (alcohol use), medications or genetic/metabolic disorders. It can be inferred that in overweight/obese children, the diagnosis of MAFLD is superior to that of NAFLD, as the former allows the presence of a joint diagnosis. In addition, the proportion of congenital or drug-induced hepatic steatosis is relatively small. Using MAFLD for diagnosis and treatment could reduce the burden from patients and doctors. If weight-loss treatment cannot reverse the progress of hepatic steatosis with three months of lifestyle intervention, doctors could launch more investigations on the etiologies of hepatic steatosis. Intriguingly, in the present study, 41 participants diagnosed with non-MD-NAFLD seemed healthier than those with MAFLD or NAFLD in terms of metabolic characteristics: they had the lowest BMI z-scores, lowest non-invasive scores, lowest levels of liver transaminase, and they were all under-/normal-weight children, diagnosed with so-called “lean NAFLD”. Genetic susceptibility and other metabolic abnormalities unrelated to weight gain play key roles in these children [30]. The most researched genetic variant for NAFLD is patatin-like phospholipase domain-containing 3 (PNPLA3), which can activate the transcription of thermogenic pathway genes in subcutaneous and brown adipose tissues in TghPNPLA3-I148M mice [31], indicating that genetic NAFLD is highly prone to progress to “lean NAFLD”. Another missense variant, the TM6SF2 rs58542926 (T) allele, could protect against diet-associated obesity [32]. In addition, complex factors including altered energy balance, gut microbiota dysbiosis and insulin resistance all contribute to the development of “lean NAFLD” [33,34,35]. Since these children have a distinct metabolomic profile, the levels of phosphatidylcholine, tyrosine and valine were different from those in the obese group [36]. A study based on NHANES showed that the prevalence of NAFLD among lean children during 2005 to 2014 cycles was $8\%$ [37]. In the present study, the prevalence of NAFLD and MAFLD among lean children in NHANES 2017–2018 was $11.8\%$ and $1.0\%$, respectively. Although the participants with “lean NAFLD” did not initially show significant metabolic abnormalities, they should not be ignored on account of all possible outcomes of the disease in the long term [38]. In addition, when replacing the diagnosis of NAFLD with MAFLD, these participants with non-MD-NAFLD, accounting for approximately 10–$20\%$ of NAFLD, would be missed. The present study is the first study to compare the characteristics between MAFLD and NAFLD in two different populations. However, several drawbacks need to be mentioned. First, since the data on alcohol use were restrictive for individuals under 18 years old in NHANES 2017–2018, we may have included those who drank excessively. However, we reviewed the summary data from NHANES 2017–2018 for adolescents aged 18 years; only 4 participants drank too much, and none of them was diagnosed with fatty liver. For children aged 12–17 years, only 12 children may have had excessive alcohol use, meaning that alcohol consumption had little effect on the results. Second, we cannot identify other causes of hepatic steatosis, which might increase the disparities between non-MD-NAFLD and MAFLD. Third, VCTE and ultrasound were used to diagnose hepatic steatosis in the present study. Compared with the gold standard liver biopsy, the accuracy of VCTE and ultrasound is limited. However, due to shortcomings such as sampling errors, severe complications and the high cost involved in liver biopsy, non-invasive diagnostic methods such as VCTE or ultrasound are the main detection methods for clinical applications, particularly in children. Fourth, children with diseases such as congenital, viral and drug-induced hepatic steatosis were excluded from the Chinese study by self-report. However, misclassification may occur if the participant had never had the relevant examination at the hospital or if the participant’s recall was biased. ## 5. Conclusions In conclusion, this study shows that most characteristics were similar between MAFLD and NAFLD. However, approximately $19\%$ of children with NAFLD do not have MAFLD, and this proportion fluctuates with steatosis induced by other causes, such as HCV infection and congenital disorders. Thus, to gain greater benefits from renaming NAFLD to MAFLD in pediatrics, doctors need to understand the prevalence of different causes of hepatic steatosis among children first, and then the appropriate treatment can be delivered to the patients in a timely manner, especially in cases where hepatic steatosis was not alleviated when a lifestyle intervention was carried out and weight loss was achieved. In addition, the presumed causes of hepatic steatosis in children should have a rank list for each area or country, which could help pediatricians treat hepatic steatosis among children in an appropriate and timely way. ## References 1. Fang Y.L., Chen H., Wang C.L., Liang L.. Pathogenesis of non-alcoholic fatty liver disease in children and adolescence: From “two hit theory” to “multiple hit model”. World J. Gastroenterol. (2018.0) 24 2974-2983. DOI: 10.3748/wjg.v24.i27.2974 2. Mann J.P., Valenti L., Scorletti E., Byrne C.D., Nobili V.. Nonalcoholic Fatty Liver Disease in Children. Semin. Liver Dis. (2018.0) 38 1-13. DOI: 10.1055/s-0038-1627456 3. Zhang X., Wu M., Liu Z., Yuan H., Wu X., Shi T., Chen X., Zhang T.. Increasing prevalence of NAFLD/NASH among children, adolescents and young adults from 1990 to 2017: A population-based observational study. BMJ Open (2021.0) 11 e042843. DOI: 10.1136/bmjopen-2020-042843 4. Anderson E.L., Howe L.D., Jones H.E., Higgins J.P., Lawlor D.A., Fraser A.. The Prevalence of Non-Alcoholic Fatty Liver Disease in Children and Adolescents: A Systematic Review and Meta-Analysis. PLoS ONE (2015.0) 10. DOI: 10.1371/journal.pone.0140908 5. Wang Y., Yang Z.R., Chen R.H.. Meta-analysis of the prevalence of non-alcoholic fatty liver disease in children in China. Chin. J. Child. Health Care (2022.0) 30 764-769 6. Eslam M., Sanyal A.J., George J.. MAFLD: A Consensus-Driven Proposed Nomenclature for Metabolic Associated Fatty Liver Disease. Gastroenterology (2020.0) 158 1999-2014.e1991. DOI: 10.1053/j.gastro.2019.11.312 7. Eslam M., Alkhouri N., Vajro P., Baumann U., Weiss R., Socha P., Marcus C., Lee W.S., Kelly D., Porta G.. Defining paediatric metabolic (dysfunction)-associated fatty liver disease: An international expert consensus statement. Lancet Gastroenterol. Hepatol. (2021.0) 6 864-873. DOI: 10.1016/S2468-1253(21)00183-7 8. Ramirez-Mejia M.M., Diaz-Orozco L.E., Barranco-Fragoso B., Mendez-Sanchez N.. A Review of the Increasing Prevalence of Metabolic-Associated Fatty Liver Disease (MAFLD) in Children and Adolescents Worldwide and in Mexico and the Implications for Public Health. Med. Sci. Monit. (2021.0) 27 e934134. DOI: 10.12659/MSM.934134 9. Le Garf S., Negre V., Anty R., Gual P.. Metabolic Fatty Liver Disease in Children: A Growing Public Health Problem. Biomedicines (2021.0) 9. DOI: 10.3390/biomedicines9121915 10. Guan L., Shang X.R., Liu F.H., Song J.Y., Ma J., Wang H.J.. Association of INSIG2 rs9308762 with ALT level independent of BMI. J. Pediatr. Gastroenterol. Nutr. (2014.0) 58 155-159. DOI: 10.1097/MPG.0b013e3182a87b71 11. Ciardullo S., Monti T., Perseghin G.. Prevalence of Liver Steatosis and Fibrosis Detected by Transient Elastography in Adolescents in the 2017–2018 National Health and Nutrition Examination Survey. Clin. Gastroenterol. Hepatol. (2021.0) 19 384-390.e381. DOI: 10.1016/j.cgh.2020.06.048 12. BMI-for-Age 5–19 Years (Z-Scores) 13. Eslam M., Newsome P.N., Sarin S.K., Anstee Q.M., Targher G., Romero-Gomez M., Zelber-Sagi S., Wai-Sun Wong V., Dufour J.F., Schattenberg J.M.. A new definition for metabolic dysfunction-associated fatty liver disease: An international expert consensus statement. J. Hepatol. (2020.0) 73 202-209. DOI: 10.1016/j.jhep.2020.03.039 14. MEC Laboratory Procedures Manual [Brochure]. (2017.0) 15. Friedewald W.T., Levy R.I., Fredrickson D.S.. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin. Chem. (1972.0) 18 499-502. DOI: 10.1093/clinchem/18.6.499 16. Wang H.J., Zhang H., Zhang S.W., Pan Y.P., Ma J.. Association of the common genetic variant upstream of INSIG2 gene with obesity related phenotypes in Chinese children and adolescents. Biomed. Environ. Sci. (2008.0) 21 528-536. DOI: 10.1016/S0895-3988(09)60013-1 17. Wai C.T., Greenson J.K., Fontana R.J., Kalbfleisch J.D., Marrero J.A., Conjeevaram H.S., Lok A.S.F.. A simple noninvasive index can predict both significant fibrosis and cirrhosis in patients with chronic hepatitis C. Hepatology (2003.0) 38 518-526. DOI: 10.1053/jhep.2003.50346 18. Angulo P., Hui J.M., Marchesini G., Bugianesi E., George J., Farrell G.C., Enders F., Saksena S., Burt A.D., Bida J.P.. The NAFLD fibrosis score: A noninvasive system that identifies liver fibrosis in patients with NAFLD. Hepatology (2007.0) 45 846-854. DOI: 10.1002/hep.21496 19. Harrison S.A., Oliver D., Arnold H.L., Gogia S., Neuschwander-Tetri B.A.. Development and validation of a simple NAFLD clinical scoring system for identifying patients without advanced disease. Gut (2008.0) 57 1441-1447. DOI: 10.1136/gut.2007.146019 20. Sun W., Cui H., Li N., Wei Y., Lai S., Yang Y., Yin X., Chen D.F.. Comparison of FIB-4 index, NAFLD fibrosis score and BARD score for prediction of advanced fibrosis in adult patients with non-alcoholic fatty liver disease: A meta-analysis study. Hepatol. Res. (2016.0) 46 862-870. DOI: 10.1111/hepr.12647 21. Meng L., Na L., Jie M.. Impacts of Types and Degree of Obesity on Non-Alcoholic Fatty Liver Disease and Related Dyslipidemia in Chinese School-Age Children?. Biomed. Environ. Sci. (2011.0) 24 22-30. PMID: 21440836 22. Ciardullo S., Carbone M., Invernizzi P., Perseghin G.. Impact of the new definition of metabolic dysfunction-associated fatty liver disease on detection of significant liver fibrosis in US adolescents. Hepatol Commun. (2022.0) 6 2070-2078. DOI: 10.1002/hep4.1969 23. Ziolkowska S., Binienda A., Jablkowski M., Szemraj J., Czarny P.. The Interplay between Insulin Resistance, Inflammation, Oxidative Stress, Base Excision Repair and Metabolic Syndrome in Nonalcoholic Fatty Liver Disease. Int. J. Mol. Sci. (2021.0) 22. DOI: 10.3390/ijms222011128 24. Valenti L., Bugianesi E., Pajvani U., Targher G.. Nonalcoholic fatty liver disease: Cause or consequence of type 2 diabetes?. Liver Int. (2016.0) 36 1563-1579. DOI: 10.1111/liv.13185 25. Newton K.P., Hou J., Crimmins N.A., Lavine J.E., Barlow S.E., Xanthakos S.A., Africa J., Behling C., Donithan M., Clark J.M.. Prevalence of Prediabetes and Type 2 Diabetes in Children with Nonalcoholic Fatty Liver Disease. JAMA Pediatr. (2016.0) 170 e161971. DOI: 10.1001/jamapediatrics.2016.1971 26. Yodoshi T., Orkin S., Arce-Clachar A.C., Bramlage K., Xanthakos S.A., Valentino P.L., Mouzaki M.. Alternative Etiologies of Liver Disease in Children with Suspected NAFLD. Pediatrics (2021.0) 147 e2020009829. DOI: 10.1542/peds.2020-009829 27. Ioannou G.N.. Hepatitis B virus in the United States: Infection, exposure, and immunity rates in a nationally representative survey. Ann. Intern. Med. (2011.0) 154 319-328. DOI: 10.7326/0003-4819-154-5-201103010-00006 28. Lu P.J., Yankey D., Jeyarajah J., O’Halloran A., Elam-Evans L., Greby S.M., Singleton J.A., Murphy T.V.. Hepatitis B vaccination among adolescents 13-17 years, United States, 2006–2012. Vaccine (2015.0) 33 1855-1864. DOI: 10.1016/j.vaccine.2015.02.021 29. Schwimmer J.B., Newton K.P., Awai H.I., Choi L.J., Garcia M.A., Ellis L.L., Vanderwall K., Fontanesi J.. Paediatric gastroenterology evaluation of overweight and obese children referred from primary care for suspected non-alcoholic fatty liver disease. Aliment. Pharmacol. Ther. (2013.0) 38 1267-1277. DOI: 10.1111/apt.12518 30. Kim D., Kim W.R.. Nonobese Fatty Liver Disease. Clin. Gastroenterol. Hepatol. (2017.0) 15 474-485. DOI: 10.1016/j.cgh.2016.08.028 31. Liu Z., Zhang Y., Graham S., Wang X., Cai D., Huang M., Pique-Regi R., Dong X.C., Chen Y.E., Willer C.. Causal relationships between NAFLD, T2D and obesity have implications for disease subphenotyping. J. Hepatol. (2020.0) 73 263-276. DOI: 10.1016/j.jhep.2020.03.006 32. Musso G., Cipolla U., Cassader M., Pinach S., Saba F., De Michieli F., Paschetta E., Bongiovanni D., Framarin L., Leone N.. TM6SF2 rs58542926 variant affects postprandial lipoprotein metabolism and glucose homeostasis in NAFLD. J. Lipid Res. (2017.0) 58 1221-1229. DOI: 10.1194/jlr.M075028 33. Ahadi M., Molooghi K., Masoudifar N., Namdar A.B., Vossoughinia H., Farzanehfar M.. A review of non-alcoholic fatty liver disease in non-obese and lean individuals. J. Gastroen. Hepatol. (2021.0) 36 1497-1507. DOI: 10.1111/jgh.15353 34. Xu R., Pan J., Zhou W., Ji G., Dang Y.. Recent advances in lean NAFLD. Biomed. Pharmacother. (2022.0) 153 113331. DOI: 10.1016/j.biopha.2022.113331 35. Safari Z., Gerard P.. The links between the gut microbiome and non-alcoholic fatty liver disease (NAFLD). Cell. Mol. Life Sci. (2019.0) 76 1541-1558. DOI: 10.1007/s00018-019-03011-w 36. Feldman A., Eder S.K., Felder T.K., Kedenko L., Paulweber B., Stadlmayr A., Huber-Schonauer U., Niederseer D., Stickel F., Auer S.. Clinical and Metabolic Characterization of Lean Caucasian Subjects with Non-alcoholic Fatty Liver. Am. J. Gastroenterol. (2017.0) 112 102-110. DOI: 10.1038/ajg.2016.318 37. Conjeevaram Selvakumar P.K., Kabbany M.N., Lopez R., Rayas M.S., Lynch J.L., Alkhouri N.. Prevalence of Suspected Nonalcoholic Fatty Liver Disease in Lean Adolescents in the United States. J. Pediatr. Gastroenterol. Nutr. (2018.0) 67 75-79. DOI: 10.1097/MPG.0000000000001974 38. Younes R., Govaere O., Petta S., Miele L., Tiniakos D., Burt A., David E., Vecchio F.M., Maggioni M., Cabibi D.. Caucasian lean subjects with non-alcoholic fatty liver disease share long-term prognosis of non-lean: Time for reappraisal of BMI-driven approach?. Gut (2022.0) 71 382-390. DOI: 10.1136/gutjnl-2020-322564
--- title: 'HDL-Associated Proteins in Subjects with Polycystic Ovary Syndrome: A Proteomic Study' authors: - Alexandra E. Butler - Abu Saleh Md Moin - Željko Reiner - Thozhukat Sathyapalan - Tannaz Jamialahmadi - Amirhossein Sahebkar - Stephen L. Atkin journal: Cells year: 2023 pmcid: PMC10047209 doi: 10.3390/cells12060855 license: CC BY 4.0 --- # HDL-Associated Proteins in Subjects with Polycystic Ovary Syndrome: A Proteomic Study ## Abstract Introduction. Serum lipoproteins, with the exception of high-density lipoprotein cholesterol (HDL-C), are increased in polycystic ovary syndrome (PCOS) and their levels may reflect the associated obesity and insulin resistance, but the nature of this association is not fully explained. Therefore, proteomic analysis of key proteins in lipoprotein metabolism was performed. Methods. In this cohort study, plasma was collected from 234 women (137 with PCOS and 97 controls without PCOS). Somalogic proteomic analysis was undertaken for the following 19 proteins involved in lipoprotein, and particularly HDL, metabolism: alpha-1-antichymotrypsin; alpha-1-antitrypsin; apolipoproteins A-1, B, D, E, E2, E3, E4, L1, and M; clusterin; complement C3; hemopexin; heparin cofactor II; kininogen-1; serum amyloid A-1; amyloid beta A-4; and paraoxonase-1. Results. The levels of apolipoprotein E were higher in PCOS ($$p \leq 0.012$$). However, the other isoforms of ApoE, ApoE2, E3, and E4, did not differ when compared with controls. ApoM was lower in PCOS ($$p \leq 0.000002$$). Complement C3 was higher in PCOS ($$p \leq 0.037$$), as was heparin cofactor II (HCFII) ($$p \leq 0.0004$$). The levels of the other proteins associated with lipoprotein metabolism did not differ between PCOS and controls. Conclusions. These data contribute to the concern of the deleterious dyslipidemia found in PCOS, with the novel combination reported here of higher levels of ApoE, C3 and HCFII together with lower ApoM. The dysregulation of these proteins could circumvent the protective effect of HDL-C and contribute to a more atherogenic profile that may increase cardiovascular risk. ## 1. Introduction Polycystic ovary syndrome (PCOS) is the most common endocrine disorder in premenopausal women, with its prevalence reported as 10–$15\%$ [1]. It is associated with an increased prevalence of type 2 diabetes, hypertension, insulin resistance, metabolic syndrome, and potentially cardiovascular disease (CVD) [2]. This may be mediated by inflammation, although the underlying pathophysiological mechanism remains unclear [3]. PCOS is also associated with nonalcoholic fatty liver disease (NAFLD) [4] and nonalcoholic steatohepatitis [5] and these diseases are associated with dyslipidemia and CVD. Dyslipidemia, the most important risk factor for CVD, is common in young adult women with PCOS [6,7]. In particular, the role of high-density lipoprotein cholesterol (HDL-C) particles is critical because decreased cholesterol efflux capacity has been shown in young women with PCOS, which increases their risk of atherosclerosis and CVD [8]. Both retrospective and prospective cohort studies report increased atherosclerotic coronary heart disease (ASCHD) in women with PCOS [9]. Apolipoproteins are the main components determining the metabolic fate of serum lipoprotein particles, as well as other proteins involved in lipoprotein metabolism (Figure 1). Dietary fats (exogenous triglycerides) are carried by nascent chylomicrons synthesized by the intestine. Chylomicrons are assembled in the intestinal mucosal cells and carry dietary triacylglycerol (TAGs), cholesterol, fat soluble vitamins (vitamin A, D, E, and K), and cholesteryl esters (CEs). TAGs account for close to $90\%$ of the lipids in a chylomicron. Chylomicrons are hydrolyzed by an extracellular enzyme, lipoprotein lipase (LPL), which is anchored by heparan sulphate to the capillary wall of most tissues. LPL is activated by insulin and apolipoprotein CII (CII) on circulating chylomicrons and hydrolyzes the TAGs to yield free fatty acids (FFAs) and glycerol. In muscle, fatty acids are oxidized for energy; in adipose tissue, they are re-esterized as TAGs for storage. Upon being hydrolyzed by LPL, a chylomicron is converted into a chylomicron remnant (which contains lipoproteins, Apo B-48, and Apo E). Chylomicron remnant (Apo B-48, Apo E) enters the liver through the low-density lipoprotein (LDL) receptor (Apo B-100, E). Apo CII from chylomicrons is taken back by HDL. De-novo synthesis of fatty acids (endogenous triglycerides) in the liver is carried out by nascent very-low-density lipoprotein (VLDL) (containing lipoproteins Apo B-100, Apo CII, and Apo E). Lipoprotein lipase hydrolyzes TAG to FFAs and the remnant of VLDL is called intermediate density lipoprotein IDL (containing Apo B-100, Apo E), which has lost triglyceride, but is rich in cholesterol. IDL or VLDL remnant (B-100, E) has two fates: largely, it loses Apo-E and is converted to LDL (B-100) and, minorly, it is taken up by liver via the LDL receptor. LDL (B-100) has two fates: $80\%$ of LDL enters the liver through the LDL receptor (ApoB-100, E), while $20\%$ enters extrahepatic tissues through the LDL receptor (ApoE). HDL acts as a reservoir for different apoproteins and exchanges them with other lipoproteins. It provides Apo CII and E to nascent chylomicrons and VLDL to form chylomicron (B-48, CII, E) and VLDL (B-100, CII, E). Besides being a reservoir for apoproteins, it serves the function of reverse cholesterol transport. HDL takes cholesterol from extrahepatic tissues through the ATP binding cassette (ABC-1) transporter. Lectin-cholesterol acyltransferase (LCAT) in HDL (stimulated by A-1 and Apo D) converts cholesterol to cholesteryl ester. Cholesteryl ester transport protein (CETP) mediates the exchange of cholesteryl ester for triglyceride from HDL with other lipoproteins. Reuptake of HDL-derived C occurs in the liver through scavenger receptor class B type 1 (SR-B1). The other apoproteins of HDL involve Apo L1 and Apo M. Two other HDL-associated proteins, paraoxonase-1 (PON1) and serum amyloid A1 (SAA1), are synthesized in the liver and later associated with the HDL moiety. Alzheimer’s disease (AD)-related protein amyloid beta (Aβ) has also been reported to interact with normal human plasma HDL [10]. Amyloid beta A4, also known as amyloid precursor protein (APP), functions as a cell surface receptor and transmembrane precursor protein, which is cleaved by secretases to form amyloid beta (Aβ) fiber. Human APOE lipoprotein isoforms, APOE2, APOE3, and APOE4, are involved in the pathobiology of AD. While APOE2 has a protective effect against amyloid fibril formation, APOE3 has no effect in amyloid fibril generation. APOE4, which promotes Aβ aggregation, constitutes the most significant genetic risk factor for AD. APOJ (also known as clusterin) has also been found to positively interact with Aβ (Figure 2). The regulatory proteins of Aβ, α-1 antichymotrypsin (ACT), a plasma serine protease inhibitor, targets neutrophil cathepsin G as well as mast cell chymase. ACT is an integral component of the amyloid deposits in Alzheimer’s disease (AD) and had been shown to catalyze amyloid beta (Aβ) polymerization. Alpha 1 antitrypsin (A1AT) is a serum proteinase inhibitor, especially for neutrophil elastase. While women with PCOS have shown worse cognitive performance compared with non-PCOS women across domains of memory, executive function, and brain structure [11], the association between HDL- and AD-related proteins in PCOS has not been explored before. HDL acts as an anticoagulant and its antithrombotic property involves modulation of platelet reactivity and endothelial function [12]. Kininogen-1 (encoded by the KNG1 gene in humans) is a α-2-thiol proteinase inhibitor and a constituent of the blood coagulation system as well as the kinin-kallikrein system. KNG1 gene undergoes alternative splicing to generate high-molecular-weight kininogen (HMWK) and low-molecular-weight kininogen. HMWK in turn is cleaved by the enzyme kallikrein (synthesized by pre-kallikrein with the help of coagulation factor XIIa) to produce bradykinin (Figure 2). Bradykinin is a potent endothelium-derived vasodilator and is involved in many biological processes including blood coagulation, inflammation, and blood pressure control [13]. As the risk of developing hypertension in young women with PCOS is higher than in controls, the role of proteins in the kinin-kallikrein system has not been elucidated in PCOS. An inverse association has been reported between high-density lipoprotein cholesterol (HDL-C) and the risk of atherosclerotic coronary artery disease (ASCAD) in multiple clinical and epidemiological studies [14,15], indicative of the protective effect of HDL-C. Accordingly, low levels of HDL-C underlie the most frequent form of familial dyslipidemia in younger patients with myocardial infarction [16]. In prospective studies, low levels of HDL-C and high levels of C-reactive protein were independent factors elevating the risk of a second event in patients with known ASCAD [17,18]. Furthermore, in a large interventional trial, elevating the level of HDL-C was shown to decrease the incidence of ASCAD [19]. The protective effects of HDL-C on ASCAD are considered to be a consequence of HDL-C’s role in reverse cholesterol transport, removing cholesterol from the periphery to the liver for processing and excretion in the bile [20]. However, further evidence suggests that HDL-C additionally exerts a protective influence on endothelial function [21]. As PCOS is associated with CVD, determining the levels of serum lipoproteins in PCOS subjects is important. Therefore, the aim of this study was to utilize state-of-the-art proteomics to determine circulating levels of proteins specifically involved in lipoprotein metabolism, particularly of HDL, in subjects with PCOS. ## 2. Materials and Methods Plasma levels of proteins involved in lipoprotein, and particularly HDL, metabolism were measured in women with ($$n = 137$$) and without PCOS ($$n = 97$$) from a PCOS biobank (ISRCTN70196169: 2012–2017, approved by the Newcastle and North Tyneside Ethics Committee [22]); all subjects provided written informed consent. The women were all Caucasian [22], with PCOS diagnosed according to the Rotterdam consensus, based on two out of three of the criteria; namely, clinical and biochemical evidence of hyperandrogenism (Ferriman–Gallwey score > 8), free androgen index > 4 (total testosterone > 1.5 nmol/L), and oligomenorrhea or amenorrhoea and polycystic ovaries diagnosed by transvaginal ultrasound. Confounding diagnoses such as nonclassical 21-hydroxylase deficiency were appropriately screened in detail previously [22]. The demographic data for the PCOS and control cohorts are shown in Table 1 [22]. Controls had regular menses, normal physical examination, and polycystic ovaries excluded by ultrasound, and were not on any medications. Blood was withdrawn fasting and prepared by centrifugation at 3500× g for 15 min, aliquoted, and stored at −80 °C. Analysis for sex hormone binding globulin (SHBG), insulin (DPC Immulite 200 analyser, Euro/DPC, Llanberis, UK), and plasma glucose (Synchron LX20 analyser, Beckman-Coulter, High Wycombe, UK) was undertaken. Free androgen index (FAI) was derived from total testosterone divided by SHBG ×100. Insulin resistance (IR) was calculated using the homeostasis model assessment (HOMA-IR). Serum testosterone was quantified using isotope-dilution liquid chromatography tandem mass spectrometry (LC-MS/MS) [23]. Plasma lipid-related proteins were measured by the slow off-rate modified aptamer (SOMA)-scan platform [24]. Calibration was based on standards as previously described [25]. SOMAscan technology offers significant advantages in sample size, cost, time, multiplexing capability, dynamic range, and flexibility of readout over many alternate protein biomarker platforms. The protein quantification was performed using a slow off-rate modified aptamer (SOMAmer)-based protein array, as previously described [26,27]. Briefly, EDTA plasma samples were diluted and the following assay steps were performed: [1] Binding—analytes and primer bead (PB)/SOMAmers (fully synthetic fluorophore-labeled SOMAmer coupled to a biotin moiety through a photocleavable linker) were equilibrated. [ 2] Catch I—all analyte/SOMAmers complexes were immobilized on a streptavidin-substituted support. Washing steps removed proteins not stably bound to PB/SOMAmers and the bound protein was biotinylated. [ 3] Cleave—long-wave ultraviolet light was applied to release analyte/SOMAmer complexes into the solution. [ 4] Catch II—analyte/SOMAmer complexes were selectively immobilized on streptavidin support via the introduced analyte-borne biotinylation. Further washing was continued to select against unspecific analyte/SOMAmer complexes. [ 5] Elution–denaturation caused disruption of analyte/SOMAmer complexes. Released SOMAmers serve as surrogates for the quantification of analyte concentrations. [ 6] Quantification–hybridization to custom arrays of SOMAmer-complementary oligonucleotides. Normalization of raw intensities, hybridization, median signal, and calibration signal were performed based on the standard samples included on each plate, as previously described [24,25]. Version 3.1 of the SOMAscan assay, targeting those proteins specifically involved in lipoprotein and particularly HDL metabolism in the SOMAscan panel, was used. These 19 proteins were alpha-1-antichymotrypsin; alpha-1-antitrypsin; apolipoproteins A-1, B, D, E, E2, E3, E4, L1, and M; clusterin; complement C3; hemopexin; heparin cofactor II; kininogen-1; serum amyloid A-1; amyloid beta A-4; and paraoxonase-1 (PON1). ## 3. Statistics Power was based on C3 protein changes reported to be different in PCOS [28] (nQuery version 9, Statsols, Boston, MA, USA). For an alpha of 0.05, with a common standard deviation (SD) of 0.37 based on $80\%$ power, a total of 23 subjects per arm were required. Visual inspection of the data was undertaken followed by Student’s t-tests for normally distributed data and Mann–Whitney tests for non-normally distributed data, as determined by the Kolmogorov–Smirnov test. All analyses were performed using Graphpad Prism version 9.4.1 (San Diego, CA, USA). ## 4. Results Baseline data for the 146 PCOS subjects and 97 controls are shown in Table 1. The two cohorts were age-matched, but subjects with PCOS had a greater BMI, had increased insulin resistance, hyperandrogenemia, and increased C-reactive protein (CRP, an inflammatory marker). Regarding the circulating lipoprotein profile, triglycerides (TGs) were higher ($$p \leq 0.001$$) and high-density lipoprotein cholesterol (HDL-C) was lower ($p \leq 0.0001$) in PCOS, while total cholesterol and low-density lipoprotein cholesterol (LDL-C) were comparable ($$p \leq 0.22$$ and 0.16, respectively). The TG/HDL-C ratio was higher in PCOS versus controls ($$p \leq 0.001$$). The results of the Somascan analysis of lipid-metabolism-related proteins are shown as violin plots in Figure 3 and numerically in Supplementary Table S1 for PCOS subjects and control subjects. ## 4.1. Levels of Proteins Involved in Lipid Metabolism in PCOS The levels of apolipoprotein E (ApoE) were higher in PCOS (39,054 ± 17,973 vs. 33,577 ± 13,945 RFU, $$p \leq 0.012$$, PCOS vs. control). However, the isoforms of ApoE-ApoE2, E3, and E4- were not different in PCOS compared with women without PCOS (26,1934 ± 50,357 vs. 259,099 ± 51,595 RFU of ApoE2, $$p \leq 0.67$$, PCOS vs. control; 217,377 ± 67,477 vs. 201,576 ± 61,141 RFU of ApoE3, $$p \leq 0.06$$, PCOS vs. control; 219,789 ± 58305 vs. 210,604 ± 55,689 RFU of ApoE4, $$p \leq 0.22$$, PCOS vs. control). ApoM was lower in PCOS (7878 ± 3039 vs. 9868 ± 3277 RFU, $$p \leq 0.000002$$, PCOS vs. control). Complement C3 (C3) was higher in PCOS (71,028 ± 25,536 vs. 63,896 ± 26,822 RFU of C3, $$p \leq 0.037$$, PCOS vs. control), as was heparin cofactor II (HCFII) (4156 ± 773 vs. 3821 ± 618 RFU of HCFII, $$p \leq 0.0004$$, PCOS vs. control) (Figure 3 and Supplementary Table S1). The levels of other proteins associated with lipid metabolism, namely, alpha-1-antichymotrypsin; alpha-1-antitrypsin; apolipoproteins A-1, B, D, E2, E3, E4, and L1; clusterin; hemopexin; kininogen-1; serum amyloid A-1; amyloid beta A-4; and paraoxonase-1, were comparable between PCOS subjects and controls (Figure 3 and Supplementary Table S1). ## 4.2. Correlation Analyses For the four proteins that differed between PCOS subjects and control women (ApoE, ApoM, C3, and HCFII), correlations with age; BMI; insulin resistance (HOMA-IR); testosterone; and circulating levels of TG, cholesterol, HDL-C, LDL-C, and CRP were performed. ApoE correlated positively with BMI in controls ($r = 0.27$, $$p \leq 0.01$$); correlated positively with total cholesterol in controls ($r = 0.38$, $$p \leq 0.0002$$) and PCOS ($r = 0.59$, $p \leq 0.0001$); correlated positively with TG in controls ($r = 0.61$, $p \leq 0.0001$) and PCOS ($r = 0.62$, $p \leq 0.0001$); correlated negatively with HDL-C in controls (r = −0.28, $$p \leq 0.014$$); and correlated positively with LDL-C in controls ($r = 0.34$, $$p \leq 0.003$$) and PCOS ($r = 0.53$ $p \leq 0.0001$) (Figure 4). ApoM correlated negatively with BMI in controls (r = −0.36, $$p \leq 0.0004$$) and PCOS (r = −0.57, $p \leq 0.0001$); correlated negatively with TG in controls (r = −0.50, $p \leq 0.0001$) and PCOS (r = −0.46, $p \leq 0.0001$); correlated positively with HDL-C in controls ($r = 0.54$, $p \leq 0.0001$) and PCOS ($r = 0.67$, $p \leq 0.0001$); correlated negatively with CRP in controls (r =−0.32, $$p \leq 0.003$$) and PCOS (r = −0.41, $p \leq 0.0001$); and correlated negatively with HOMA-IR in controls (r = −0.50, $$p \leq 0.007$$) and PCOS (r = −0.45, $$p \leq 0.008$$) (Figure 5). Complement C3 correlated positively with BMI in controls ($r = 0.33$, $$p \leq 0.001$$); correlated positively with TG in controls ($r = 0.44$, $p \leq 0.0001$); and correlated positively with HOMA-IR in controls ($r = 0.38$, $$p \leq 0.049$$) (Figure 6). HCFII correlated negatively with age in PCOS (r =−0.21, $$p \leq 0.02$$); correlated positively with BMI in controls ($r = 0.30$, $$p \leq 0.004$$); correlated positively with cholesterol in PCOS ($r = 0.22$, $$p \leq 0.01$$); correlated positively with TG in controls ($r = 0.52$, $p \leq 0.0001$) and PCOS ($r = 0.36$, $p \leq 0.0001$); correlated positively with CRP in controls ($r = 0.31$, $$p \leq 0.004$$) and PCOS ($r = 0.29$, $$p \leq 0.001$$); and correlated positively with HOMA-IR in controls ($r = 0.40$, $$p \leq 0.03$$) (Figure 7). ## 5. Discussion The results presented here show that the levels of ApoE (but not the isoforms ApoE2, ApoE3, and Apo E4), complement C3, and heparin cofactor II were higher, while the levels of ApoM were lower in women with PCOS versus control women. Several studies have been performed on women with PCOS analyzing their serum lipoproteins. One most recently published retrospective study on 700 Chinese women indicated that most of these subjects had low HDL-C; that subjects with clinical hyperandrogenism also had lower ApoA levels; and that the levels of TG, LDL-C, and ApoB were increased in women with PCOS with insulin resistance. This study also showed that TG and ApoB levels showed a trend towards an increase with BMI and that ApoAI, TG/HDL-C, and ApoB/ApoA ratios were linked to certain features of PCOS, specifically insulin resistance and obesity [29]. However, these authors did not compare women with PCOS to healthy women, neither did they perform a proteomic study. In one of the earliest studies on PCOS and dyslipidemia performed on south-west Chinese women with PCOS, now published a decade ago, HDL-C was decreased and TG was increased in women with PCOS versus women without PCOS. PCOS subjects who exhibited dyslipidemia had not only higher TG/HDL-C ratios, but also lower HDL-C and ApoAI levels versus controls or PCOS subjects without dyslipidemia. They also had higher BMI and fasting and 2 h insulin and glucose concentrations, as well as increased homeostatic model assessment IR (HOMA-IR); atherogenic indexes; TG, LDL-C, and ApoB concentrations; and ApoB/ApoA-I ratios versus controls [30]. The association between Apo E, and particularly ApoE2 (one of the three isoforms of the APOE gene), and atherogenesis is still controversial, although evidence suggests that apolipoprotein gene rs7412 (E2) and rs429358 (E4) single nucleotide polymorphisms (SNPs) may be associated with CVD risk. The results of a cohort study of south-west Chinese women showed no associations of any ApoE genotype with PCOS, and accord with the results of the present study [31]. In a study on subjects with PCOS in Western Anatolia, Turkey, the ApoE3 allele was reported at a higher frequency in PCOS subjects versus controls, although no significant difference was found in lipid or other CVD risk factors with regard to allele and genotype data [32]. Another study reported similar findings in PCOS with no significant difference in E3, E4, and E2 alleles of ApoE genes [33]. Of note, higher apoE in whole plasma and a higher apoE content in HDL particles were associated with lower dementia risk [34] and, conversely, reduced plasma ApoE plus the APOE ε4 allele were associated with elevated risk for Alzheimer’s disease [35]. Polymorphism of the ApoE gene is a major risk for Alzheimer’s disease, with the strongest risk being for the ε4 allele associated with lower levels of ApoE, while the ε2 allele is protective with higher levels of ApoE [36]. The various ApoE isoforms differ in their ability to bind lipids and amyloid-β, a critical protein in Alzheimer’s disease [36]. This finding of increased levels of ApoE may be pertinent in protecting against the potential increased risk of Alzheimer’s disease in PCOS, where it has been shown that these subjects may share some common risk factors [37]. Here, no association of PCOS and ApoB was found, though ApoB showed a trend to increase; this result differs from a study of young girls with PCOS, where elevated plasma apoB48-lipoprotein remnants were found to be highly associated with cardiometabolic risk and had ~2-fold elevated prevalence compared with girls without PCOS. Therefore, this may predispose them to prematurely developing atherosclerosis and CVD [38]. A recent study including proteomics of potential insulin resistance biomarkers in PCOS women showed that only ApoC3 was flagged as potentially being a diagnostic marker for PCOS-insulin-resistant subjects [39]; however, in this study, ApoC3 was not available in the Somascan panel. A single nucleotide polymorphism with a reduction in ApoM transcriptional activity and a decrease in serum ApoM levels has been reported as a potential biomarker for coronary artery disease [40]. Decreased ApoM has not been reported in PCOS previously and, in this context, may contribute to the increased atherogenic dyslipidemia seen in PCOS. ApoM is an apolipoprotein primarily located in HDL particles and required in their formation and HDL-mediated reverse cholesterol transport. ApoM is linked to the anti-atherosclerotic, anti-inflammatory, and anti-oxidant effects of HDL particles, and is associated with several diseases, including ASCVD [41]. It is interesting that plasma ApoM levels are low in subjects with type 2 diabetes mellitus (T2DM), a phenomenon that, according to some authors, is likely caused by diabetes and is not a consequence of the dyslipidemia that often accompanies T2DM [42]. However, others consider that no causal association exists between plasma ApoM and an elevated risk of T2DM [43]. ApoM, the major plasma carrier of the bioactive lipid mediator sphingosine-1-phosphate (S1P), seems to underlie several HDL-associated protective functions in the endothelium, including the regulation of adhesion molecule quantity, leukocyte–endothelial adhesion, and the endothelial barrier limiting endothelial inflammation by delivering S1 to the S1P receptor 1 [44]. Lower levels of ApoM, as found in this study, probably lead to lower ApoM/S1 levels. It has been suggested that the ApoM/S1 complex has a protective role against the development of insulin resistance, a common feature of PCOS [45]; therefore, lower ApoM levels may contribute to the insulin resistance commonly seen in PCOS and that is an independent cardiovascular risk marker [46]. Others have reported that single nucleotide polymorphisms with a reduction in ApoM transcriptional activity and a decrease in serum ApoM levels may be biomarkers for coronary artery disease [40]. Cardiac insulin resistance generates damage by at least three different mechanisms, including signal transduction alteration, impaired regulation of substrate metabolism, and altered delivery of substrates to the myocardium [40]. Higher levels of complement C3 were found here in PCOS and have been reported previously [47], which may reflect an association with the higher risk of obesity and atherosclerosis. In a PCOS population, complement C3 was associated with coronary artery calcification [48]. HDL subspecies that contain C3 are associated with higher CHD risk versus HDL without complement C3 [49]. One possible explanation might be that complement C3 is associated with ApoC2 and indirectly with ApoE [50]. The complement C3 fragment, C3a-desArg, acts as a hormone with insulin-like effects and aids triglyceride metabolism, but also promotes the production of inflammatory initiators such as the anaphylatoxin C3a, potentiating atherogenesis [51]. It has been shown that complement C3 is linked to an adverse lipoprotein profile, featuring increased triglyceride-enriched lipoproteins though fewer large HDL particles [52]. In addition to elevated triglycerides, lower levels of large HDL particles seem to be linked with increased CHD risk [53,54]. Increased levels of complement C3 have a greater association with insulin resistance than markers of inflammation such as C-reactive protein [55], and a recent study has shown a strong association of serum complement C3 with serum insulin and insulin resistance in subjects with PCOS, suggesting that this inflammatory marker might predict future diabetes and CVD complication risk in subjects with PCOS [56]; therefore, it is not surprising that, in this study, complement C3 was also elevated in PCOS versus controls. This also suggests that increased insulin resistance is mediated by complement C3 rather than complement C3, being an epiphenomenon of inflammation. In accord with the increased insulin resistance associated with reduced ApoM noted above, the increased complement C3 may augment the insulin resistance seen and its associated cardiovascular risk. Studies suggest that high plasma HCII levels are protective against in-stent restenosis and atherosclerosis [57], with HCII deficiency promoting atherogenesis in mice [58], indicative of the important role that HCII may play in vascular homeostasis. Higher heparin cofactor II (HCFII) levels were found in PCOS; however, there is scant literature on the role of HCFII in the metabolism of serum lipoproteins and development of atherogenesis. In a single study, heparin cofactor II activity and HDL-C were negatively correlated with maximum atherosclerotic plaque thickness; plasma heparin cofactor II activity and HDL-C concentration independently contributed to plaque thickness, with the antiatherogenic effects of heparin cofactor II activity being greater than the effect of HDL-C [59]. It has been reported that the HCII level was negatively correlated with a higher vulnerability of carotid plaques and that plasma HCII may be a potential biomarker for the evaluation of the vulnerability of carotid plaques [60]. Thus, in this study, the high levels of HCII may mitigate and protect against the potential detrimental effects of the raised complement C3 and the low ApoM. In this study, we did not find a decrease in PON1 activity (an antioxidant that serves to prevent lipoprotein oxidation and to hydrolyze atherogenic products generated from oxidative lipid modification). Such a decrease was, however, found in a meta-analysis of PCOS women [61]. Limited studies have suggested that antioxidant therapy in PCOS, such as with alpha lipoic acid, improves oxidative stress and insulin resistance, promotes follicular maturation, and improves glucose and lipid metabolism and vascular endothelial function, though robust and sufficiently powered studies are necessary to confirm these findings [62]. Serum complement C3 was reported to have a stronger link with insulin resistance than with high-sensitivity C-reactive protein (hsCRP) in PCOS women [55]. Therefore, it is not surprising that, in this study, complement C3 was also elevated in PCOS versus controls. A recent study has shown a strong association of serum complement C3 with serum insulin and insulin resistance in subjects with PCOS, suggesting that this inflammatory marker might predict future diabetes and CVD complication risk in subjects with PCOS [56]. The level of heparin cofactor-II in subjects with PCOS was shown to be increased [63]; however, in a multivariate analysis where BMI, inflammation, and insulin resistance were accounted for, no correlation of PCOS with either heparin cofactor II or any coagulation proteins was seen, suggesting that hypercoagulability is not an intrinsic facet of PCOS. Thus, PCOS should not be considered as a risk factor on the Veno Thrombo embolism risk assessment unless associated with other risk factors such as obesity, hormonal stimulation, or smoking. The limitations of this study include that all participants were Caucasian and, therefore, the findings may differ, to a greater or lesser extent, in other ethnic groups. Moreover, protein measurements are reported as relative fluorescent units (RFUs) by the SOMAscan software version 4.1 and cannot be converted to protein concentrations. In conclusion, these data contribute to the concern of the deleterious dyslipidemia found in women with PCOS, with the novel combination of higher levels of ApoE, C3, and HCFII together with lower ApoM being reported here; these dysregulated proteins may circumvent the protective effect of HDL and contribute to a more atherogenic profile that may increase cardiovascular risk and, at least in part, explain the atherogenic, inflammatory, insulin-resistant, and prothrombotic characteristics of PCOS. To mitigate these adverse risk factors, lifestyle modification including dietary modification and exercise should be emphasized for the management of women with PCOS with further interventional studies needed to determine if these findings are reversible. ## References 1. Azziz R.. Polycystic ovary syndrome. Obstet. Gynecol. (2018) 132 321-336. DOI: 10.1097/AOG.0000000000002698 2. Sathyapalan T., Atkin S.L.. Recent advances in cardiovascular aspects of polycystic ovary syndrome. Eur. J. Endocrinol. Eur. Fed. Endocr. Soc. (2012) 166 575-583. DOI: 10.1530/EJE-11-0755 3. Carter A.M.. Complement activation: An emerging player in the pathogenesis of cardiovascular disease. Scientifica (2012) 2012 402783. DOI: 10.6064/2012/402783 4. Yuan L., Kardashian A., Sarkar M.. NAFLD in Women: Unique Pathways, Biomarkers, and Therapeutic Opportunities. Curr. Hepatol. Rep. (2019) 18 425-432. DOI: 10.1007/s11901-019-00495-9 5. Maldonado S.S., Grab J., Wang C.W., Huddleston H., Cedars M., Sarkar M.. Polycystic ovary syndrome is associated with nonalcoholic steatohepatitis in women of reproductive age. Hepatol. Commun. (2022) 6 2634-2639. DOI: 10.1002/hep4.2039 6. Luo X., Cai W.Y., Wu X.K., Kim J.J., Choi Y.M.. Prevalence, Pattern and Predictors for Dyslipidemia of Chinese Women With Polycystic Ovary Syndrome. Front. Cardiovasc. Med. (2021) 8 790454. DOI: 10.3389/fcvm.2021.790454 7. Kim J.J., Choi Y.M.. Dyslipidemia in women with polycystic ovary syndrome. Obstet. Gynecol. Sci. (2013) 56 137-142. DOI: 10.5468/ogs.2013.56.3.137 8. Roe A., Hillman J., Butts S., Smith M., Rader D., Playford M., Mehta N.N., Dokras A.. Decreased cholesterol efflux capacity and atherogenic lipid profile in young women with PCOS. J. Clin. Endocrinol. Metab. (2014) 99 E841-E847. DOI: 10.1210/jc.2013-3918 9. Gunning M., Fauser B.. Are women with polycystic ovary syndrome at increased cardiovascular disease risk later in life?. Climacteric (2017) 20 222-227. DOI: 10.1080/13697137.2017.1316256 10. Koudinov A.R., Berezov T.T., Kumar A., Koudinova N.V.. Alzheimer’s amyloid beta interaction with normal human plasma high density lipoprotein: Association with apolipoprotein and lipids. Clin. Chim. Acta (1998) 270 75-84. DOI: 10.1016/S0009-8981(97)00207-6 11. Perović M., Wugalter K., Einstein G.. Review of the effects of polycystic ovary syndrome on Cognition: Looking beyond the androgen hypothesis. Front. Neuroendocrinol. (2022) 67 101038. DOI: 10.1016/j.yfrne.2022.101038 12. Van der Stoep M., Korporaal S.J., Van Eck M.. High-density lipoprotein as a modulator of platelet and coagulation responses. Cardiovasc. Res. (2014) 103 362-371. DOI: 10.1093/cvr/cvu137 13. Bryant J.W., Shariat-Madar Z.. Human plasma kallikrein-kinin system: Physiological and biochemical parameters. Cardiovasc. Hematol. Agents Med. Chem. (2009) 7 234-250. DOI: 10.2174/187152509789105444 14. Assmann G., Schulte H., von Eckardstein A., Huang Y.. High-density lipoprotein cholesterol as a predictor of coronary heart disease risk. The PROCAM experience and pathophysiological implications for reverse cholesterol transport. Atherosclerosis (1996) 124 S11-S20. DOI: 10.1016/0021-9150(96)05852-2 15. Gordon D.J., Rifkind B.M.. High-density lipoprotein--the clinical implications of recent studies. N. Engl. J. Med. (1989) 321 1311-1316. DOI: 10.1056/NEJM198911093211907 16. Genest J.J., Martin-Munley S.S., McNamara J.R., Ordovas J.M., Jenner J., Myers R.H., Silberman S.R., Wilson P.W., Salem D.N., Schaefer E.J.. Familial lipoprotein disorders in patients with premature coronary artery disease. Circulation (1992) 85 2025-2033. DOI: 10.1161/01.CIR.85.6.2025 17. Bolibar I., von Eckardstein A., Assmann G., Thompson S.. Short-term prognostic value of lipid measurements in patients with angina pectoris. Thromb. Haemost. (2000) 84 955-960. DOI: 10.1055/s-0037-1614155 18. Ridker P.M.. High-sensitivity C-reactive protein: Potential adjunct for global risk assessment in the primary prevention of cardiovascular disease. Circulation (2001) 103 1813-1818. DOI: 10.1161/01.CIR.103.13.1813 19. Robins S.J.. Targeting low high-density lipoprotein cholesterol for therapy: Lessons from the Veterans Affairs High-density Lipoprotein Intervention Trial. Am. J. Cardiol. (2001) 88 19-23. DOI: 10.1016/S0002-9149(01)02148-8 20. Silver D.L., Jiang X.C., Arai T., Bruce C., Tall A.R.. Receptors and lipid transfer proteins in HDL metabolism. Ann. N. Y. Acad. Sci. (2000) 902 103-112. DOI: 10.1111/j.1749-6632.2000.tb06305.x 21. Calabresi L., Gomaraschi M., Franceschini G.. Endothelial protection by high-density lipoproteins: From bench to bedside. Arterioscler. Thromb. Vasc. Biol. (2003) 23 1724-1731. DOI: 10.1161/01.ATV.0000094961.74697.54 22. Sathyapalan T., Al-Qaissi A., Kilpatrick E.S., Dargham S.R., Atkin S.L.. Anti-Mullerian hormone measurement for the diagnosis of polycystic ovary syndrome. Clin. Endocrinol. (2018) 88 258-262. DOI: 10.1111/cen.13517 23. Cunningham T.K., Allgar V., Dargham S.R., Kilpatrick E., Sathyapalan T., Maguiness S., Mokhtar Rudin H.R., Abdul Ghani N.M., Latiff A., Atkin S.L.. Association of Vitamin D Metabolites With Embryo Development and Fertilization in Women With and Without PCOS Undergoing Subfertility Treatment. Front. Endocrinol. (Lausanne) (2019) 10 13. DOI: 10.3389/fendo.2019.00013 24. Kahal H., Halama A., Aburima A., Bhagwat A.M., Butler A.E., Graumann J., Suhre K., Sathyapalan T., Atkin S.L.. Effect of induced hypoglycemia on inflammation and oxidative stress in type 2 diabetes and control subjects. Sci. Rep. (2020) 10 4750. DOI: 10.1038/s41598-020-61531-z 25. Kraemer S., Vaught J.D., Bock C., Gold L., Katilius E., Keeney T.R., Kim N., Saccomano N.A., Wilcox S.K., Zichi D.. From SOMAmer-based biomarker discovery to diagnostic and clinical applications: A SOMAmer-based, streamlined multiplex proteomic assay. PLoS ONE (2011) 6. DOI: 10.1371/journal.pone.0026332 26. Gold L., Ayers D., Bertino J., Bock C., Bock A., Brody E.N., Carter J., Dalby A.B., Eaton B.E., Fitzwater T.. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS ONE (2010) 5. DOI: 10.1371/journal.pone.0015004 27. Suhre K., Arnold M., Bhagwat A.M., Cotton R.J., Engelke R., Raffler J., Sarwath H., Thareja G., Wahl A., DeLisle R.K.. Connecting genetic risk to disease end points through the human blood plasma proteome. Nat. Commun. (2017) 8 14357. DOI: 10.1038/ncomms14357 28. Lewis R.D., Narayanaswamy A.K., Farewell D., Rees D.A.. Complement activation in polycystic ovary syndrome occurs in the postprandial and fasted state and is influenced by obesity and insulin sensitivity. Clin. Endocrinol. (2021) 94 74-84. DOI: 10.1111/cen.14322 29. Guo F., Gong Z., Fernando T., Zhang L., Zhu X., Shi Y.. The Lipid Profiles in Different Characteristics of Women with PCOS and the Interaction Between Dyslipidemia and Metabolic Disorder States: A Retrospective Study in Chinese Population. Front. Endocrinol. (2022) 13 892125. DOI: 10.3389/fendo.2022.892125 30. Zhang J., Fan P., Liu H., Bai H., Wang Y., Zhang F.. Apolipoprotein AI and B levels, dyslipidemia and metabolic syndrome in south-west Chinese women with PCOS. Hum. Reprod. (2012) 27 2484-2493. DOI: 10.1093/humrep/des191 31. Liu H.-W., Zhang F., Fan P., Bai H., Zhang J.-X., Wang Y.. Effects of apolipoprotein E genotypes on metabolic profile and oxidative stress in south-west Chinese women with polycystic ovary syndrome. Eur. J. Obstet. Gynecol. Reprod. Biol. (2013) 170 146-151. DOI: 10.1016/j.ejogrb.2013.04.016 32. Cetinkalp S., Karadeniz M., Erdogan M., Zengi A., Cetintas V., Tetik A., Eroglu Z., Kosova B., Ozgen A.G., Saygili F.. Apolipoprotein E gene polymorphism and polycystic ovary syndrome patients in Western Anatolia, Turkey. J. Assist. Reprod. Genet. (2009) 26 1-6. DOI: 10.1007/s10815-008-9280-8 33. Heinonen S., Korhonen S., Hippeläinen M., Hiltunen M., Mannermaa A., Saarikoski S.. Apolipoprotein E alleles in women with polycystic ovary syndrome. Fertil. Steril. (2001) 75 878-880. DOI: 10.1016/S0015-0282(01)01691-0 34. Koch M., DeKosky S.T., Fitzpatrick A.L., Furtado J.D., Lopez O.L., Kuller L.H., Mackey R.H., Hughes T.M., Mukamal K.J., Jensen M.K.. Apolipoproteins and Alzheimer’s pathophysiology. Alzheimer’s Dement. Diagn. Assess. Dis. Monit. (2018) 10 545-553. DOI: 10.1016/j.dadm.2018.07.001 35. Giannisis A., Al-Grety A., Carlsson H., Patra K., Twohig D., Sando S.B., Lauridsen C., Berge G., Grøntvedt G.R., Bråthen G.. Plasma apolipoprotein E levels in longitudinally followed patients with mild cognitive impairment and Alzheimer’s disease. Alzheimer’s Res. Ther. (2022) 14 115. DOI: 10.1186/s13195-022-01058-9 36. Husain M.A., Laurent B., Plourde M.. APOE and Alzheimer’s Disease: From Lipid Transport to Physiopathology and Therapeutics. Front. Neurosci. (2021) 15 630502. DOI: 10.3389/fnins.2021.630502 37. Sarahian N., Sarvazad H., Sajadi E., Rahnejat N., Eskandari Roozbahani N.. Investigation of common risk factors between polycystic ovary syndrome and Alzheimer’s disease: A narrative review. Reprod. Health (2021) 18 156. DOI: 10.1186/s12978-021-01203-x 38. Vine D.F., Beilin L.J., Burrows S., Huang R.-C., Hickey M., Hart R., Proctor S.D., Mori T.A.. ApoB48-lipoproteins are associated with cardiometabolic risk in adolescents with and without polycystic ovary syndrome. J. Endocr. Soc. (2020) 4 bvaa061. DOI: 10.1210/jendso/bvaa061 39. Li L., Zhang J., Zeng J., Liao B., Peng X., Li T., Li J., Tan Q., Li X., Yang Y.. Proteomics analysis of potential serum biomarkers for insulin resistance in patients with polycystic ovary syndrome. Int. J. Mol. Med. (2020) 45 1409-1416. DOI: 10.3892/ijmm.2020.4522 40. Zhang Y., Huang L.Z., Yang Q.L., Liu Y., Zhou X.. Correlation analysis between ApoM gene-promoter polymorphisms and coronary heart disease. Cardiovasc. J. Afr. (2016) 27 228-237. DOI: 10.5830/CVJA-2016-001 41. Christoffersen C., Obinata H., Kumaraswamy S.B., Galvani S., Ahnström J., Sevvana M., Egerer-Sieber C., Muller Y.A., Hla T., Nielsen L.B.. Endothelium-protective sphingosine-1-phosphate provided by HDL-associated apolipoprotein M. Proc. Natl. Acad. Sci. USA (2011) 108 9613-9618. DOI: 10.1073/pnas.1103187108 42. Zhang P., Gao J., Pu C., Feng G., Wang L., Huang L., Tao Q., Zhang Y.. Effects of hyperlipidaemia on plasma apolipoprotein M levels in patients with type 2 diabetes mellitus: An independent case–control study. Lip. Health Dis. (2016) 15 1-8. DOI: 10.1186/s12944-016-0325-1 43. Hajny S., Christoffersen M., Dalila N., Nielsen L.B., Tybjærg-Hansen A., Christoffersen C.. Apolipoprotein M and risk of type 2 diabetes. J. Clin. Endocrinol. Metab. (2020) 105 3046-3057. DOI: 10.1210/clinem/dgaa433 44. Ruiz M., Frej C., Holmér A., Guo L.J., Tran S., Dahlbäck B.. High-density lipoprotein–associated apolipoprotein M limits endothelial inflammation by delivering sphingosine-1-phosphate to the sphingosine-1-phosphate receptor 1. Arterioscler. Thromb. Vasc. Biol. (2017) 37 118-129. DOI: 10.1161/ATVBAHA.116.308435 45. Kurano M., Tsukamoto K., Shimizu T., Kassai H., Nakao K., Aiba A., Hara M., Yatomi Y.. Protection against insulin resistance by apolipoprotein M/sphingosine-1-phosphate. Diabetes (2020) 69 867-881. DOI: 10.2337/db19-0811 46. Ormazabal V., Nair S., Elfeky O., Aguayo C., Salomon C., Zuñiga F.A.. Association between insulin resistance and the development of cardiovascular disease. Cardiovasc. Diabetol. (2018) 17 122. DOI: 10.1186/s12933-018-0762-4 47. Moin A.S.M., Sathyapalan T., Butler A.E., Atkin S.L.. Classical and alternate complement factor overexpression in non-obese weight matched women with polycystic ovary syndrome does not correlate with vitamin D. Front. Endocrinol. (2022) 13 935750. DOI: 10.3389/fendo.2022.935750 48. Snyder M.L., Shields K.J., Korytkowski M.T., Sutton-Tyrrell K., Talbott E.O.. Complement protein C3 and coronary artery calcium in middle-aged women with polycystic ovary syndrome and controls. Gynecol. Endocrinol. (2014) 30 511-515. DOI: 10.3109/09513590.2014.895985 49. Sacks F.M., Liang L., Furtado J.D., Cai T., Davidson W.S., He Z., McClelland R.L., Rimm E.B., Jensen M.K.. Protein-Defined Subspecies of HDLs (High-Density Lipoproteins) and Differential Risk of Coronary Heart Disease in 4 Prospective Studies. Arterioscler. Thromb. Vasc. Biol. (2020) 40 2714-2727. DOI: 10.1161/ATVBAHA.120.314609 50. Furtado J.D., Yamamoto R., Melchior J.T., Andraski A.B., Gamez-Guerrero M., Mulcahy P., He Z., Cai T., Davidson W.S., Sacks F.M.. Distinct Proteomic Signatures in 16 HDL (High-Density Lipoprotein) Subspecies. Arterioscler. Thromb. Vasc. Biol. (2018) 38 2827-2842. DOI: 10.1161/ATVBAHA.118.311607 51. Barbu A., Hamad O.A., Lind L., Ekdahl K.N., Nilsson B.. The role of complement factor C3 in lipid metabolism. Mol. Immunol. (2015) 67 101-107. DOI: 10.1016/j.molimm.2015.02.027 52. Xin Y., Hertle E., van der Kallen C.J.H., Vogelzangs N., Arts I.C.W., Schalkwijk C.G., Stehouwer C.D.A., van Greevenbroek M.M.J.. C3 and alternative pathway components are associated with an adverse lipoprotein subclass profile: The CODAM study. J. Clin. Lipidol. (2021) 15 311-319. DOI: 10.1016/j.jacl.2021.01.011 53. Watanabe H., Soderlund S., Soro-Paavonen A., Hiukka A., Leinonen E., Alagona C., Salonen R., Tuomainen T.P., Ehnholm C., Jauhiainen M.. Decreased high-density lipoprotein (HDL) particle size, prebeta-, and large HDL subspecies concentration in Finnish low-HDL families: Relationship with intima-media thickness. Arterioscler. Thromb. Vasc. Biol. (2006) 26 897-902. DOI: 10.1161/01.ATV.0000209577.04246.c0 54. Reiner Z.. Hypertriglyceridaemia and risk of coronary artery disease. Nat. Rev. Cardiol. (2017) 14 401-411. DOI: 10.1038/nrcardio.2017.31 55. Yang S., Li Q., Song Y., Tian B., Cheng Q., Qing H., Zhong L., Xia W.. Serum complement C3 has a stronger association with insulin resistance than high-sensitivity C-reactive protein in women with polycystic ovary syndrome. Fertil. Steril. (2011) 95 1749-1753. DOI: 10.1016/j.fertnstert.2011.01.136 56. Khichar A., Gupta S., Mishra S., Meena M.. Assessment of Inflammatory Markers in Women with PCOS and their Correlation with Insulin Resistance. Clin. Lab. (2021) 67. DOI: 10.7754/Clin.Lab.2021.210310 57. Takamori N., Azuma H., Kato M., Hashizume S., Aihara K., Akaike M., Tamura K., Matsumoto T.. High plasma heparin cofactor II activity is associated with reduced incidence of in-stent restenosis after percutaneous coronary intervention. Circulation (2004) 109 481-486. DOI: 10.1161/01.CIR.0000109695.39671.37 58. Vicente C.P., He L., Tollefsen D.M.. Accelerated atherogenesis and neointima formation in heparin cofactor II deficient mice. Blood (2007) 110 4261-4267. DOI: 10.1182/blood-2007-04-086611 59. Aihara K., Azuma H., Takamori N., Kanagawa Y., Akaike M., Fujimura M., Yoshida T., Hashizume S., Kato M., Yamaguchi H.. Heparin cofactor II is a novel protective factor against carotid atherosclerosis in elderly individuals. Circulation (2004) 109 2761-2765. DOI: 10.1161/01.CIR.0000129968.46095.F3 60. Li J., Liu X., Xiang Y., Ding X., Wang T., Liu Y., Yin M., Tan C., Deng F., Chen L.. Alpha-2-macroglobulin and heparin cofactor II and the vulnerability of carotid atherosclerotic plaques: An iTRAQ-based analysis. Biochem. Biophys. Res. Commun. (2017) 483 964-971. DOI: 10.1016/j.bbrc.2017.01.042 61. Murri M., Luque-Ramírez M., Insenser M., Ojeda-Ojeda M., Escobar-Morreale H.F.. Circulating markers of oxidative stress and polycystic ovary syndrome (PCOS): A systematic review and meta-analysis. Hum. Reprod. Update (2013) 19 268-288. DOI: 10.1093/humupd/dms059 62. Cheng X., He B.. Clinical and Biochemical Potential of Antioxidants in Treating Polycystic Ovary Syndrome. Int. J. Women’s Health (2022) 14 467-479. DOI: 10.2147/IJWH.S345853 63. Moin A.S.M., Sathyapalan T., Diboun I., Elrayess M.A., Butler A.E., Atkin S.L.. Metabolic consequences of obesity on the hypercoagulable state of polycystic ovary syndrome. Sci. Rep. (2021) 11 5320. DOI: 10.1038/s41598-021-84586-y
--- title: Comparative Analysis on the Effect of Sarcopenia in Patients with Knee Osteoarthritis before and after Total Knee Arthroplasty authors: - Chrysanthi Liliana Tzartza - Nikolaos Karapalis - Gavriela Voulgaridou - Christiana Zidrou - Anastasios Beletsiotis - Ioanna P. Chatziprodromidou - Constantinos Giaginis - Sousana K. Papadopoulou journal: Diseases year: 2023 pmcid: PMC10047231 doi: 10.3390/diseases11010036 license: CC BY 4.0 --- # Comparative Analysis on the Effect of Sarcopenia in Patients with Knee Osteoarthritis before and after Total Knee Arthroplasty ## Abstract Introduction: Primary sarcopenia is an age-related disease that occurs mainly in older adults, while its possibility of appearance increases with age. Secondary sarcopenia is related to the presence of a disease. At times, studies have implied a connection between various diseases and the appearance of sarcopenia. Due to pain, patients with knee osteoarthritis limit their everyday activities, leading to a decrease in muscle mass and physical function. Purpose: This study aimed to investigate the impact of the coexistence of sarcopenia and osteoarthritis on patients’ rehabilitation and symptoms, such as pain, after total knee arthroplasty, compared with patients with osteoarthritis without sarcopenia. Methodology: This cross-sectional study material consisted of 20 patients with osteoarthritis, who were hospitalized at Papageorgiou Hospital of Thessaloniki for total knee arthroplasty from November 2021 to April 2022. The patients were evaluated for sarcopenia according to the FNIH criteria. The two groups were asked to complete the KOOS score questionnaire in order to evaluate the condition of their knee in two phases, before surgery and 3 months after surgery. Results: The two groups, 5 sarcopenic patients and 15 non-sarcopenic, did not show a statistically significant difference in muscle strength measurements. However, the lean mass indices, ALM (15.18 ± 3.98 versus 19.96 ± 3.65, respectively; $$p \leq 0.023$$) and ALM/height2 (5.53 ± 1.40 versus 6.98 ± 0.75, respectively; $$p \leq 0.007$$) had significant differences, since the sarcopenic group showed a reduced lean mass, especially in patients with a comorbidity of cancer. Sarcopenic patients showed a smaller increase in KOOS score compared to non-sarcopenic patients before (0.38 ± 0.09 vs. 0.35 ± 0.09, respectively; $$p \leq 0.312$$) and after surgery (0.54 ± 0.08 vs. 0.59 ± 0.10, respectively; $$p \leq 0.909$$), but without a statistically significant difference. The score increased for both groups, with the time factor playing a greater role than the group. Conclusions: Both the sarcopenic group and the control group did not show significant differences in their scores for the assessment of the affected limb in any of the two phases while completing the questionnaire. However, there was an improvement in their osteoarthritis symptoms before and after arthroplasty in both groups. Further research with a larger sample and longer recovery time is needed to draw more accurate conclusions and confirm the present results. ## 1. Introduction Sarcopenia constitutes a pathological condition accompanied by a gradual decrease in muscle mass, as well as loss of muscle function and strength [1]. The main factors or situations that lead to the degradation of muscle tissue are still under research [2]. Although there is no sufficient explanation, there are many possible proving factors, such as satellite cells, inflammation, fibroblast growth factors (FGF), hormonal factors, autophagy, myosteatosis, reactive oxygen species (ROS), p38 mitogen-activated protein kinases (p38MAPK), and p16Iu4a [2]. The multifactorial pathophysiological pathways of sarcopenia make it difficult to find out which specific mechanism causes this syndrome. In addition, different lifestyle factors augment even more the difficulty of interpreting sarcopenia. Thus, the study of lifestyle factors needs extended research. Lifestyle factors such as physical activity, nutrition/diet, sleep duration and quality, unhealthy habits or substances (nicotine and alcohol), etc., affect the expression of sarcopenia, reversing or exacerbating the symptoms. Patients’ conditions worsen with advanced age, as the prevalence of the disease increases sharply after the age of 60 years [3]. The prevalence of sarcopenia in different regions ranges around $10\%$ for people over 60 years regardless of gender, with lower rates for Asians compared to the rest of world [4]. A recent meta-analysis showed small differences between men and women. The prevalence in men is slightly higher ($11\%$) compared to women ($9\%$) and much higher in elderly nursing home residents and patients (31–$51\%$ and 23–$24\%$ for men and women, respectively) [5]. As age increases, various changes occur; physical activity decreases, dietary habits change, the gut microbiome changes, and the hormonal balance in the body is disturbed, resulting in a change in body composition and a decrease in muscle mass and strength [6]. Sarcopenia coexists with several other diseases and worsens the condition of these patients, increasing hospitalization time and mortality rates [7,8]. Diabetes mellitus is associated with adverse changes in body composition due to metabolic disturbances that can cause inflammation and oxidative stress [9]. Furthermore, the prevalence of sarcopenia appears to be quite elevated in patients with cancer, exceeding $50\%$ in elderly cancer patients [10]. Sarcopenic cancer patients have increased rates of complications and mortality compared to non-sarcopenic patients [11]. There is a two-way relationship between sarcopenia and cancer, since patients with cancer are at higher risk of developing sarcopenia than the general population due to the increased inflammatory response of the body and invasive treatments that patients undergo [12]. Early diagnosis challenges the health care system because of the different parameters and criteria that exist. Hence, the prevention of sarcopenia would be life-saving in order to avoid the burden that it imposes on both the patient and the health care system [13]. Osteoarthritis is also a very common disease that occurs in older adults and is the most frequent reason of physical inactivity. Due to low physical levels, patients with osteoarthritis have a more than ~$20\%$ higher mortality rate compared to patients without osteoarthritis [14]. Osteoarthritis is a disease characterized by a reduction in the articular cartilage surface, as well as damage to the subchondral bone and synovial membrane. Inflammatory factors lead to an imbalance between anabolic and catabolic activities, resulting in a reduction in chondrocytes [15]. Pro-inflammatory cytokines, such as interleukin 6 (IL-6), interleukin 8 (IL-8), monokine induced by gamma (MIG), macrophage inflammatory protein-1β (MIP-1β), vascular endothelial growth factor (VEGF), interferon-gamma-inducible Protein 10 (IP-10), and monocyte chemoattractant protein-1 (MCP-1), are in higher levels in patients with osteoarthritis [16]. Clinicians have to perform differential diagnosis of osteoarthritis from other diseases with knee pain, such as inflammatory arthritis (i.e., rheumatoid and psoriatic arthritis) or infectious arthritis (i.e., gout arthritis). The incidence of osteoarthritis increases with age and is affected by gender, with women experiencing higher rates than men [17]. The prevalence of osteoarthritis varies in different countries and increases as age limits increase and life expectancy increases, resulting in higher rates in older people worldwide [18]. Other studies define the prevalence of osteoarthritis at 30–$50\%$ for people under 65 years, while $80\%$ of the elderly develop osteoarthritis in at least one joint [19]. Regarding the knee, osteoarthritis is the most common condition occurring in the knee joint, and the incidence increases with age [20]. The prevalence of knee osteoarthritis ranges from $16\%$ to $22.9\%$ worldwide [21]. Furthermore, obese patients are in more danger of the onset of osteoarthritis compared to non-obese patients ($19.7\%$ vs. $10.9\%$, respectively) [22]. According to race, African American women have a higher prevalence of knee osteoarthritis compared to Caucasian women ($51\%$ vs. $46.8\%$, respectively), whereas African American men have also a higher prevalence than the Caucasian men ($40.9\%$ vs. $36.7\%$, respectively), but distinctly lower than the prevalence in women [23]. The reduction in lean muscle mass, a main feature of sarcopenia, plays an important role in the occurrence of osteoarthritis, and often, these two diseases coexist, although only a few studies have been conducted to show the parallel comorbidity and how they are interrelated [24]. With regard to knee osteoarthritis in particular, it appears that individuals with sarcopenia and obesity are more prone to it. [ 25]. The onset of osteoarthritis is manifested by pain in the affected joint and the progressive limitation of the individual’s movements and functionality. The definitive treatment of the disease is admission for surgery for joint replacement, and at the early stage, relief can be provided by weight loss and exercise [26]. The loss of functionality of the individual and the surgery, with rehabilitation required, is a costly procedure, which places a significant burden on the health care system. Patients, in their lifespan, spend around USD 15,000 for medical costs [14]. On the other hand, sarcopenia has been shown to prolong hospitalization and rehabilitation time for hospitalized and surgical patients, thus increasing the cost of health care [27]. Total knee arthroplasty is a widely used operation and a life-saving procedure for many patients with chronic osteoarthritis, the rates of which are increasing as the average age of the population increases. As the rate of surgery increases and the years go by, the rates of revision total knee arthroplasty are also increasing, either due to the advanced age of the implements or due to an infection [28,29]. Infection of the operated joint or area and loosening of the joint are the most common complications, which can lead to a second surgery in a shorter than expected time frame. A total of $30\%$ of cases that undergo revision have suffered an infection, which is the most common failure factor for both the first surgery and revision arthroplasty [30]. In view of these considerations, it is important to investigate the risk factors that slow down the patient’s recovery and prolong the hospitalization and/or the recovery time after total knee arthroplasty. Sarcopenia, since it often coexists with osteoarthritis and affects the patient’s general health status, should be diagnosed early and, if possible, be prevented before the onset of symptoms. However, there are limited available data about sarcopenia in patients who undergo total knee arthroplasty. In addition, the impact of sarcopenia coexisting with osteoarthritis on the time of recovery after surgery, as well as on other osteoarthritis symptoms, remains unclear. Thus, we aimed to investigate whether patients with coexisting sarcopenia and osteoarthritis had an improvement in osteoarthritis symptoms (i.e., pain, difficulty in daily activities, etc.) after total knee arthroplasty surgery compared to patients without sarcopenia, and the impact of sarcopenia on patients’ recovery after surgery. ## 2.1. Study Design and Participants To conduct this study, 20 Greek subjects, (11 women and 9 men; mean difference (MD) of sarcopenic group: 64.8; MD of non-sarcopenic group: 76.4) who were admitted to Papageorgiou Hospital’s 2nd Orthopedic Department for total knee arthroplasty from November 2021 to April 2022, were enrolled and examined. All patients had been previously diagnosed with knee osteoarthritis and had undergone the necessary tests for its diagnosis and for planning the surgery. Patients excluded from the study were those who did not eventually undergo surgery due to concomitant health problems. Out of all the patients enrolled, 20 patients with osteoarthritis who were to undergo total knee arthroplasty surgery were evaluated. These patients were divided into two categories based on the Foundation for the National Institutes of Health (FNIH) criteria, sarcopenic and non-sarcopenic [31]. Among the patients, there were 9 males and 11 females, with ages ranging from 49 to 84 years. Follow-up was performed after a period of rehabilitation of three months. All patients gave written consent for their participation in this study. This study was conducted according to the Declaration of Helsinki, and it was approved by the Ethic Institutional Committee ($\frac{28551}{22}$ September 2022). ## 2.2. Anthropometry The patients were weighed on a precision electronic balance (Seca 711) to the nearest 0.1 Kg, with light clothing and without footwear. Height was measured with a calibrated tape measure to the nearest 0.1 cm, without shoes. Waist and hip circumferences were also measured in the upright position with a calibrated tape measure. For waist circumference, the tape measure was placed at the narrowest point under the ribs, while for hip circumference, it was placed at the widest part of the buttocks. The measurements were taken at baseline, one day before the total knee arthroplasty. ## 2.3. Sarcopenia Measurement The patients were divided into two categories, sarcopenic and non-sarcopenic, based on FNIH criteria [31]. We used both handgrip strength and lean mass indices, appendicular lean mass (ALM) and ALM adjusted to body mass index (ALM/BMI), to categorize patients. Based on FNIH criteria, cut-off values for grip strength were <26 kg and <16 kg for men and women, respectively. Cut-off values for ALM were <19.75 kg for men and <15.02 kg for women, and cut-off points for ALM/BMI were <0.789 and <0.512 for men and women, respectively [31]. Handgrip strength was measured by using an electronic hand dynamometer (Takei 5401). The Takei 5401 dynamometer has a digital display with a measuring range from 5 to 100 kg and can be used with both the left and right hands [32]. Body composition measurement was performed with a Bodystat Quadscan 4000 bioelectric impedance analyzer. BIA gives reliable measurements of extracellular fluid and can estimate a patient’s body fat and muscle mass [33]. Both ALM and ALM/BMI were used to diagnose and categorize patients. These indices showed us the amount of muscle mass. An algorithm based on height, resistance (R) to bioelectric impedance at 50 Hz, weight, and gender was used to calculate ALM in subjects up to 80 years of age. ALM = 4.957 + (0.196 × height2/R) + (0.060 × weight) − (2.554 × gender)where for gender, men = 1, and women = 0 [34]. For patients older than 80 years, the algorithm is different. ALM = 0.827 + (0.19 × impedance index) + (2.101 × gender) + (0.060 × weight) where for gender, similarly, men = 1, and women = 0 [35]. All patients were measured with a digital handgrip dynamometer (Takei 5401), and body composition was measured by the BIA method (Quadscan 4000). ## 2.4. Knee Injury and Osteoarthritis Outcome Score (KOOS) The KOOS is divided into five chapters/categories and includes a total of 42 questions. The 1st chapter refers to the pain experienced by the patient throughout daily movements and/or activities, the 2nd to accompanying symptoms, the 3rd to the difficulty the patient experiences in carrying out daily activities, the 4th to the patient’s functionality in sports activities, and the 5th to the reduction in quality of life due to limited mobility. To estimate the total score, each chapter is first calculated separately and then reduced to a percentage [36]. One day before surgery, the participants were asked to complete the KOOS to assess their knee condition [37].The KOOS score has been described as a reliable and valid criterion for assessing pain, as well as other symptoms, such as stiffness and functional limitations when performing daily activities, in patients with knee osteoarthritis [38]. The patients were asked to complete the questionnaire again three months after surgery, and the new score was compared with the first one in order to evaluate the improvement of each patient’s condition. The KOOS is a reliable criterion to evaluate a patient’s functional status and pain, both before, as mentioned above, and after surgery [39]; this way, we are able to estimate the patient’s rehabilitation progress and the success or not of total knee arthroplasty. An increase in the KOOS compared to pre-surgery levels implies an improvement in the patient’s condition. The results were pooled, and a comparison was made between the two groups of patients. ## 2.5. Statistical Analysis Statistical analysis was performed using the statistical package SPSS v.27. Initially, a Shapiro–Wilk test was performed in order to examine the normal distribution of all dependent variables. It was found that all dependent variables except “Illness biomarker” ($p \leq 0.001$ for non-sarcopenic patients) and “Waist-to-Hip” ($$p \leq 0.005$$ for non-sarcopenic and $$p \leq 0.046$$ for sarcopenic patients) followed the normal distribution within each group ($p \leq 0.05$ for all cases). Based on this, the following statistical tests were performed. Student’s t-test was performed on independent samples to compare the mean values of all dependent variables between the two groups (non-sarcopenic and sarcopenic), except for “Disease Index” and “Waist/Speech Ratio”. The non-parametric Mann–Whitney test was used for the comparison between the two groups for the variables “Disease Index” and “Waist/heel ratio”. Two-way repeated measures analysis of variance (ANOVA) was performed to compare the mean values of the dependent variable KOOS between the two groups (non-sarcopenic and sarcopenic) before and after surgery (main effects and interaction), as previously described [40]. Results are presented as mean ± standard deviation, and the level of statistical significance was set at the $p \leq 0.05$ level. ## 3. Results Of the total of 20 patients that were evaluated, 11 women and 9 men, 5 patients were found to belong to the sarcopenic group according to the FNIH criteria, and 15 to the control group. Three patients of the sarcopenic group were male, and two were female. For each group, the mean of anthropometric and metabolic variables was calculated, as presented in Table 1. We can see the comparison of the parameters between the two groups. The two groups showed a statistically significant difference in terms of age ($$p \leq 0.038$$), with the sarcopenic subjects being older than the non-sarcopenic subjects; however, no statistically significant differences were observed in terms of other body and metabolic characteristics ($p \leq 0.01$). The patients’ muscle strength was calculated based on handgrip strength. The limits of muscle strength differed for the two genders, with 26 kg of handgrip strength defined as the minimum normal for men and 16 kg for women. No statistically significant difference was observed between sarcopenic (19.5 ± 8.2) and non-sarcopenic (25.2 ± 9.2) patients in terms of handgrip strength ($$p \leq 0.257$$). In contrast, a statistically significant difference was observed between sarcopenic and non-sarcopenic patients in lean mass, measured either as ALM (15.2 ± 4.00 vs. 20.0 ± 3.7, respectively ($$p \leq 0.023$$)) or as ALM/ΒΜΙ (0.50 ± 0.20 vs.0.70 ± 0.14, respectively; $$p \leq 0.007$$) (Figure 1a–c). Two-way repeated measures ANOVA showed a statistically significant interaction (time x group) ($$p \leq 0.03$$) and a significant main effect of “time” ($p \leq 0.01$). In contrast, there was no statistically significant main effect of “group” ($$p \leq 0.74$$). Post-hoc analyses showed that the groups did not differ significantly before (0.38 ± 0.09 vs. 0.35 ± 0.09; $$p \leq 0.312$$) or after (0.54 ± 0.08 vs. 0.59 ± 0.10; $$p \leq 0.909$$) surgery. However, both sarcopenic ($$p \leq 0.06$$) and non-sarcopenic ($p \leq 0.001$) patients showed improvement after surgery on the KOOS index (Table 2 and Figure 2). ## 4. Discussion The aim of this study was to investigate the improvement of osteoarthritis symptoms, such as pain, in patients with and without sarcopenia after arthroplasty surgery. In the present study, patients with confirmed osteoarthritis were examined and categorized into sarcopenic and non-sarcopenic groups based on muscle strength and lean mass. Muscle strength was calculated using a handheld dynamometer. Comparison of the results for handgrip strength showed that the mean values in the non-sarcopenic group were higher than in the sarcopenic group, but this difference between the two groups was not statistically significant. The lean mass of the patients was calculated based on the ALM index, according to the measurements made with the bioelectric impedance device. Lean mass showed a statistically significant difference between the sarcopenic patients and the non-sarcopenic group. The two groups showed no statistically significant difference in the KOOS, before and after surgery, but a significant difference showed after surgery in each group. Sarcopenic patients had a slightly higher mean score. The score of both groups increased significantly after surgery. The non-sarcopenic patients showed a greater improvement in score 3 months after; however, the effect of time seemed to be more significant than the effect of group. There are not sufficient data available about sarcopenia in patients undergoing arthroplasty, although sarcopenia is associated with osteoarthritis, as shown by previous studies [32]. The prevalence of patients with sarcopenia awaiting for total knee arthroplasty is $35.5\%$ in patients aged 56–87 years old; the prevalence of severe sarcopenia increases with age [6]. The presence of sarcopenia in patients with osteoarthritis could be secondary, as severe pain and poor physical function limit the ability of patients for activities, increasing the risk of its appearance [6]. A cohort study with patients >65 years old showed that patients with sarcopenia had worse outcomes after knee arthroplasty than the non-sarcopenic patients [41]. In contrast to this study, Ho et al. [ 2021] [42] found that both patients, aged >50 years old, with and without sarcopenia improved their functional capacity after total knee arthroplasty surgery. This result is in total agreement with our results. A meta-analysis to be carried out by Wang et al. [ 2022] [32] may provide a better explanation about the association between osteoarthritis and sarcopenia. Moreover, it is known that obesity and sarcopenic obesity are risk factors for the onset of osteoarthritis, especially for women [43]. Nevertheless, sarcopenia has, independently, a negative impact on the therapeutic effects of rehabilitation on physical mobility in patients undergoing total knee arthroplasty, as shown by a previous study [44]. On the other hand, the KOOS is a reliable questionnaire that assesses the patients’ condition before and after total knee arthroplasty surgery [44]. Notably, one year after surgery, the patients’ condition was significantly improved in terms of functionality and pain [44]. To the best of our knowledge, this was the first study that tried to investigate the impact of sarcopenia in patients with osteoarthritis on their rehabilitation progress, and osteoarthritis symptoms after total knee arthroplasty, compared to patients with osteoarthritis without sarcopenia. Regarding the limitations of this study, it is worth noting that age group plays an important role in both sarcopenia and osteoarthritis. The mean age of the two study groups, sarcopenic and non-sarcopenic, showed a statistically significant difference, with a higher mean age for the sarcopenic group. So, the two types of sarcopenia were probably mixed. However, in the non-sarcopenic group, there was only one patient who was 49 years old, which significantly lowered the mean. The handgrip strength measurement for the two groups showed a small difference, not significant in the mean measurements. The patients with sarcopenia had a lower mean handgrip strength. In addition, while both groups showed a significant increase in KOOS after surgery, the sarcopenic group showed a smaller mean increase in score. However, the difference between the two groups was not statistically significant, which may be due to the sample size. The study sample was limited, with only 20 patients, due to the reduced number of surgeries performed in the previous period. Finally, the KOOS is based on patients’ reports, which may be influenced to some extent by their current psychology and/or the different sense of pain that undoubtedly exists between people. ## 5. Conclusions Both the sarcopenic patient group and the control group showed no significant differences between them in the scores for the assessment of the affected limb in either phase of the questionnaire, but a significant improvement was depicted in each group after arthroplasty surgery. The sarcopenic patients had a smaller increase in KOOS, but no statistically significant difference. Although there was a statistically significant grouptime interaction, it appeared that there was no significant difference between the groups. Further investigation is strongly recommended with a larger sample and longer recovery time in order for more precise conclusions to be drawn and confirm the present results. ## References 1. Rosenberg I.H.. Sarcopenia: Origins and Clinical Relevance. J. Nutr.* (1997) 127 990S-991S. DOI: 10.1093/jn/127.5.990S 2. Kim J.W., Kim R., Choi H., Lee S.-J., Bae G.-U.. Understanding of Sarcopenia: From Definition to Therapeutic Strategies. Arch. Pharm. Res. (2021) 44 876-889. DOI: 10.1007/s12272-021-01349-z 3. Mayhew A.J., Amog K., Phillips S., Parise G., McNicholas P.D., de Souza R.J., Thabane L., Raina P.. The Prevalence of Sarcopenia in Community-Dwelling Older Adults, an Exploration of Differences between Studies and within Definitions: A Systematic Review and Meta-Analyses. Age Ageing (2019) 48 48-56. DOI: 10.1093/ageing/afy106 4. Abellan Van Kan G.. Epidemiology and Consequences of Sarcopenia. J. Nutr. Health Aging (2009) 13 708-712. DOI: 10.1007/s12603-009-0201-z 5. Papadopoulou S.K., Tsintavis P., Potsaki G., Papandreou D.. Differences in the Prevalence of Sarcopenia in Community-Dwelling, Nursing Home and Hospitalized Individuals. A Systematic Review and Meta-Analysis. J. Nutr. Health Aging (2020) 24 83-90. DOI: 10.1007/s12603-019-1267-x 6. Daily J.W., Park S.. Sarcopenia Is a Cause and Consequence of Metabolic Dysregulation in Aging Humans: Effects of Gut Dysbiosis, Glucose Dysregulation, Diet and Lifestyle. Cells (2022) 11. DOI: 10.3390/cells11030338 7. Gariballa S., Alessa A.. Sarcopenia: Prevalence and Prognostic Significance in Hospitalized Patients. Clin. Nutr. (2013) 32 772-776. DOI: 10.1016/j.clnu.2013.01.010 8. Xu J., Wan C.S., Ktoris K., Reijnierse E.M., Maier A.B.. Sarcopenia Is Associated with Mortality in Adults: A Systematic Review and Meta-Analysis. Gerontology (2022) 68 361-376. DOI: 10.1159/000517099 9. Mesinovic J., Zengin A., De Courten B., Ebeling P.R., Scott D.. Sarcopenia and Type 2 Diabetes Mellitus: A Bidirectional Relationship. Diabetes Metab. Syndr. Obes. Targets Ther. (2019) 12 1057-1072. DOI: 10.2147/DMSO.S186600 10. Dunne R.F., Loh K.P., Williams G.R., Jatoi A., Mustian K.M., Mohile S.G.. Cachexia and Sarcopenia in Older Adults with Cancer: A Comprehensive Review. Cancers (2019) 11. DOI: 10.3390/cancers11121861 11. Shachar S.S., Williams G.R., Muss H.B., Nishijima T.F.. Prognostic Value of Sarcopenia in Adults with Solid Tumours: A Meta-Analysis and Systematic Review. Eur. J. Cancer (2016) 57 58-67. DOI: 10.1016/j.ejca.2015.12.030 12. Taylor J.M., Song A., David A.R., Chen V.E., Lu B., Werner-Wasik M.. Impact of Sarcopenia on Survival in Patients With Early-Stage Lung Cancer Treated With Stereotactic Body Radiation Therapy. Cureus (2020) 12 e10712. DOI: 10.7759/cureus.10712 13. Norman K., Otten L.. Financial Impact of Sarcopenia or Low Muscle Mass—A Short Review. Clin. Nutr. (2019) 38 1489-1495. DOI: 10.1016/j.clnu.2018.09.026 14. Katz J.N., Arant K.R., Loeser R.F.. Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. JAMA (2021) 325 568-578. DOI: 10.1001/jama.2020.22171 15. Sandell L.J., Aigner T.. Articular Cartilage and Changes in Arthritis: Cell Biology of Osteoarthritis. Arthritis Res. Ther. (2001) 3 107. DOI: 10.1186/ar148 16. Sohn D.H., Sokolove J., Sharpe O., Erhart J.C., Chandra P.E., Lahey L.J., Lindstrom T.M., Hwang I., Boyer K.A., Andriacchi T.P.. Plasma Proteins Present in Osteoarthritic Synovial Fluid Can Stimulate Cytokine Production via Toll-like Receptor 4. Arthritis Res. Ther. (2012) 14 R7. DOI: 10.1186/ar3555 17. Litwic A., Edwards M.H., Dennison E.M., Cooper C.. Epidemiology and Burden of Osteoarthritis. Br. Med. Bull. (2013) 105 185-199. DOI: 10.1093/bmb/lds038 18. Safiri S., Kolahi A.-A., Smith E., Hill C., Bettampadi D., Mansournia M.A., Hoy D., Ashrafi-Asgarabad A., Sepidarkish M., Almasi-Hashiani A.. Global, Regional and National Burden of Osteoarthritis 1990-2017: A Systematic Analysis of the Global Burden of Disease Study 2017. Ann. Rheum. Dis. (2020) 79 819-828. DOI: 10.1136/annrheumdis-2019-216515 19. Loeser R.F.. Age-Related Changes in the Musculoskeletal System and the Development of Osteoarthritis. Clin. Geriatr. Med. (2010) 26 371-386. DOI: 10.1016/j.cger.2010.03.002 20. Verma D.K., Kumari P., Kanagaraj S.. Engineering Aspects of Incidence, Prevalence, and Management of Osteoarthritis: A Review. Ann. Biomed. Eng. (2022) 50 237-252. DOI: 10.1007/s10439-022-02913-4 21. Cui A., Li H., Wang D., Zhong J., Chen Y., Lu H.. Global, Regional Prevalence, Incidence and Risk Factors of Knee Osteoarthritis in Population-Based Studies. EClinicalMedicine (2020) 29–30 100587. DOI: 10.1016/j.eclinm.2020.100587 22. Losina E., Weinstein A.M., Reichmann W.M., Burbine S.A., Solomon D.H., Daigle M.E., Rome B.N., Chen S.P., Hunter D.J., Suter L.G.. Lifetime Risk and Age at Diagnosis of Symptomatic Knee Osteoarthritis in the US. Arthritis Care Res. (2013) 65 703-711. DOI: 10.1002/acr.21898 23. Jordan J.M., Helmick C.G., Renner J.B., Luta G., Dragomir A.D., Woodard J., Fang F., Schwartz T.A., Abbate L.M., Callahan L.F.. Prevalence of Knee Symptoms and Radiographic and Symptomatic Knee Osteoarthritis in African Americans and Caucasians: The Johnston County Osteoarthritis Project. J. Rheumatol. (2007) 34 172-180. PMID: 17216685 24. Pickering M.-E., Chapurlat R.. Where Two Common Conditions of Aging Meet: Osteoarthritis and Sarcopenia. Calcif. Tissue Int. (2020) 107 203-211. DOI: 10.1007/s00223-020-00703-5 25. Misra D., Fielding R.A., Felson D.T., Niu J., Brown C., Nevitt M., Lewis C.E., Torner J., Neogi T.. Risk of Knee Osteoarthritis With Obesity, Sarcopenic Obesity, and Sarcopenia. Arthritis Rheumatol. (2019) 71 232-237. DOI: 10.1002/art.40692 26. Hunter D.J., Bierma-Zeinstra S.. Osteoarthritis. Lancet (2019) 393 1745-1759. DOI: 10.1016/S0140-6736(19)30417-9 27. Ardeljan A.D., Polisetty T.S., Palmer J., Vakharia R.M., Roche M.W.. Comparative Analysis on the Effects of Sarcopenia Following Primary Total Knee Arthroplasty: A Retrospective Matched-Control Analysis. J. Knee Surg. (2022) 35 128-134. DOI: 10.1055/s-0040-1713355 28. Cram P., Lu X., Kates S.L., Singh J.A., Li Y., Wolf B.R.. Total Knee Arthroplasty Volume, Utilization, and Outcomes among Medicare Beneficiaries, 1991–2010. JAMA (2012) 308 1227. DOI: 10.1001/2012.jama.11153 29. Hossain F., Patel S., Haddad F.S.. Midterm Assessment of Causes and Results of Revision Total Knee Arthroplasty. Clin. Orthop. (2010) 468 1221-1228. DOI: 10.1007/s11999-009-1204-0 30. Studenski S.A., Peters K.W., Alley D.E., Cawthon P.M., McLean R.R., Harris T.B., Ferrucci L., Guralnik J.M., Fragala M.S., Kenny A.M.. The FNIH Sarcopenia Project: Rationale, Study Description, Conference Recommendations, and Final Estimates. J. Gerontol. Ser. A (2014) 69 547-558. DOI: 10.1093/gerona/glu010 31. Wang H., Wang N., Wang Y., Li H.. Association between Sarcopenia and Osteoarthritis: A Protocol for Meta-Analysis. PLoS ONE (2022) 17. DOI: 10.1371/journal.pone.0272284 32. Abu Khaled M., McCutcheon M.J., Reddy S., Pearman P.L., Hunter G.R., Weinsier R.L.. Electrical Impedance in Assessing Human Body Composition: The BIA Method. Am. J. Clin. Nutr. (1988) 47 789-792. DOI: 10.1093/ajcn/47.5.789 33. Scafoglieri A., Clarys J.P., Bauer J.M., Verlaan S., Malderen L.V., Vantieghem S., Cederholm T., Sieber C.C., Mets T., Bautmans I.. Predicting Appendicular Lean and Fat Mass with Bioelectrical Impedance Analysis in Older Adults with Physical Function Decline—The PROVIDE Study. Clin. Nutr. (2017) 36 869-875. DOI: 10.1016/j.clnu.2016.04.026 34. Vermeiren S., Beckwée D., Vella-Azzopardi R., Beyer I., Knoop V., Jansen B., Delaere A., Antoine A., Bautmans I., Scafoglieri A.. Evaluation of Appendicular Lean Mass Using Bio Impedance in Persons Aged 80+: A New Equation Based on the BUTTERFLY-Study. Clin. Nutr. (2019) 38 1756-1764. DOI: 10.1016/j.clnu.2018.07.029 35. Roos E.M., Toksvig-Larsen S.. Knee Injury and Osteoarthritis Outcome Score (KOOS)—Validation and Comparison to the WOMAC in Total Knee Replacement. Health Qual. Life Outcomes (2003) 1 17. DOI: 10.1186/1477-7525-1-17 36. Moutzouri M., Tsoumpos P., Billis E., Papoutsidakis A., Gliatis J.. Cross-Cultural Translation and Validation of the Greek Version of the Knee Injury and Osteoarthritis Outcome Score (KOOS) in Patients with Total Knee Replacement. Disabil. Rehabil. (2015) 37 1477-1483. DOI: 10.3109/09638288.2014.972583 37. Lyman S., Lee Y.-Y., Franklin P.D., Li W., Cross M.B., Padgett D.E.. Validation of the KOOS, JR: A Short-Form Knee Arthroplasty Outcomes Survey. Clin. Orthop. (2016) 474 1461-1471. DOI: 10.1007/s11999-016-4719-1 38. Roos E.M., Roos H.P., Lohmander L.S., Ekdahl C., Beynnon B.D.. Knee Injury and Osteoarthritis Outcome Score (KOOS)—Development of a Self-Administered Outcome Measure. J. Orthop. Sports Phys. Ther. (1998) 28 88-96. DOI: 10.2519/jospt.1998.28.2.88 39. Mazziotta C., Pellielo G., Tognon M., Martini F., Rotondo J.C.. Significantly Low Levels of IgG Antibodies Against Oncogenic Merkel Cell Polyomavirus in Sera From Females Affected by Spontaneous Abortion. Front. Microbiol. (2021) 12 3542. DOI: 10.3389/fmicb.2021.789991 40. He Z., Cai J., Wang X., Lu D.. Clinical and Functional Outcomes of Total Knee Arthroplasty in Sarcopenia: A Case–Control Retrospective Cohort Study. J. Knee Surg. (2021). DOI: 10.1055/s-0041-1740928 41. Ho K.K.-W., Lau L.C.-M., Chau W.-W., Poon Q., Chung K.-Y., Wong R.M.-Y.. End-Stage Knee Osteoarthritis with and without Sarcopenia and the Effect of Knee Arthroplasty—A Prospective Cohort Study. BMC Geriatr. (2021) 21. DOI: 10.1186/s12877-020-01929-6 42. Jin W.S., Choi E.J., Lee S.Y., Bae E.J., Lee T.-H., Park J.. Relationships among Obesity, Sarcopenia, and Osteoarthritis in the Elderly. J. Obes. Metab. Syndr. (2017) 26 36-44. DOI: 10.7570/jomes.2017.26.1.36 43. Liao C.-D., Chen H.-C., Huang S.-W., Liou T.-H.. Impact of Sarcopenia on Rehabilitation Outcomes after Total Knee Replacement in Older Adults with Knee Osteoarthritis. Ther. Adv. Musculoskelet. Dis. (2021) 13 1-12. DOI: 10.1177/1759720X21998508 44. Goldhahn S., Takeuchi R., Nakamura N., Nakamura R., Sawaguchi T.. Responsiveness of the Knee Injury and Osteoarthritis Outcome Score (KOOS) and the Oxford Knee Score (OKS) in Japanese Patients with High Tibial Osteotomy. J. Orthop. Sci. (2017) 22 862-867. DOI: 10.1016/j.jos.2017.04.013
--- title: An Automated Method for Classifying Liver Lesions in Contrast-Enhanced Ultrasound Imaging Based on Deep Learning Algorithms authors: - Mădălin Mămuleanu - Cristiana Marinela Urhuț - Larisa Daniela Săndulescu - Constantin Kamal - Ana-Maria Pătrașcu - Alin Gabriel Ionescu - Mircea-Sebastian Șerbănescu - Costin Teodor Streba journal: Diagnostics year: 2023 pmcid: PMC10047233 doi: 10.3390/diagnostics13061062 license: CC BY 4.0 --- # An Automated Method for Classifying Liver Lesions in Contrast-Enhanced Ultrasound Imaging Based on Deep Learning Algorithms ## Abstract Background: Contrast-enhanced ultrasound (CEUS) is an important imaging modality in the diagnosis of liver tumors. By using contrast agent, a more detailed image is obtained. Time-intensity curves (TIC) can be extracted using a specialized software, and then the signal can be analyzed for further investigations. Methods: The purpose of the study was to build an automated method for extracting TICs and classifying liver lesions in CEUS liver investigations. The cohort contained 50 anonymized video investigations from 49 patients. Besides the CEUS investigations, clinical data from the patients were provided. A method comprising three modules was proposed. The first module, a lesion segmentation deep learning (DL) model, handled the prediction of masks frame-by-frame (region of interest). The second module performed dilation on the mask, and after applying colormap to the image, it extracted the TIC and the parameters from the TIC (area under the curve, time to peak, mean transit time, and maximum intensity). The third module, a feed-forward neural network, predicted the final diagnosis. It was trained on the TIC parameters extracted by the second model, together with other data: gender, age, hepatitis history, and cirrhosis history. Results: For the feed-forward classifier, five classes were chosen: hepatocarcinoma, metastasis, other malignant lesions, hemangioma, and other benign lesions. Being a multiclass classifier, appropriate performance metrics were observed: categorical accuracy, F1 micro, F1 macro, and Matthews correlation coefficient. The results showed that due to class imbalance, in some cases, the classifier was not able to predict with high accuracy a specific lesion from the minority classes. However, on the majority classes, the classifier can predict the lesion type with high accuracy. Conclusions: The main goal of the study was to develop an automated method of classifying liver lesions in CEUS video investigations. Being modular, the system can be a useful tool for gastroenterologists or medical students: either as a second opinion system or a tool to automatically extract TICs. ## 1. Introduction In recent years, with the evolution of technology and algorithms, artificial intelligence has had a significant impact in the healthcare field. Machine learning or deep learning algorithms are used to either predict the malignancy of a lesion or to segment a lesion in a medical imaging [1,2,3,4]. Ultrasound liver investigation is a non-invasive method with relatively low cost, and it can be effective in evaluating multiple types of liver lesions. Contrast-enhanced ultrasound (CEUS) has become an increasingly important imaging modality in the diagnosis and management of liver tumors. CEUS combines the use of a contrast agent that is administered intravenously with ultrasound imaging to create a more detailed image of the liver [5]. Second-generation contrasting agents use stabilized microbubbles that oscillate under the compressive effect of an ultrasound (US) beam and create a non-linear response, thus providing vascular contrast. This technique has been shown to be highly accurate in diagnosing and staging liver tumors, which have different vascularity compared to normal parenchyma, allowing it to become a part of the standard of care in many medical facilities [5]. The type of pattern within a lesion reflects in the maximum contrast uptake values within the time needed for the agent to clear the surveilled area. Thus, a time-intensity curve (TIC) can be generated by a computerized system, and comparing the measurements from within the tumor and from the surrounding parenchyma may provide a quantification usable in diagnosis, staging, and prognosis [6]. The use of CEUS for the diagnosis of liver tumors has been supported by recent medical guidelines and the literature. Both the American Association for the Study of Liver Diseases (AASLD) [7] and the European Association for the Study of the Liver (EASL) [8] recommend the use of CEUS for the evaluation of focal liver lesions. The low cost of operation, large availability, and virtual lack of side effects or contraindications made CEUS a valid choice of imaging when compared to other techniques, such as computer tomography (CT) and magnetic resonance imaging (MRI) in detecting early-stage liver tumors. However, a steep learning curve combined with an uneven training curricula and lack of consistency in the operation of the US machine can decrease the diagnostic yield of the procedure. The development and implementation of an artificial intelligence (AI)-based system that can aid the clinician in diagnosing liver malignancies through US and CEUS can thus be considered a possible solution. ## 2. Materials and Methods The anonymized dataset used in the study was provided by the University of Medicine and Pharmacy Craiova, and it included 50 video files from CEUS liver investigations performed on the same Hitachi Aloka Arieta V70 350 (Hitachi Medical Corp.; Tokyo, Japan) with the convex C 251 probe and using SonoVue (Bracco SpA, Milan, Italy) as intravascular contrast agent. The total number of patients was 49, with 59 liver lesions. For every patient involved in the study, besides the CEUS video investigation, detailed medical results and history were provided. The anonymized video files were used to build a dataset for training a U-Net image segmentation model—basically, each frame consisted of a side-by-side representation of the B-mode US and the CEUS windows. Frames from each video investigation were automatically extracted and cropped. The mask for each frame extracted was performed manually by a senior gastroenterologist (L.D.S.) using QuPath [9]. The labels for the dataset (masks) were automatically exported from QuPath. A more detailed presentation of this method was conducted in our previous work [10]. The demographic information of the patients is presented in Table 1. Data were extracted from a cohort of patients with either liver cirrhosis or chronic hepatitis of any etiology that had at least one focal liver lesion identifiable by US. All patients underwent a complete CEUS investigation. Other inclusion criteria were age over 18, irrespective of gender or other comorbidities, and confirmation of a liver tumor by other imaging (contrast-enhanced CT, MRI or a combination of the two), liver biopsy or surgical confirmation depending on each case, with a follow-up of at least 12 months. Patients that had missing medical records or incomplete CEUS were excluded. For every patient involved in the study, besides the CEUS video investigation, detailed medical results and history were provided. ## 3. Proposed Method The current study presents a system based on artificial intelligence algorithms for liver lesions detection and classification in CEUS investigations. The proposed system includes three modules. The first module is an image segmentation deep learning (DL) model based on a U-Net [11] model trained on an in-house dataset. The second module represents a frame processing algorithm and a time-intensity curve (TIC) extraction algorithm. The last component, a classifier based on feed-forward neural networks, requires as input the data extracted from the TICs and clinical data of the patient. The architecture of the proposed method is presented in Figure 1. The U-Net [11] image segmentation model obtained from our previous work [10] was used to obtain region of interest for each frame in the video investigation file. By passing a frame from the ultrasound video investigation through the presented image segmentation model, a mask of 256 by 256 pixels is obtained. This mask can be further used to isolate the lesion from parenchyma in that specific video frame. Running a video investigation through the image segmentation model, the lesion is automatically isolated from the parenchyma, frame-by-frame, in the entire file; therefore, a time-intensity curve can be extracted from the investigation. A sample of inputs and outputs of the U-Net segmentation model are presented in Figure 2A–F. To create the dataset and for extracting the lesion from each frame in the video investigation, boundaries were defined for each region in the video investigation. The boundaries for B-mode image were defined in our previous work [10] and were used to create the dataset but also for predicting the mask. For the contrast image section in the video investigation file, the boundaries were determined experimentally as for the B-mode image. A frame was plotted from a video investigation file, and the minimum and maximum x and y coordinates were defined. The determined values are presented in Table 2. A sample frame with the determined coordinates is presented in Figure 3. As mentioned earlier, the goal of the present study was to create a system to automatically extract time-intensity curves and predict the malignancy of a lesion from contrast-enhanced ultrasound imaging. Therefore, besides cropping the B-mode and contrast mode images from the video investigation, the color scale was also needed. For each ultrasound device, the color scale is present in the video investigation file. To crop and obtain the color scale, the same principle was applied as for B-mode imaging and contrast mode imaging: experimentally determine the minimum and maximum x and y coordinates. The determined values for the color scale are presented in Table 3. In addition, a sample frame with the determined coordinates is presented in Figure 3. With all the values for each section in the video investigation file determined and with the predicted mask from the image segmentation model, the lesion in B-mode or contrast mode could be extracted effortlessly. Extracting the time-intensity curves in liver contrast-enhanced ultrasound imaging is not a simple task due to multiple perturbations: patient breathing or operator moving the probe. While the lesion represented the mask predicted by the trained U-Net segmentation model, the parenchyma could not be represented by the remainder of the image (after removing the lesion). Therefore, a small region around the predicted mask had to be chosen. As the goal of the study is to present an automatically TIC extraction and prediction system, an algorithm had to be used for parenchyma determination. Dilation, which is a morphological operation, was applied on the predicted mask. Dilation is a mathematical morphology operation performed on images in which a kernel is applied to the image to obtain an output (a new image). In most of the cases, the dilation operation is applied on a binary image. Assuming that A represents the mask predicted by the U-Net segmentation model and K the kernel (structuring element), dilation is given by Equation [1]. Ak represented the translation of A by k. In our study, the size of K, the kernel, was 10 pixels by 10 pixels. [ 1]A ⊕K=∪k ∈KAk After applying dilation on the predicted mask, a new image was obtained: the dilated mask. By applying dilation to the predicted mask pixels, the boundaries of the lesion are increased. Since both images, the predicted mask and the dilated mask, are binary images, a difference between the dilated mask and the predicted mask can be calculated to obtain a mask which represented the parenchyma surrounding the lesion. Two samples are shown in Figure 4. We present the B-mode images of the lesions in Figure 4A,D. In Figure 4B,E the predicted masks by the segmentation model for 4A and 4D are presented, and 4C presents the mask for the parenchyma surrounding the lesions. By knowing the coordinates for each image type, B-mode or contrast mode, in the video investigation file, the mask obtained for the B-mode was applied for the contrast mode by using the proper coordinates (defined in Table 1). In addition, by knowing the coordinates of B-mode and contrast mode images, the position in the image of the B-mode and the contrast mode have been determined (left or right side). By applying the two masks obtained earlier to the contrast mode image, two images were obtained: the first image represented the lesion, isolated from the rest of the liver tissue, while the second image represented the surrounding parenchyma. To apply the colormap provided by the ultrasound device manufacturer, a custom lookup table was applied for both images. The values for the lookup table are presented in Table 1 (color map). A lookup table represents an array of data used for mapping input values to output values, and it is mostly used for functions which require a long time to compute. In our study, the lookup table was used for applying the color map to the images but also to improve the overall execution time. For each type (lesion and parenchyma), the intensity was obtained by performing the mean of the pixels in the cropped images after applying the color map. Two sets of values were obtained: the intensity for the parenchyma and the intensity for the detected lesion. The extraction of the intensity values for each type (parenchyma and lesion) was performed with a rate of one per second to give the system enough time to extract the TIC values. In CEUS, 10 frames per video investigation can be enough to extract a reliable TIC [12]. In some cases in the video investigations, the probe was lifted by the operator, and therefore computing the intensity value in this case was difficult. To treat this situation properly, an intensity value was considered valid if the difference between the current value and previous was under $75\%$ from the previous value. If the difference was greater than $75\%$, the next frame was analyzed. This value was determined experimentally by analyzing the variations of the intensity in the entire frame in each second of the video file. Extracting intensity values with a frequency of one second and checking the values for possible probe lifting by the operator resulted in a curve with less noise on which parameters could be extracted. Before extracting the parameters, the TIC values were filtered using a Savitzky–Golay filter [13] with 51 window sizes and a polynomial order of 3. The two sets of values extracted from the video investigations (intensity for the parenchyma and intensity for the lesions) were saved into comma separated values (csv) files. After saving the values into csv files, the following values were extracted from the TIC: maximum intensity, time to peak (TTP), area under the curve (AUC), and mean transit time (MTT). The entire flow described earlier has two purposes. The main purpose was to extract all the values to build a custom dataset. The second purpose was to be able to reuse parts of the method to predict the lesion type based on the video investigation file and patient data. The dataset fields are presented in Table 4, and the labels are presented in Table 5. This dataset was then used to train a feed-forward neural network model. The proposed neural network architecture was a feed-forward neural network with eight neurons as input layer, two hidden layers with six neurons each, and 5 neurons as the output layer (one neuron for each output class presented in Table 5). The architecture of the proposed classifier is presented in Figure 5. When choosing the architecture of the neural network model, both the inference time and performance of the model were considered. The activation functions used were rectified linear unit (ReLu) for all the layer except for the output layer. For this layer, a Softmax activation function was used. The dataset created was imbalanced ($40.67\%$ of the cases were diagnosed with hepatocarcinoma). To deal with class imbalance, oversampling techniques were applied. Therefore, random samples from each class in the dataset were duplicated. This technique has been proven to be effective when dealing with class imbalance [14,15]. The dataset was split in $70\%$ samples for training and $30\%$ samples for validation. The loss function used in our proposed feed-forward classifier was focal cross-entropy (FCE) proposed by Tsung-Yi Lin et al. [ 16]. Typically, in a multiclass classification problem, the cross-entropy loss function (Equation [2]) is used. Being a logarithmic loss function, cross-entropy introduces a large penalty for errors close to 1 and smaller penalties when the error is closer to 0. However, the cross-entropy loss function penalized all the classes equally, including the imbalanced classes. [ 2]CE=−∑$i = 1$Ntilogpi where ti is the expected label, pi represents the probability (Softmax output function) for ith class, and N is the total number of classes. Focal cross-entropy reduced the loss for well-classified samples and increased the loss for samples defined as “hard-to-classify” [16]. Focal cross-entropy was used in object detection tasks in which the class imbalance introduced a very difficult problem. Objects in the background were hard to detect if the object detection model was trained using cross-entropy [16]. FCE is given by Equation [3]. [ 3]FCE=−∑$i = 1$Nαii−piγlogpi where N represents the total number of classes, pi represents the probability for the ith class, α is the weighting factor, and γ represents the focusing parameter. The values for the weighting factor and the focusing parameter were chosen as 0.25 and 2.0, respectively. The proposed feed-forward model was trained for 100 epochs with a batch size of 50, and a learning rate of 0.0001 with the optimizer RMSProp [17]. The training was performed in the cloud, using Google Colab [18] without a graphical processing unit (GPU) and with Tensorflow version 2.9.2 [19] and Python version 3.8. The segmentation model was trained on Nvidia 3050Ti GPU with Tensorflow version 2.7.0 [19]. The feed-forward neural network from the current study was a multi-class classifier. Therefore, the performance metrics applied to binary classifiers could not be used to correctly evaluate this model. For the evaluation of the model, categorical accuracy, F1 macro, F1 micro, and Matthews correlation coefficient (MCC) were used. Categorical accuracy is defined as the number of correct predictions divided by the total number of predictions. If n represents the total number of correct predictions and N the total number of predictions, categorical accuracy (CA) is defined by Equation [4]. [ 4]CA=nN F1 micro and F1 macro scores are defined as the harmonic mean between precision micro and recall micro, respectively, precision macro and recall macro. Assuming the following notations: TP—true positives, FP—false positives, FN—false negatives, and N—the total number of classes, F1 micro and F1 macro are given by Equations [5]–[10]. [ 5]Prcμ=∑$i = 1$NTPi∑$i = 1$NTPi+FPi [6]PrcMacro=∑$i = 1$NTPiTPi+FPiN [7]Recallμ=∑$i = 1$NTPi∑$i = 1$NTPi+FNi [8]RecallMacro=∑$i = 1$NTPiTPi+FNiN [9]F1μ=2 Prcμ· RecallμPrcμ+Recallμ [10]F1Macro=2 PrcMacro· RecallMacroPrcMacro+RecallMacro Matthews correlation coefficient for N classes is given by Equation [11] in which N represents the total number of classes, c represents the total number of samples predicted correctly, S is the total number of samples in the batch, ti represents the number of appearances for class I, and pi is the number of predictions for class i. [11]MCC=c ·S−∑$i = 1$Npi·tiS2−∑$i = 1$Npi2·s2−∑$i = 1$Nti2 ## 4. Results The results presented in this chapter are divided into two parts: the results obtained by extracting the time-intensity curves and the results obtained by the proposed method. For the TIC, several curves are presented for different types of lesions. All the curves are extracted using the proposed method. In Figure 6A, a TIC extracted with the proposed method is presented. The video investigation file was from a patient diagnosed with hepatocarcinoma. Figure 6B shows the same TIC filtered with Savitzky–Golay. In Figure 7A, another TIC extracted with the proposed method is presented. The patient was diagnosed with hemangioma. Figure 7B presents the same TIC but with the filtering applied. The results for both training and validation are presented below in Table 6, and the evolution of the loss during training and validation is presented in Figure 8. ## 5. Discussion The purpose of the presented study is to introduce an automated method of classifying liver lesions based on time-intensity curves and deep learning algorithms in CEUS video investigations. The entire system includes a U-Net segmentation model [11] and a feed-forward neural network model. The segmentation model was trained on B-mode frames extracted from contrast-enhanced ultrasound video investigations. A more detailed description of the model can be found in our previous published study [10]. A feed-forward neural network model was then trained with data from two sources: values extracted from the time-intensity curves and clinical information of the patient. Five output labels were defined for the proposed system: hepatocarcinoma, metastasis, other malignant lesions, hemangioma, and other benign lesions as per liver lesions in our dataset. The ultrasound investigations were performed by the Department of Gastroenterology of the Emergency Clinical County Hospital of Craiova using Hitachi Arietta V 70, convex probe C250. The study involved 49 patients with 59 liver lesions. Most of the patients were male, 31 aged between 38 and 85. The mean tumor size was 51.65 mm, and $22.44\%$ of the patients had a history of previous malignancy. The lesions in the study included hepatic hemangioma, liver cysts, focal nodular hyperplasia, liver adenoma, liver abscess, hepatocellular carcinoma, liver metastases, cholangiocarcinoma, and malignant liver adenoma. The most dominant diagnosis was hepatocellular carcinoma ($40.67\%$), while the least dominant diagnosis was liver adenoma ($1.69\%$). A more detailed description of the patient cohort was presented in our previous studies [10,20]. Being modular, the proposed system can be enhanced to detect other liver lesions by performing transfer learning only on the feed-forward neural network model. In their study, Hang-Tong Hu et al. [ 21] trained a deep learning model to classify focal liver lesions as malignant or benign. Their cohort contained 363 patients with CEUS video investigations from four phases: plain scan, arterial image, portal image, and delayed image. Residual neural network architecture (ResNet) was used. Their results showed that the trained DL model had a performance comparable with senior radiologists [21]. In other study, Kaizhi Wu et al. [ 22] introduced a classification system of liver lesions based on CEUS investigations. Their study contained 22 patients with 26 lesions. For extracting the TICs, sparse non-negative matrix factorizations were used, and for classifying the lesions, the TICs were analyzed using a DL model. Their presented results showed that the proposed method outperformed other classifiers such as support vector machine (SVM) or K-nearest neighbors (KNN). B. Schmauch et al. [ 23] proposed a method for diagnosis of focal liver lesions in ultrasound investigations based on a ResNet50 DL model with an attention block. While their dataset contained only 367 liver images, the results presented showed that the DL model can perform class separation (benign and malignant) with high accuracy. Cătălin Daniel Căleanu et al. [ 24] proposed a convolutional neural network architecture which achieved high accuracy in classifying liver lesions in CEUS investigations. Compared with other studies, their proposed method could classify an investigation into five different classes (hepatocellular carcinoma, hypervascular metastases, hypovascular metastases, hemangiomas, and focal nodular hyperplasia). Our proposed method contains a TIC extraction component and a feed-forward neural network classifier. Therefore, the presented method analyzes the CEUS video investigation (for extracting the TIC) and the clinical data for each patient. It requires minimal intervention from the human operator as the frames are extracted from each video investigation file. The results presented in Table 6 and Table 7 show that the feed-forward classifier cannot predict with high accuracy the exact type of lesion. However, the model can perform class separation between malignant and benign lesions. Two random samples were selected for analyzing: one from the malignant lesions and one from the benign lesions. The video investigations for each patient were run though the proposed system. The results are presented in Table 7. The first selected patient (PATIENT1) was a female, 79 years old, with a cirrhotic liver. In our dataset, the patient was diagnosed with hepatocarcinoma. The second patient (PATIENT2) was a 62-year-old male with no history of cirrhosis or hepatitis. In our dataset, this patient was diagnosed with hepatic adenoma. The results presented in Table 6 and Table 7 show that the proposed system is able to perform class separation between malignant and benign. However, in some cases, the system has difficulties detecting a specific lesion in these classes, especially in benign ones. This can be due to relatively low samples of benign lesions in our dataset. While other studies [22] proposed binary classifiers (benign or malignant lesions), the presented system can classify liver lesions from CEUS video investigations into five classes. Besides CEUS video investigations, it uses clinical data of the patient to perform the classification. Because it uses a DL image segmentation model, it requires no intervention from the human operator during data analysis for prediction. U-Net, the image segmentation model used in the proposed system, was tested in our previous study [10] in terms of execution time and GPU requirements. It can perform an image segmentation in between 32.50 and 76.43 milliseconds (depending on the GPU) [10]. The feed-forward classifier added in the current study has 203 trainable parameters. Thus, together with the TIC extraction module, the group adds a small overhead to the entire system. Analyzing the evolution of the ultrasound signal over time is not new in the medical field. However, generating the graphical representation manually is a difficult task and can introduce errors [25,26]. Automatically extracting TICs from CEUS liver investigations by using either image processing techniques or AI algorithms is an active research area. Simona Turco et al. [ 27] proposed a method with minimal manual input from the operator. The authors avoided motion compensation by using spatial and spatiotemporal analysis. While the metrics observed in their study showed that it can separate classes with high accuracy, the method could classify the lesion into two classes: malignant or benign. A limitation of our study is the unbalanced classes. Separating the labels from the dataset from binary (malignant or benign) to multiclass (five classes—three malignant and two benign) unbalanced the dataset even more. To counter this, we used the focal cross-entropy loss function, which is usually used in highly unbalanced classes. In addition, we applied oversampling techniques to the dataset. ## 6. Conclusions The aim of the study was to develop an automated method for classifying liver lesions in CEUS video investigations. A dataset provided by the University of Medicine and Pharmacy Craiova with 50 CEUS video investigations was used. The proposed method included three components or modules: a lesion segmentation module, a TIC extraction module, and a classifier module. A U-Net segmentation model was trained to perform lesion segmentation in B-mode frame-by-frame from the video investigation. By using this segmentation model, tracking the lesion’s movement in the CEUS investigation was not an issue because the segmentation model predicted the mask frame-by-frame. Because we knew the coordinates of each window type in the video frame, B-mode or contrast mode images could be extracted from it. Dilation was applied on each analyzed frame, and together with the predicted mask from the segmentation model, a TIC could be extracted. A feed-forward neural network model was trained with the parameters extracted from the TICs and with clinical data from each patient in the study. The results show that due to class imbalance, the classifier was not able to predict with high accuracy a specific lesion from the minority classes. However, in the majority of classes, the classifier could predict the lesion type with high accuracy. The proposed system can be a useful tool for gastroenterologists or medical students to either provide a prediction of the lesion type or to automatically extract TICs. ## References 1. Lv B., Liu F., Gou F., Wu J.. Multi-Scale Tumor Localization Based on Priori Guidance-Based Segmentation Method for Osteosarcoma MRI Images. Mathematics (2022.0) 10. DOI: 10.3390/math10122099 2. Wu J., Xiao P., Huang H., Gou F., Zhou Z., Dai Z.. An Artificial Intelligence Multiprocessing Scheme for the Diagnosis of Osteosarcoma MRI Images. IEEE J. Biomed. Health Informatics (2022.0) 26 4656-4667. DOI: 10.1109/JBHI.2022.3184930 3. Georgeanu V.A., Mămuleanu M., Ghiea S., Selișteanu D.. Malignant Bone Tumors Diagnosis Using Magnetic Resonance Imaging Based on Deep Learning Algorithms. Medicina (2022.0) 58. DOI: 10.3390/medicina58050636 4. Nica R.-E., Șerbănescu M.-S., Florescu L.-M., Camen G.-C., Streba C.T., Gheonea I.-A.. Deep Learning: A Promising Method for Histological Class Prediction of Breast Tumors in Mammography. J. Digit. Imaging (2021.0) 34 1190-1198. DOI: 10.1007/s10278-021-00508-4 5. Dietrich C.F., Nolsøe C.P., Barr R.G., Berzigotti A., Burns P.N., Cantisani V., Chammas M.C., Chaubal N., Choi B.I., Clevert D.-A.. Guidelines and Good Clinical Practice Recommendations for Contrast-Enhanced Ultrasound (CEUS) in the Liver–Update 2020 WFUMB in Cooperation with EFSUMB, AFSUMB, AIUM, and FLAUS. Ultrasound Med. Biol. (2020.0) 46 2579-2604. DOI: 10.1016/j.ultrasmedbio.2020.04.030 6. Jung E.-M., Weber M.-A., Wiesinger I.. Contrast-enhanced ultrasound perfusion imaging of organs. Radiologe (2021.0) 61 19-28. DOI: 10.1007/s00117-021-00891-7 7. Marrero J.A., Kulik L.M., Sirlin C.B., Zhu A.X., Finn R.S., Abecassis M.M., Roberts L.R., Heimbach J.K.. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology (2018.0) 68 723-750. DOI: 10.1002/hep.29913 8. Galle P.R., Forner A., Llovet J.M., Mazzaferro V., Piscaglia F., Raoul J.-L., Schirmacher P., Vilgrain V.. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J. Hepatol. (2018.0) 69 182-236. DOI: 10.1016/j.jhep.2018.03.019 9. Bankhead P., Loughrey M.B., Fernández J.A., Dombrowski Y., McArt D.G., Dunne P.D., McQuaid S., Gray R.T., Murray L.J., Coleman H.G.. QuPath: Open source software for digital pathology image analysis. Sci. Rep. (2017.0) 7 16878. DOI: 10.1038/s41598-017-17204-5 10. Mămuleanu M., Urhuț C.M., Săndulescu L.D., Kamal C., Pătrașcu A.-M., Ionescu A.G., Șerbănescu M.-S., Streba C.T.. Deep Learning Algorithms in the Automatic Segmentation of Liver Lesions in Ultrasound Investigations. Life (2022.0) 12. DOI: 10.3390/life12111877 11. Ronneberger O., Fischer P., Brox T., Navab N., Hornegger J., Wells W.M., Frangi A.F.. U-Net: Convolutional networks for biomedical image segmentation. Medical Image Computing and Computer-Assisted Intervention 2015 (2015.0) 234-241. DOI: 10.1007/978-3-319-24574-4_28 12. Salvatore V., Borghi A., Sagrini E., Galassi M., Gianstefani A., Bolondi L., Piscaglia F.. Quantification of enhancement of focal liver lesions during contrast-enhanced ultrasound (CEUS). Analysis of ten selected frames is more simple but as reliable as the analysis of the entire loop for most parameters. Eur. J. Radiol. (2012.0) 81 709-713. DOI: 10.1016/j.ejrad.2011.01.097 13. Press W.H., Teukolsky S.A.. Savitzky-Golay Smoothing Filters. Comput. Phys. (1990.0) 4 669. DOI: 10.1063/1.4822961 14. Butaru A.E., Mămuleanu M., Streba C.T., Doica I.P., Diculescu M.M., Gheonea D.I., Oancea C.N.. Resource Management through Artificial Intelligence in Screening Programs—Key for the Successful Elimination of Hepatitis C. Diagnostics (2022.0) 12. DOI: 10.3390/diagnostics12020346 15. Mamuleanu M., Ionete C., Albita A., Selisteanu D.. Distributed Deep Learning Model for Predicting the Risk of Diabetes, Trained on Imbalanced Dataset. Proceedings of the 2022 23rd International Carpathian Control Conference (ICCC) 315-318 16. Lin T.Y., Goyal P., Girshick R., He K., Dollar P.. Focal loss for dense object detection. IEEE Trans. Pattern Anal. Mach. Intell. (2020.0) 42 318-327. DOI: 10.1109/TPAMI.2018.2858826 17. Hinton G., Srivastava N., Swersky K.. Neural Networks for Machine Learning Lecture 6a Overview of Mini-Batch Gradient Descent 18. Google Colaboratory 19. Abadi M., Barham P., Chen J., Chen Z., Davis A., Dean J., Devin M., Ghemawat S., Irving G., Isard M.. TensorFlow: A System for Large-Scale Machine Learning. Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16) 265-283 20. Ciocalteu A., Iordache S., Cazacu S.M., Urhut C.M., Sandulescu S.M., Ciurea A.-M., Saftoiu A., Sandulescu L.D.. Role of Contrast-Enhanced Ultrasonography in Hepatocellular Carcinoma by Using LI-RADS and Ancillary Features: A Single Tertiary Centre Experience. Diagnostics (2021.0) 11. DOI: 10.3390/diagnostics11122232 21. Hu H.T., Wang W., da Chen L., Ruan S.M., Chen S.L., Li X., Lu M.-D., Xie X.Y., Kuang M.. Artificial intelligence assists identifying malignant versus benign liver lesions using contrast-enhanced ultrasound. J. Gastroenterol. Hepatol. (2021.0) 36 2875-2883. DOI: 10.1111/jgh.15522 22. Wu K., Chen X., Ding M.. Deep learning based classification of focal liver lesions with contrast-enhanced ultrasound. Optik (2014.0) 125 4057-4063. DOI: 10.1016/j.ijleo.2014.01.114 23. Schmauch B., Herent P., Jehanno P., Dehaene O., Saillard C., Aubé C., Luciani A., Lassau N., Jégou S.. Diagnosis of focal liver lesions from ultrasound using deep learning. Diagn. Interv. Imaging (2019.0) 100 227-233. DOI: 10.1016/j.diii.2019.02.009 24. Căleanu C.D., Sîrbu C.L., Simion G.. Deep Neural Architectures for Contrast Enhanced Ultrasound (CEUS) Focal Liver Lesions Automated Diagnosis. Sensors (2021.0) 21. DOI: 10.3390/s21124126 25. Ignee A., Jedrejczyk M., Schuessler G., Jakubowski W., Dietrich C.F.. Quantitative contrast enhanced ultrasound of the liver for time intensity curves—Reliability and potential sources of errors. Eur. J. Radiol. (2010.0) 73 153-158. DOI: 10.1016/j.ejrad.2008.10.016 26. Streba C.T.. Contrast-enhanced ultrasonography parameters in neural network diagnosis of liver tumors. World J. Gastroenterol. (2012.0) 18 4427. DOI: 10.3748/wjg.v18.i32.4427 27. Turco S., Tiyarattanachai T., Ebrahimkheil K., Eisenbrey J., Kamaya A., Mischi M., Lyshchik A., El Kaffas A.. Interpretable Machine Learning for Characterization of Focal Liver Lesions by Contrast-Enhanced Ultrasound. IEEE Trans. Ultrason. Ferroelectr. Freq. Control. (2022.0) 69 1670-1681. DOI: 10.1109/TUFFC.2022.3161719
--- title: Heparanase Modulates Chromatin Accessibility authors: - Honglian Li - Hua Zhang - Amelie Wenz - Ziqi Kang - Helen Wang - Israel Vlodavsky - Xingqi Chen - Jinping Li journal: Cells year: 2023 pmcid: PMC10047235 doi: 10.3390/cells12060891 license: CC BY 4.0 --- # Heparanase Modulates Chromatin Accessibility ## Abstract Heparanase is the sole endoglucuronidase that degrades heparan sulfate in the cell surface and extracellular matrix (ECM). Several studies have reported the localization of heparanase in the cell nucleus, but the functional role of the nuclear enzyme is still obscure. Subjecting mouse embryonic fibroblasts (MEFs) derived from heparanase knockout (Hpse-KO) mice and applying transposase-accessible chromatin with sequencing (ATAC-seq), we revealed that heparanase is involved in the regulation of chromatin accessibility. Integrating with genome-wide analysis of chromatin states revealed an overall low activity in the enhancer and promoter regions of Hpse-KO MEFs compared with wild-type (WT) MEFs. Western blot analysis of MEFs and tissues derived from Hpse-KO vs. WT mice confirmed reduced expression of H3K27ac (acetylated lysine at N-terminal position 27 of the histone H3 protein). Our results offer a mechanistic explanation for the well-documented attenuation of inflammatory responses and tumor growth in Hpse-KO mice. ## 1. Introduction Heparanase, the sole mammalian heparan sulfate (HS)-degrading endoglycosidase, belongs to the wider class of enzymes known as ‘retaining glycosidases,’ which catalyze hydrolytic cleavage of glycosidic bonds with net retention of anomeric stereochemistry [1]. It employs a conserved ‘double displacement mechanism’ involving two key catalytic amino acid residues—a nucleophile (Glu343) and a general acid/base proton donor (Glu225)—and transient formation of a covalent enzyme–substrate intermediate during the catalytic cycle [1]. The heparin/HS binding domains (HBD1, HBD2) are situated close to the active site micro-pocket fold. The heparanase mRNA encodes a 65 kDa pro-enzyme that is cleaved by cathepsin L into 8 and 50 kDa subunits that non-covalently associate to form the active enzyme. Structurally, heparanase is composed of a TIM-barrel fold that contains the enzyme’s active site and a flexible C-terminus domain required for secretion and the signaling function of the protein. Heparanase regulates multiple biological activities in cancers, e.g., enhancing tumor growth, angiogenesis, and metastasis [2,3,4]. Heparanase expression is found to be elevated in almost all cancers examined including various carcinomas, sarcomas, and hematological malignancies. Numerous clinical association studies have consistently demonstrated that upregulation of heparanase expression correlates with increased tumor size, tumor angiogenesis, enhanced metastasis, and poor prognosis. In contrast, knockdown of heparanase or treatments of tumor-bearing mice with heparanase-inhibiting compounds markedly attenuate tumor progression, further underscoring the potential of anti-heparanase therapy for multiple types of cancer. Heparanase-neutralizing monoclonal antibodies block myeloma and lymphoma tumor growth and dissemination, attributable to a combined effect on the tumor cells and/or cells of the tumor microenvironment [5]. In fact, much of the impact of heparanase on tumor progression is related to its function in mediating tumor–host crosstalk, priming the tumor microenvironment to better support tumor growth, metastasis, and chemoresistance [4]. Given their heterogeneity and versatility, heparan sulfate proteoglycans (HSPGs) serve as important functional components of the cell surface and ECM. Hence, cleavage of HS by heparanase not only contributes to disassembly of the ECM, thereby facilitating cancer metastasis, but also affects diverse physiological and pathological processes ranging from gene transcription to DNA damage [6]. The known repertoire of the physio-pathological activities of heparanase is expanding. It activates cells of the innate immune system, promotes the formation of exosomes and autophagosomes, and stimulates signal transduction pathways via enzymatic and non-enzymatic activities. These effects dynamically impact multiple regulatory pathways that together drive tumor survival, growth, dissemination, and drug resistance [2,4,7,8]. Collectively, the emerging premise is that heparanase expressed by tumor cells, innate immune cells, activated endothelial cells, and other cells of the tumor microenvironment is a master regulator of the aggressive phenotype of cancer, an important contributor to the poor outcome of cancer patients, and a prime target for therapy. Apart from its role in cancer outcomes described in cancer-related studies, heparanase has also been found to play a significant role in modulating inflammatory responses [9,10]. The enzyme appears to fulfill some normal functions associated, for example, with vesicular traffic, lysosomal-based secretion, stress response, and heparan sulfate turnover. Based on these findings, heparanase inhibitors are developed for treatment of cancers and inflammatory diseases. Heparin-like compounds that inhibit heparanase activity are being evaluated in clinical trials for various types of cancer. Heparanase-neutralizing monoclonal antibodies are being evaluated in pre-clinical studies, and heparanase-inhibiting small molecules have been developed based on the crystal structure of the heparanase protein. A key avenue through which heparanase accomplishes its multiple effects on cells and tissues is by regulating the bioavailability of HS-bound growth factors, chemokines, and cytokines, priming the tissue microenvironment. HS resides on the cell surface and the ECM in the form of proteoglycans (HSPGs), where, among other activities, it modulates cytokine bioavailability and functions as a co-receptor [11]. Binding of cytokines/growth factors to their specific cell membrane receptors is promoted by HS, followed by endocytosis of the growth factor–HS–receptor complex into late endosomes and lysosomes where it is catabolized by endo- and exo-glycosidases [12]. In this way, heparanase mediates tumor–host crosstalk, and promotes basic cellular processes (i.e., exosome formation, autophagy, and immune responses) that together orchestrate tissue remodeling [13]. Several studies have demonstrated the localization of HS and HSPGs in the cell nucleus [14,15]. It is unknown whether HS and HSPGs are transported into the cell nucleus after internalization and/or directly from the Golgi after synthesis. Nevertheless, it was demonstrated that exogenously added HS is transported into the nucleus [16], suggesting that secreted HS can be internalized into the cell nucleus. Likewise, heparanase has been found in the cell nucleus [17]. Nuclear heparanase is thought to fulfill functional roles, such as regulating chromatin remodeling [18], chromosome stability [19], and gene transcription [20], depending or not depending on heparanase’s enzymatic activity. Early on, heparanase was believed to function primarily as a metabolic enzyme involved in HS/heparin catabolism [21]. Later on, a number of non-enzymatic activities (i.e., signal transduction, tumorigenesis, and gene transcription) were identified and ascribed to the enzyme C-terminus domain [22] and/or nuclear localization. However, little is known about the role of heparanase and HS within the nucleus. Using myeloma cell lines, it was found that: (i) within the nucleus, heparanase is present in the soluble fraction, and it is also bound to insoluble chromatin; (ii) the presence of nuclear heparanase enhances acetylation of histone H3 and promotes an open chromatin conformation; (iii) heparanase binds the promoter region of syndecan-1, MMP9, and CCND1, three genes whose expression is upregulated by heparanase; and (iv) heparanase increases phosphorylation of PTEN, leading to enhanced PTEN stability and thereby diminishing its function as a tumor suppressor [18]. Examination of available gene expression databases reveals that myeloma patients with high heparanase expression exhibited enhanced expression of acetyltransferase complexes and signaling pathways associated with myeloma growth and progression [18]. Together, these findings indicate that nuclear heparanase plays an important role in tumorigenesis by promoting chromatin remodeling that opens its conformation, allowing access to promotors of genes that drive tumor progression. To gain insight regarding the role of nuclear heparanase, we applied the ATAC-seq for genome-wide analysis of the transposase-accessible chromatin genome-wide [23] in mouse embryonic fibroblasts (MEFs) isolated from wild-type (WT) vs. heparanase knockout (Hpse-KO) [24] mice. The results show an overall low activity in the enhancer and promoter regions in Hpse-KO MEFs compared with WT MEFs. Significantly lower levels of H3K27ac were detected in Hpse-KO embryonic cells and organs in comparison with WT tissues. ## 2. Materials and Methods Preparation of mouse embryonic fibroblasts (MEFs): Heparanase knockout (Hpse-KO) (C57BL/6 genetic background) mice were generated as previously described [24] and maintained by littermates breeding in the animal facility of the Biomedical Center, Uppsala University, Sweden. Wild-type (WT) C57BL/6 mice were used as control. Embryonic fibroblasts (MEFs) were isolated from embryos of WT and Hpse-KO mice at day 14.5, as described [25]. Purified cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Thermo Fisher, Waltham, MA, USA, 11995073) containing $10\%$ FBS (Thermo Fisher, 10500064) and 100 U/mL penicillin–streptomycin (Thermo Fisher, 15070063) in a humidified atmosphere containing $5\%$ CO2 at 37 °C. The experimental protocol was conducted according to local ethical regulations. Transposase-accessible chromatin using ATAC-seq: ATAC-seq was performed as previously described [26]. Briefly, both Hpse-KO and WT cells [50,000] were applied per ATAC-seq reaction. The transposition reaction was carried out according to the ATAC-seq protocol [26]. After transposition, the DNA was purified with MinElute PCR Purification kit (Qiagen, Hilden, Germany, 28004) and eluted in 10 µL Qiagen EB elution buffer. Sequencing libraries were prepared as described in the original ATAC-seq protocol. Sequencing was performed on Illumina NovaSeq 6000, and at least 20 million paired-end sequencing reads were generated for each ATAC-seq library. Data processing: ATAC-seq raw data were mapped to mm10 with Bowtie2. The reads with mapping quality over 30 were kept and duplication reads were removed with Picard. MACS2 was used to perform peak calling with -q 0.01 -nomodel -shift 0 parameters. The peaks that overlapped with ENCODE genome blacklist were discarded. The read counts within peaks were generated with bedtools and the genome tracks were visualized using vig IGV. Differential peak analysis and genomic annotation: Differential peak analysis was performed with DEseq2 package, keeping the differential peak list with threshold values of \|logFC\| > 1, FDR < 0.01. Genomic annotation was performed with the ChIPseeker package in R with seven genomic features: 3′ UTR, 5′ UTR, exon, intergenic region, intron, TSS, and TTS. The ChIP-seq data for ChromHMM were downloaded from Gene Expression Omnibus with the number GSE90895 and only the wild-type samples were downloaded. In total, 15 states were identified: Genic Enhancer 1 (EnhG1), Strong Transcription (Tx), Bivlant Transcription (BivTx), Repressed Polycomb (ReprPC), Bivlant Enhancer (EnhBiv), Weak Enhancer (EnhWk), Weak Transcription (TxWk), Active enhancer 1 (EnhA1), Genic Enhancer 2 (EnhG2), Active enhancer 2 (EnhA2), Active TSS (TssA), Flanking TSS upstream (TssFlnkU), Quiescent/low (Quies), ZNF genes & repeats (ZNF/Rpts), and Heterochromatin (Het). Bedtools documentation was applied to perform overlapping and to assign chromatin states into peaks. KEGG analysis was conducted with clusterProfiler package in R. Super-enhancer calling and function characterization: Super-enhancers were identified with ROSE software with H3K27ac ChIP-seq data (GSE90803). Bedtools were used to analyze overlapping peaks between super-enhancer regions and chromatin-accessible regions. Ontological enrichment was performed by g: Profiler and visualized by Cytoscape setting FDR range from 0 to 0.001. Homer mm10 database and the super-enhancer bed files were used to calculate motif occurrences with a cut-off of p-value < 0.01. Western blotting: Tissues of embryos and adult organs were collected from mice and homogenized with Polytron in ice-cold lysis buffer containing 20 mM Tris pH 7.4, 150 mM NaCl, $1\%$ NP-40, $0.5\%$ Sodium Deoxycholate, $0.1\%$ SDS, and protease inhibitors. Cells were lysed in the same buffer and incubated on ice for 30 min. After centrifugation, the supernatants were collected and protein concentration was determined using the BCA method (Thermo Fisher, 23227). The lysates of tissues (60 µg total protein) and cells (20 µg total protein) were heated at 98 °C for 5 min before SDS-PAGE, followed by transferring to a polyvinylidene fluoride (PVDF) membrane (Cytiva, Marlborough, MA, USA, 10600023). After blocking with $5\%$ non-fat milk for 1 h, the membranes were probed sequentially with primary and secondary antibodies. Antibody dilutions were: anti-H3K27ac 1:2500 (Abcam, Cambridge, UK, ab4729), anti-H3K27me3 1:1000 (Abcam, ab6002), anti-H3K4me1 1:1000 (Abcam, ab8895), anti-H3K9me3 1:1000 (Abcam, ab8898), anti-H3 1:1000 (Abcam, ab1791), and HRP-conjugated goat anti-rabbit IgG 1:10000 (Thermo Fisher, 31460). Signals were visualized using chemiluminescent HRP substrate (Sigma, Saint Louis, MS, USA, WBKLS0500). Band intensity was quantified by the Image Lab software (Bio-Rad, Hercules, CA, USA). ## Statistical Analysis Statistical significance was analyzed by the 2-tailed unpaired Student’s t-test. Values of $p \leq 0.05$ were considered significant. ## 3.1. Suppressed Activity of Promoters in Hpse-KO MEFs Using the primary MEFs derived from Hpse-KO and WT mice, we collected two sets of ATAC-seq data. Examination of the heparanase gene loci identified enriched peaks in the heparanase promoter region of WT cells, which were missing in Hpse-KO cells, confirming the null expression of heparanase in the Hpse-KO cells (Figure 1A). Detection of the Coq2 gene is shown as a reference (Figure 1A). The high quality of ATAC-seq was reflected by highly concordant enrichment of the sequence reads at the transcription start sites (TSS) (Figure 1B) and the fraction of reads in peaks (FRiP) (Figure 1C). Furthermore, heatmap analysis showed good reproducibility of ATAC-seq from both WT and Hpse-KO samples (Figure 1D). ## 3.2. Identification and Characterization of Significantly Differential Peaks A total of 13,000 ATAC-seq peaks downregulated or upregulated were identified (Figure 2A,B). Representative different peak regions are displayed with the genome browser (Figure 2C). To understand global functional changes in Hpse-KO cells, we systematically annotated the genomic feature distribution of ATAC-seq peaks and differential peaks and found the distinctive accessible chromatin patterns between WT and Hpse-KO cells (Figure 2D). A dramatic downregulation of the genes in the promoter region of Hpse-KO cells indicates that expression of a large number of genes may be reduced. In contrast, significant upregulation was detected in the distal intergenic region (Figure 2D). As this region contains multiple enhancers or inhibitory elements and is important in the regulation of gene expression, detailed genomic annotation is required to reveal the functional elements in this region of the Hpse-KO cells. Using chromHMM to characterize detailed chromatin states, we found 15 states that are located at the TSS and Flanking TSS upstream region (Figure 2E). Notably, among the 15 states, there are six histone markers, of which all are H3 with different modifications. The 15 chromatin states were used to re-annotate the ATAC-seq peaks and the significantly differential peaks to highlight differences in the distribution of the chromatin states (Figure 2F). In Hpse-KO cells, the quiescent (Quies) regions of chromosomes increased significantly and an increase at the weak enhancer (EnhWk) region was also observed in comparison with WT cells (Figure 2F). In contrast, active enhancers (EnhA1, EnhA2) and transcription start sites (TssA, TssFlnkU) regions were significantly reduced in the Hpse-KO vs. WT cells. KEGG enrichment analysis of peaks in the quiescent region revealed a significantly enriched pathway of neuroactive ligand–receptor interaction. In comparison, KEGG enrichment analysis of the peaks in the weak enhancer regions showed major enrichment of several signaling pathways, including PI3K-Akt, MAPK focal adhesion, and calcium signaling (Figure 2G). In contrast, the active enhancer (EhA) and transcription start site (TSS) regions were mostly downregulated in Hpse-KO cells. KEGG analysis detected enrichment of a number of pathways, including PI3K-Akt, MAPK, and Ras signaling. These pathways are also enriched in the weak enhancer regions in Hpse-KO vs. WT cells (Figure 2H). ## 3.3. Suppressed Super-Enhancer in Hpse-KO MEFs Representative SE-associated genes are marked as rank ordering of super-enhancers (Figure 3A). Gene ontology enrichment (GO) analysis of the SE-associated genes is shown in Figure 3B. Ontological analysis of SE-associated genes revealed their relationship with genes involved in important biological processes, including several biosynthetic and metabolic pathways as well as protein phosphorylation. The heatmap shows noticeably reduced chromatin accessibility levels of super-enhancers in Hpse-KO cells by overlapping and calculating peak counts (Figure 3C). The lower chromatin accessibility in Hpse-KO vs. WT cells can be seen in the super-enhancer region (e.g., Slc11a1) (Figure 3D). The pattern of super-enhancer transcription factor binding sequence obtained by motif analysis is shown in Figure 3E. ## 3.4. Histone Modifications of H3 in Hpse-KO MEFs and Organs To determine if the epigenetic analysis of H3 in the EhA and TSS regions of Hpse-KO samples is of relevance to biological functions, we examined histone modification of H3 by Western blot analysis of the MEF cells. Using several anti-H3K antibodies recognizing different epitopes, we found that the level of H3K27ac was significantly reduced, while the level of H3K27me3 was substantially elevated in the Hpse-KO compared with WT cells. There was no difference in the levels of H3K4me1 or H3K9me3 between the two cell types (Figure 4A). These results are largely in agreement with the epigenetic analysis results. To reveal whether the reduced level of H3K27ac has a biological impact in mice, we analyzed H3K27ac in tissues dissected from the Hpse-KO and WT mice. Examination of brain, lung, heart, kidney, spleen, and pancreas did not detect a difference between Hpse-KO and WT organs. Interestingly, the level of H3K27ac was lower in the eyes of Hpse-KO compared with WT mice (Figure 4B). Examination of embryos under different developmental stages revealed a significantly lower level of H3K27ac in whole Hpse-KO vs. WT embryos (Figure 5), possibly indicating a role of H3K27ac in Hpse-KO mice during embryonic development. ## 4. Discussion The sole endoglucuronidase in mammalians, heparanase, was thought to play a vital role in HS catabolism; surprisingly, elimination of the heparanase gene in mice did not lead to accumulation of HS, and only resulted in averagely longer HS chains [24]. The single observed phenotypic defect in Hpse-KO mice was in the retinal pigment epithelium [27], which may be associated with a low level of H3K27ac in the eyes (Figure 4B). Obviously, it appears that heparanase is not critical for HS catabolism and has no vital biological functions during embryonic development. The overall normal development of the Hpse-KO mice seems unaffected by the lowered level of H3K27ac detected in embryos (Figure 5). Importantly, heparanase also displays non-enzymatic biological actions (e.g., cytokine production, signal transduction) [28]. Several observed impacts of heparanase expression on cellular signaling may or may not depend on the enzymatic activity [29,30]. The observed nuclear localization of heparanase further emphasizes the potential of heparanase’s non-enzymatic functions apart from its HS-degrading activity. Several studies have reported regulatory activities of nuclear heparanase, such as the induction of mammary cancer cell differentiation [31], regulation of glucose metabolism in endothelial cells [32], and gene transcription [33]. Some of these functions are attributed to the modulatory role of heparanase in chromatin packaging and remodeling [18], collectively, suggesting a stimulatory effect of nuclear heparanase on gene expression. To gain some insight into the epigenetic regulation of heparanase function, we applied the advanced technique of ATAC-seq (assay for transposase-accessible chromatin using sequencing) to analyze the local accessibility of chromatin in embryonic fibroblasts (MEFs) derived from Hpse-KO and WT mice. Since the regions of gene transcription are controlled by cis-acting DNA elements, including enhancers, silencers, and promoters, we compared gene expression profiles in these regions. Super-enhancer is a special region in mammalian genomes driving transcription of important genes, controlling and defining cell identity. The EhA and TSS regions are chromatin states enriched with H3K27ac, the most frequently used marker of super-enhancer recognition [34]. Genes marked with this broad epigenetic domain, including H3K27 modification, are involved in cell identity and essential functions with strong clinical relevance [35]. The observed decreased level of H3K27ac in Hpae-KO MEFs and embryos is in line with previous reports showing that H3 methylation at actively transcribed genes is affected by heparanase [20] and that histone acetyltransferase (HAT) activity is modulated by heparanase through degradation of syndecan-1 [36]. Increased HAT activity in Hpse-high myeloma cells resulted in upregulation of transcription of multiple genes (i.e., VEGF, HGF, MMP-9, RANKL) that drive an aggressive tumor phenotype [36]. A novel set of genes under heparanase regulation has been characterized in T cells [20]. In this context, nuclear heparanase was shown to regulate the transcription of a cohort of inducible immune response genes by controlling histone H3 methylation, further expanding the transcriptional potential of heparanase. Heparanase was also shown to interact with promoters of multiple genes and micro-RNAs that upregulate gene transcription and control T cell differentiation. During herpes simplex virus-1 infection of corneal epithelial cells, heparanase translocates to the nucleus and enhances cytokine production [37]. The combined involvement of heparanase in epigenetic gene regulation and tumor progression has been further elucidated in studies demonstrating that chemotherapy, in addition to its cytotoxic effects on tumor cells, can support tumor re-growth and spread [38,39,40]. The possible involvement of heparanase in this observation is supported by our previous studies showing that heparanase expression is increased substantially in myeloma patients and cells treated with chemotherapy [40], providing a strong rationale for applying anti-heparanase therapy in combination with conventional anti-cancer drugs. In a subsequent study, we found that macrophages, an important constituent of the tumor microenvironment, are activated efficiently by chemotherapy (i.e., paclitaxel, cisplatin) and thereby support tumor growth. Strikingly, cytokine induction by chemotherapy was not observed in macrophages isolated from Hpse-KO mice [41]. Mechanistically, we found that chemotherapy (paclitaxel) stimulates the methylation of histone H3 on lysine 4 (H3K4) in wild-type but not Hpse-KO macrophages, leading to cytokine induction, and involving WDR5 [41]. This result provides another example of the involvement of heparanase in epigenetic gene regulation through histone modification. Collectively, these results suggest a regulatory role of heparanase in chromatin packaging, accessibility, and activity. The increased activity of histone acetyltransferases (HAT) in cells expressing high levels of heparanase and the suppressed expression of H3K27ac and H3K4 tri-methylation (histone methyltransferases) in Hpse-KO cells/tissues suggest an important regulatory function of heparanase in histone modification and activity. The overall low activity in regions that control gene transcription in Hpse-KO MEFs may be associated with suppressed inflammatory responses and tumor growth in Hpse-KO mice [10,42,43]. Indeed, overexpression of heparanase in mice promotes tumor initiation and growth [44,45] and stimulates inflammatory responses [7,9,46]. Collectively, we propose that heparanase may function as an epigenetic marker, modulating innate immunity and other cellular functions, in part by controlling gene expression. Given that heparanase is a well-recognized target for inhibition of cancer progression and inflammation [43,47], our findings further strengthen this notion by broadening the multitasking molecular function of the heparanase gene and protein. ## References 1. Wu L., Viola C.M., Brzozowski A.M., Davies G.J.. Structural characterization of human heparanase reveals insights into substrate recognition. Nat. Struct. Mol. Biol. (2015) 22 1016-1022. DOI: 10.1038/nsmb.3136 2. Jayatilleke K.M., Hulett M.D.. Heparanase and the hallmarks of cancer. J. Transl. Med. (2020) 18 453. DOI: 10.1186/s12967-020-02624-1 3. Mayfosh A.J., Nguyen T.K., Hulett M.D.. The Heparanase Regulatory Network in Health and Disease. Int. J. Mol. Sci. (2021) 22. DOI: 10.3390/ijms222011096 4. Vlodavsky I., Singh P., Boyango I., Gutter-Kapon L., Elkin M., Sanderson R.D., Ilan N.. Heparanase: From basic research to therapeutic applications in cancer and inflammation. Drug Resist. Updat. (2016) 29 54-75. DOI: 10.1016/j.drup.2016.10.001 5. Weissmann M., Arvatz G., Horowitz N., Feld S., Naroditsky I., Zhang Y., Ng M., Hammond E., Nevo E., Vlodavsky I.. Heparanase-neutralizing antibodies attenuate lymphoma tumor growth and metastasis. Proc. Natl. Acad. Sci. USA (2016) 113 704-709. DOI: 10.1073/pnas.1519453113 6. Agelidis A., Suryawanshi R.K., Patil C.D., Campeau A., Gonzalez D.J., Shukla D.. Dissociation of DNA damage sensing by endoglycosidase HPSE. iScience (2021) 24 102242. DOI: 10.1016/j.isci.2021.102242 7. Mayfosh A.J., Goodall K.J., Nguyen T., Baschuk N., Hulett M.D.. Heparanase is a regulator of natural killer cell activation and cytotoxicity. J. Leukoc. Biol. (2022) 111 1211-1224. DOI: 10.1002/JLB.3A0420-259RR 8. Karamanos N.K., Piperigkou Z., Passi A., Gotte M., Rousselle P., Vlodavsky I.. Extracellular matrix-based cancer targeting. Trends Mol. Med. (2021) 27 1000-1013. DOI: 10.1016/j.molmed.2021.07.009 9. Lerner I., Hermano E., Zcharia E., Rodkin D., Bulvik R., Doviner V., Rubinstein A.M., Ishai-Michaeli R., Atzmon R., Sherman Y.. Heparanase powers a chronic inflammatory circuit that promotes colitis-associated tumorigenesis in mice. J. Clin. Investig. (2011) 121 1709-1721. DOI: 10.1172/JCI43792 10. Schmidt E.P., Yang Y., Janssen W.J., Gandjeva A., Perez M.J., Barthel L., Zemans R.L., Bowman J.C., Koyanagi D.E., Yunt Z.X.. The pulmonary endothelial glycocalyx regulates neutrophil adhesion and lung injury during experimental sepsis. Nat. Med. (2012) 18 1217-1223. DOI: 10.1038/nm.2843 11. Xie M., Li J.P.. Heparan sulfate proteoglycan—A common receptor for diverse cytokines. Cell. Signal. (2019) 54 115-121. DOI: 10.1016/j.cellsig.2018.11.022 12. Freeman C., Hopwood J.. Lysosomal degradation of heparin and heparan sulphate. Adv. Exp. Med. Biol. (1992) 313 121-134. PMID: 1442257 13. Sanderson R.D., Bandari S.K., Vlodavsky I.. Proteases and glycosidases on the surface of exosomes: Newly discovered mechanisms for extracellular remodeling. Matrix Biol. (2019) 75–76 160-169. DOI: 10.1016/j.matbio.2017.10.007 14. Ishihara M., Fedarko N.S., Conrad H.E.. Transport of heparan sulfate into the nuclei of hepatocytes. J. Biol. Chem. (1986) 261 13575-13580. DOI: 10.1016/S0021-9258(18)67058-9 15. Stewart M.D., Sanderson R.D.. Heparan sulfate in the nucleus and its control of cellular functions. Matrix Biol. (2014) 35 56-59. DOI: 10.1016/j.matbio.2013.10.009 16. Li M., Yuan S., Chen J., Lv L., Guo S., Zhou W.. Design and synthesis of structurally defined heparan sulfate (HS)-FK506 conjugates as an exogenous approach to investigate biological functions of nucleus HS. Bioorg. Chem. (2018) 81 203-210. DOI: 10.1016/j.bioorg.2018.08.015 17. Schubert S.Y., Ilan N., Shushy M., Ben-Izhak O., Vlodavsky I., Goldshmidt O.. Human heparanase nuclear localization and enzymatic activity. Lab. Investig. J. Tech. Methods Pathol. (2004) 84 535-544. DOI: 10.1038/labinvest.3700084 18. Amin R., Tripathi K., Sanderson R.D.. Nuclear Heparanase Regulates Chromatin Remodeling, Gene Expression and PTEN Tumor Suppressor Function. Cells (2020) 9. DOI: 10.3390/cells9092038 19. Yang S., Liao Y., Zhao Q., Xie Y., Zheng A., Wan H.. Heparanase Is a Critical Regulator of Mitotic Spindles Required for Maintaining Chromosome Stability. DNA Cell Biol. (2018) 37 291-297. DOI: 10.1089/dna.2017.3990 20. He Y.Q., Sutcliffe E.L., Bunting K.L., Li J., Goodall K.J., Poon I.K., Hulett M.D., Freeman C., Zafar A., McInnes R.L.. The endoglycosidase heparanase enters the nucleus of T lymphocytes and modulates H3 methylation at actively transcribed genes via the interplay with key chromatin modifying enzymes. Transcription (2012) 3 130-145. DOI: 10.4161/trns.19998 21. Lindahl U., Li J.P.. Heparanase—Discovery and Targets. Adv. Exp. Med. Biol. (2020) 1221 61-69. PMID: 32274706 22. Fux L., Ilan N., Sanderson R.D., Vlodavsky I.. Heparanase: Busy at the cell surface. Trends Biochem. Sci. (2009) 34 511-519. DOI: 10.1016/j.tibs.2009.06.005 23. Buenrostro J.D., Giresi P.G., Zaba L.C., Chang H.Y., Greenleaf W.J.. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat. Methods (2013) 10 1213-1218. DOI: 10.1038/nmeth.2688 24. Zcharia E., Jia J., Zhang X., Baraz L., Lindahl U., Peretz T., Vlodavsky I., Li J.P.. Newly generated heparanase knock-out mice unravel co-regulation of heparanase and matrix metalloproteinases. PLoS ONE (2009) 4. DOI: 10.1371/journal.pone.0005181 25. Jia J., Maccarana M., Zhang X., Bespalov M., Lindahl U., Li J.P.. Lack of L-iduronic acid in heparan sulfate affects interaction with growth factors and cell signaling. J. Biol. Chem. (2009) 284 15942-15950. DOI: 10.1074/jbc.M809577200 26. Chen X., Shen Y., Draper W., Buenrostro J.D., Litzenburger U., Cho S.W., Satpathy A.T., Carter A.C., Ghosh R.P., East-Seletsky A.. ATAC-see reveals the accessible genome by transposase-mediated imaging and sequencing. Nat. Methods (2016) 13 1013-1020. DOI: 10.1038/nmeth.4031 27. Van Bergen T., Etienne I., Jia J., Li J.P., Vlodavsky I., Stitt A., Vermassen E., Feyen J.H.M.. Heparanase Deficiency Is Associated with Disruption, Detachment, and Folding of the Retinal Pigment Epithelium. Curr. Eye Res. (2020) 46 1166-1170. DOI: 10.1080/02713683.2020.1862239 28. Riaz A., Ilan N., Vlodavsky I., Li J.P., Johansson S.. Characterization of heparanase-induced phosphatidylinositol 3-kinase-AKT activation and its integrin dependence. J. Biol. Chem. (2013) 288 12366-12375. DOI: 10.1074/jbc.M112.435172 29. Cohen-Kaplan V., Jrbashyan J., Yanir Y., Naroditsky I., Ben-Izhak O., Ilan N., Doweck I., Vlodavsky I.. Heparanase induces signal transducer and activator of transcription (STAT) protein phosphorylation: Preclinical and clinical significance in head and neck cancer. J. Biol. Chem. (2012) 287 6668-6678. DOI: 10.1074/jbc.M111.271346 30. Batool T., Fang J., Barash U., Moustakas A., Vlodavsky I., Li J.P.. Overexpression of heparanase attenuated TGF-beta-stimulated signaling in tumor cells. FEBS Open Bio. (2017) 7 405-413. DOI: 10.1002/2211-5463.12190 31. Nobuhisa T., Naomoto Y., Takaoka M., Tabuchi Y., Ookawa K., Kitamoto D., Gunduz E., Gunduz M., Nagatsuka H., Haisa M.. Emergence of nuclear heparanase induces differentiation of human mammary cancer cells. Biochem. Biophys. Res. Commun. (2005) 331 175-180. DOI: 10.1016/j.bbrc.2005.03.129 32. Wang F., Wang Y., Zhang D., Puthanveetil P., Johnson J.D., Rodrigues B.. Fatty acid-induced nuclear translocation of heparanase uncouples glucose metabolism in endothelial cells. Arterioscler. Thromb. Vasc. Biol. (2012) 32 406-414. DOI: 10.1161/ATVBAHA.111.240770 33. de Mestre A.M., Khachigian L.M., Santiago F.S., Staykova M.A., Hulett M.D.. Regulation of inducible heparanase gene transcription in activated T cells by early growth response 1. J. Biol. Chem. (2003) 278 50377-50385. DOI: 10.1074/jbc.M310154200 34. Hnisz D., Abraham B.J., Lee T.I., Lau A., Saint-Andre V., Sigova A.A., Hoke H.A., Young R.A.. Super-enhancers in the control of cell identity and disease. Cell (2013) 155 934-947. DOI: 10.1016/j.cell.2013.09.053 35. Beacon T.H., Delcuve G.P., Lopez C., Nardocci G., Kovalchuk I., van Wijnen A.J., Davie J.R.. The dynamic broad epigenetic (H3K4me3, H3K27ac) domain as a mark of essential genes. Clin. Epigenetics (2021) 13 138. DOI: 10.1186/s13148-021-01126-1 36. Purushothaman A., Hurst D.R., Pisano C., Mizumoto S., Sugahara K., Sanderson R.D.. Heparanase-mediated loss of nuclear syndecan-1 enhances histone acetyltransferase (HAT) activity to promote expression of genes that drive an aggressive tumor phenotype. J. Biol. Chem. (2011) 286 30377-30383. DOI: 10.1074/jbc.M111.254789 37. Agelidis A.M., Hadigal S.R., Jaishankar D., Shukla D.. Viral Activation of Heparanase Drives Pathogenesis of Herpes Simplex Virus-1. Cell Rep. (2017) 20 439-450. DOI: 10.1016/j.celrep.2017.06.041 38. Shaked Y.. Balancing efficacy of and host immune responses to cancer therapy: The yin and yang effects. Nat. Rev. Clin. Oncol. (2016) 13 611-626. DOI: 10.1038/nrclinonc.2016.57 39. Gingis-Velitski S., Loven D., Benayoun L., Munster M., Bril R., Voloshin T., Alishekevitz D., Bertolini F., Shaked Y.. Host response to short-term, single-agent chemotherapy induces matrix metalloproteinase-9 expression and accelerates metastasis in mice. Cancer Res. (2011) 71 6986-6996. DOI: 10.1158/0008-5472.CAN-11-0629 40. Ramani V.C., Vlodavsky I., Ng M., Zhang Y., Barbieri P., Noseda A., Sanderson R.D.. Chemotherapy induces expression and release of heparanase leading to changes associated with an aggressive tumor phenotype. Matrix Biol. (2016) 55 22-34. DOI: 10.1016/j.matbio.2016.03.006 41. Bhattacharya U., Gutter-Kapon L., Kan T., Boyango I., Barash U., Yang S.M., Liu J., Gross-Cohen M., Sanderson R.D., Shaked Y.. Heparanase and Chemotherapy Synergize to Drive Macrophage Activation and Enhance Tumor Growth. Cancer Res. (2020) 80 57-68. DOI: 10.1158/0008-5472.CAN-19-1676 42. Gutter-Kapon L., Alishekevitz D., Shaked Y., Li J.P., Aronheim A., Ilan N., Vlodavsky I.. Heparanase is required for activation and function of macrophages. Proc. Natl. Acad. Sci. USA (2016) 113 E7808-E7817. DOI: 10.1073/pnas.1611380113 43. Barash U., Lapidot M., Zohar Y., Loomis C., Moreira A., Feld S., Goparaju C., Yang H., Hammond E., Zhang G.. Involvement of Heparanase in the Pathogenesis of Mesothelioma: Basic Aspects and Clinical Applications. J. Natl. Cancer Inst. (2018) 110 1102-1114. DOI: 10.1093/jnci/djy032 44. Boyango I., Barash U., Fux L., Naroditsky I., Ilan N., Vlodavsky I.. Targeting heparanase to the mammary epithelium enhances mammary gland development and promotes tumor growth and metastasis. Matrix Biol. (2018) 65 91-103. DOI: 10.1016/j.matbio.2017.08.005 45. Zhang G.L., Gutter-Kapon L., Ilan N., Batool T., Singh K., Digre A., Luo Z., Sandler S., Shaked Y., Sanderson R.D.. Significance of host heparanase in promoting tumor growth and metastasis. Matrix Biol. (2020) 93 25-42. DOI: 10.1016/j.matbio.2020.06.001 46. Hermano E., Goldberg R., Rubinstein A.M., Sonnenblick A., Maly B., Nahmias D., Li J.P., Bakker M.A.H., van der Vlag J., Vlodavsky I.. Heparanase Accelerates Obesity-Associated Breast Cancer Progression. Cancer Res. (2019) 79 5342-5354. DOI: 10.1158/0008-5472.CAN-18-4058 47. Nguyen T.K., Paone S., Chan E., Poon I.K.H., Baxter A.A., Thomas S.R., Hulett M.D.. Heparanase: A Novel Therapeutic Target for the Treatment of Atherosclerosis. Cells (2022) 11. DOI: 10.3390/cells11203198
--- title: Promising Antimicrobial Action of Sustained Released Curcumin-Loaded Silica Nanoparticles against Clinically Isolated Porphyromonas gingivalis authors: - Adileh Shirmohammadi - Solmaz Maleki Dizaj - Simin Sharifi - Shirin Fattahi - Ramin Negahdari - Mohammad Ali Ghavimi - Mohammad Yousef Memar journal: Diseases year: 2023 pmcid: PMC10047251 doi: 10.3390/diseases11010048 license: CC BY 4.0 --- # Promising Antimicrobial Action of Sustained Released Curcumin-Loaded Silica Nanoparticles against Clinically Isolated Porphyromonas gingivalis ## Abstract Background. Porphyromonas gingivalis (P. gingivalis) has always been one of the leading causes of periodontal disease, and antibiotics are commonly used to control it. Numerous side effects of synthetic drugs, as well as the spread of drug resistance, have led to a tendency toward using natural antimicrobials, such as curcumin. The present study aimed to prepare and physicochemically characterize curcumin-loaded silica nanoparticles and to detect their antimicrobial effects on P. gingivalis. Methods. Curcumin-loaded silica nanoparticles were prepared using the chemical precipitation method and then were characterized using conventional methods (properties such as the particle size, drug loading percentage, and release pattern). P. gingivalis was isolated from one patient with chronic periodontal diseases. The patient’s gingival crevice fluid was sampled using sterile filter paper and was transferred to the microbiology laboratory in less than 30 min. The disk diffusion method was used to determine the sensitivity of clinically isolated P. gingivalis to curcumin-loaded silica nanoparticles. SPSS software, version 20, was used to compare the data between groups with a p value of <0.05 as the level of significance. Then, one-way ANOVA testing was utilized to compare the groups. Results. The curcumin-loaded silica nanoparticles showed a nanometric size and a drug loading percentage of $68\%$ for curcumin. The nanoparticles had a mesoporous structure and rod-shaped morphology. They showed a relatively rapid release pattern in the first 5 days. The release of the drug from the nanoparticles continued slowly until the 45th day. The results of in vitro antimicrobial tests showed that P. gingivalis was sensitive to the curcumin-loaded silica nanoparticles at concentrations of 50, 25, 12.5, and 6.25 µg/mL. One-way ANOVA showed that there was a significant difference between the mean growth inhibition zone, and the concentration of 50 µg/mL showed the highest inhibition zone (p ≤ 0.05). Conclusion. Based on the obtained results, it can be concluded that the local nanocurcumin application for periodontal disease and implant-related infections can be considered a promising method for the near future in dentistry. ## 1. Introduction The main causes of periodontal diseases are inflammation and infection of the gums and bone surrounding the teeth. In the early stage of periodontal disease, which is called gingivitis, the gums become swollen and red and may bleed. In more advanced stages, called periodontitis, the gums can separate from the teeth, the bone can be lost, and the teeth can become loose or even fall out. The two biggest threats to dental health are tooth decay and periodontal disease [1,2]. These diseases affect the tissues supporting and protecting the teeth and can deteriorate the alveolar bone and periodontal ligament. Their frequency and severity vary greatly between communities; nonetheless, it is expected that 15 to 20 percent of adults are infected with the more severe types of illness, while 35 to 60 percent of the population is afflicted with less severe conditions [3,4]. The most important goal of treating periodontitis is to completely clean the pockets around the teeth and inhibition of surrounding bone damage. If the periodontitis has not progressed much, treatment may include less invasive and nonsurgical methods, including root planing (smoothing surfaces of the root, preventing further accumulation of bacteria and plaque), scaling (eliminating bacteria and plaque from surfaces of the tooth), and the oral or topical application of antibiotics [5]. Patients with advanced periodontitis may need dental surgery for treatment, including pocket reduction surgery (flap surgery), bone grafting, soft tissue grafting, and the regeneration of guided tissue [6]. Bacterial infections in dentistry may induce implant-associated issues, resulting in tissue and organ function loss or even implant failure. In dentistry, bacterial infections contribute to the development of caries and periodontitis, which are two of the most prevalent bacterial infections in humans [7]. Numerus bacterial strains are involved in periodontal disease development, such as Aggrigatibacter actinomycetemcomitans, Capnocytophaga species and Eikenella corrodens, P. gingivalis, Tannerella forsythia, Treponema denticola, Prevotella intermedia, Actinomyces species, and Fusobacterium nucleatum [8]. Porphyromonas gingivalis (P. gingivalis) has always been one of the leading causes of periodontal disease, and antibiotics are commonly used to control it. Numerous systems releasing antibacterial and remineralizing substances, such as fluoride (F), calcium (Ca2+) as well as phosphate (PO43–), or silver (Ag+) ions, have been described for effective avoidance or treatment of biofilm infections [9,10]. Chlorhexidine (CHX) is frequently used because of its excellent antibacterial efficacy against Gram-negative and Gram-positive microorganisms, fungi, and viruses [11]. Furthermore, the propensity to produce resistance is minor [12]. Preservatives are also included in disinfection products and oral rinses [13]. Moreover, by attaching to the enamel and pellicle, CHX suppresses the production of bacterial biofilms. The first stage in creating biofilms, i.e., the aggregation of bacterial cells on these surfaces, is inhibited [13,14]. CHX has a significant substantivity that indicates a long-term interaction with particular substrates, such as tooth surfaces or mucosa inside the oral cavity [14]. This elevated CHX is the gold standard for dentistry microbial infection prevention and treatment [14]. However, tooth staining is one drawback of CHX, which limits its long-term usage. There are also other side effects, including tongue and mucosal surface staining, changes of taste, desquamation of the mucosa, expansion of the parotid and enlarged calculus deposition supragingivally [15]. In addition, the spread of drug resistance has led to a tendency toward using new natural antimicrobials [16]. The active ingredients in plants are widely used in the treatment of various diseases [17,18]. Curcumin is a substance produced from the rhizomes of the Curcuma longa plant, and it is commonly utilized in culinary applications [17,18]. A wide variety of publications have reported on curcumin’s anti-inflammatory, wound-healing, antimicrobial, and anti-neoplastic properties, utilized in in vitro and in vivo strategies for conditions ranging from diabetes to neurological disturbances, in cancer, in autoimmune disorders, and in chronic inflammatory conditions, including Crohn’s disease, rheumatoid arthritis, and periodontal disease [19]. Curcumin’s anti-inflammatory properties have been found to diminish immune cell response to periodontal disease-associated bacterial antigens and to restrict periodontal tissue destruction in in vitro and in vivo studies [19,20]. Nevertheless, since most of this in vivo research has utilized a systemic manner of administration, curcumin’s poor pharmacodynamic properties, including hydrophobicity, low gastrointestinal absorption rate, and very short plasma half-life, may have skewed their results [21]. New designs based on nanotechnology have been discovered to improve the bioavailability of curcumin and reduce its cytotoxicity [22]. Today, nanotechnology has become important in various medical fields, such as drug delivery [23]. Nanoporous silica materials have been extensively studied [24,25,26] since their initial deployment as a drug delivery platform in 2001 [27] or as implant surface coatings [28,29]. Nanoporous silica has a variety of qualities that make it an attractive option for a controlled-release system. It has a large surface area, huge pore volumes, and variable pore sizes with contracted pore size distributions, allowing for significant cargo loading. On the other hand, uncontrolled antimicrobial chemical leaching from release mechanisms has disadvantages. Although burst release could benefit the treatment of acute infections, and it is much more efficient than protracted delivery, it is essential for controlled release systems that can stay quiescent for lengthy periods yet distribute cargo when triggered. As a result, the medicine remains in the pores and could be removed when required. Due to the antimicrobial properties of curcumin and the useful characteristics of porous silica nanoparticles as a sustained-release carrier, the present study was conducted with the aim of preparing and physicochemically identifying curcumin-loaded silica nanoparticles and evaluating their antimicrobial effect on P. gingivalis. ## 2.1. Preparation of Mesoporous Silica Nanoparticles Containing Curcumin Fifteen milligrams of powder of silica nanoparticles (Nano Sadra Company, Isfahan, Iran) and 0.75 mg of curcumin powder (Sigma Aldrich, Burlington, MA, USA) were added to 10 mL of cyclohexane. The prepared suspension was sonicated, stirred overnight, and washed with cyclohexane, and the silica particles containing curcumin were vacuum dried [30]. The nanoparticles were stored at −18 °C for further investigations. ## 2.2. Sampling of P. gingivalis To attain clinically isolated P. gingivalis, one patient with chronic periodontal disease was selected from the patients referred to the Department of Periodontics, Faculty of Dentistry, Tabriz University of Medical Sciences, Tabriz, Iran. With sterile gauze, the surface of the tooth was cleaned, and the gingival crevice fluid was then sampled using sterile filter paper and placed in a thioglycollate broth media. The samples were moved to the microbiology laboratory in less than 30 minutes and stored at −20 °C until assayed. ## 2.3. Cultivation of P. gingivalis The isolated sample from the mentioned patient was vortexed for 30 s. Selective medium for P. gingivalis containing Columbia agar base supplemented with vitamin K1, $5\%$ defibrillated sheep blood, hemin, colistin sulfate, bacitracin, and nalidixic acid was used [31]. Then, the plates were incubated under $80\%$ N2, $10\%$ CO2, $10\%$ H2 and $0\%$ O2 in anaerobic conditions provided by the Anoxomat system (MART microbiology B.V., Drachten, The Netherlands). The growth of bacterial colonies was examined at 48, 72, and 96 h. The trypsin reagent test was used to confirm the presence of P. gingivalis on the plates. Gingipain, which is produced by P. gingivalis, is a trypsin-like enzyme. The aerotolerance test and biochemical and microbiological assays (such as colony morphology, special potency disks, pigment production, fluorescent under UV light, catalase test, indole, and trypsin-like peptidase activity assay) were used to identify P. gingivalis isolates [31]. ## 2.4.1. The Particle Size of Nanoparticles The prepared nanoparticles were characterized using a dynamic light scattering (DLS) device (DLS, Malvern, Cambridge, UK) for size determination. The suspension of the nanoparticles was prepared in distilled water and poured into the device. An argon laser beam at 633 nm and a scattering angle of 90° at 25 °C were used for DLS device settings. DLS is an instrument for measuring the hydrodynamic size of molecules and submicron and nanoparticles. This test was performed three times. ## 2.4.2. Morphology and the Cytotoxicity Investigation Transmission electron microscopy (TEM) is a powerful tool to investigate the interaction of nanoparticles, their structure, and their morphology. A transmission electron microscope (TEM-2100F; JEOL, Tokyo, Japan) was used to investigate the mesoporous structure of the silica nanoparticles. For this analysis, the samples were prepared by dropping a solution of nanoparticles in deionized water on a carbon-coated copper TEM grid, followed by imaging. Size histograms for free silica nanoparticles and curcumin-loaded silica, based on TEM analysis, were also reported. Cell viability examination was used to define the cytotoxicity of the prepared nanoparticles against dental pulp stem cells. The cells were obtained from the cell bank of Shahid Beheshti University (Tehran, Iran). Then, the nanoparticles as disks were placed in the bottoms of the wells. The cells were cultured in a single layer in DMEM including serum and antibiotics. After 72 h, the washing, incubating (for 4 h at 37 °C), and adding of MTT solution (2 mg/mL PBS) were performed. As a next step, the above solution was removed and, 200 mL of DMSO and 25 mL of Sorenson glycine buffer were added to each well. The absorbance was read at 540 nm, and the percentage of living cells was evaluated. Cells grown without any material were considered as control group. ## 2.4.3. Determination of Curcumin Loading Inside the Nanoparticles One of the key parameters for drug-loaded nanoparticles is drug loading percentage, which is defined as the mass ratio of drug to drug-loaded nanoparticles. To determine the amount of curcumin loaded on silica nanoparticles, 10 mg of the prepared nanoparticles were dissolved in 20 mL of dimethyl sulfoxide. One milliliter of the dissolved nanoparticle solution was poured into a special tube of an ultraviolet spectrophotometer, and Lambda Max was adjusted to 350 nm for curcumin. This test was performed three times. ## 2.4.4. Evaluation of Release Pattern Drug release denotes the procedure in which drug solutes migrate from the initial position in the carrier system to the carrier’s outer surface and then to the release medium. To determine the pattern of drug release from curcumin-loaded silica nanoparticles, phosphate buffer (300 mL) was poured into 3 beakers. An amount of 5 mg of the prepared nanoparticles was poured into the beaker. The pH of the liquid was adjusted to 7.4, and the temperature was set to 37 °C. The stirrer was set to 100 rpm. Indeed, these parameters had to be established based on the body’s condition for a dissolution test of a drug (pH of 7.4, temperature of 37 °C, and stirring rate of 100 rpm). Samples were taken from the beaker every day (1 mL), and the absorbance was noted using a UV spectrophotometer for curcumin at 350 nm. The sample taken from the beakers was replaced with 1 mL of a new buffer medium to keep the concentration in balance. The amount of UV absorption was then changed to concentration. Subsequently, the cumulative release percentage was designed against the time (day) for the release study. The calculation method for the percentage of cumulative release (%) was according to the following equation:Cumulative percentage release (%) = Volume of sample withdrawn (mL)/The volume of release media (v) × P (t − 1) + Pt where Pt is the percentage release at time t. ## 2.4.5. The Antimicrobial Action of Nanoparticles The original method for determining susceptibility to antimicrobials was based on broth dilution methods. In this study, the disk diffusion method as a routine laboratory test was utilized to investigate the antibacterial effects of silica nanoparticles loaded with curcumin. This method identifies the action of bacteria on an antimicrobial material by creating a gradient of concentration around a disk. The bacterial isolate used in this study was isolated from a patient with chronic periodontal disease. First, a bacterial suspension of 0.5 McFarland was prepared, and then, using a sterile cotton swab, a uniform grass culture was grown on the surface of Brucella agar enriched with dried sheep blood ($5\%$), vitamin K1 (1 μg/mL), and hemin (5 μg in mL). To prepare discs containing nanoparticles, sterile blank disks were immersed in concentrations of 3.12, 6.25, 12.5, 25, and 50 μg/mL nanoparticle suspensions, and then the disks were placed on the agar surface. A blank disk was used as a negative control, and metronidazole antibiotic disks (5 μg/mL) were used as a positive control. After incubating the plates at 37 °C for 42 h, the growth inhibition zones were measured. With this method, the halos of non-growth around the discs were measured from the back of the plate with a ruler based on millimeters. In the next step, Brucella broth supplemented with hemin (5 µg/mL), vitamin K1 (1 µg/mL), and lysed horse blood ($5\%$) in the presence of a serial concentration of nanoparticles (50, 25, 12.5, and 6.25 µg/mL concentrations) was applied to obtain the MICs of the nanoparticles against P. gingivalis. The wells were incubated for 48 h at 35 °C and then observed for microbial growth turbidity. The positive control was metronidazole antibiotic, and water was considered as a negative control. ## 3. Statistical Analysis The results are stated as descriptive indices. The Shapiro–Wilk test was applied to test the normality of the units. The, we used SPSS software, version 20 (IBM Company, Armonk, NY, USA), to compare the data between groups with a p value of <0.05 as the significance level. One-way ANOVA and Tukey’s post hoc test were utilized to compare the groups. The flow chart of the study process is shown in Figure 1. ## 4. Results and Discussion The low bioavailability of curcumin is the most important concern for its clinical use. Additionally, little information is available about its safety at higher doses. Today, to reduce its toxicity and improve the bioavailability of curcumin, new designs based on its nanoformulation have been discovered [17,18]. Evaluating the physicochemical properties of nanoparticles is necessary to ensure their suitability for various uses. The interactions of nanoparticles in vitro and in vivo are related to their physicochemical properties [32]. Reducing the size of nanoparticles increases their surface area, the interaction of these nanoparticles with the environment increases, and their ways of crossing body barriers and entering cells will be different [33,34]. The average particle size of drug-free silica nanoparticles is shown in Figure 2a, and that for curcumin-loaded silica nanoparticles is shown in Figure 2b. The results showed that both types of nanoparticles had nanometric sizes. For drug-free silica nanoparticles the mean particles size was 90 ± 1.02 nm, while curcumin-loaded silica nanoparticles had a mean particle size of 110 ± 1.23 nm. Figure 3a shows the morphology of the drug-free silica nanoparticles, and the morphology of curcumin-loaded silica nanoparticles has shown in Figure 3b. The size histograms for free silica nanoparticles and curcumin-loaded silica, based on TEM analysis, are shown also in the Figure 3c and d, respectively. Our outcomes showed that the nanoparticle sizes differed in DLS analysis compared to TEM analysis. This difference may be owing to the hydrating of the outer layer of the nanoparticles in the DLS technique. In addition, the aggregation of nanoparticles and the non-spherical shape of nanoparticles could be the cause of this difference [35]. Nanoparticles exert their antimicrobial effects on bacteria by several mechanisms that depend on the size of the nanoparticles and the type of bacteria. The dose of nanoparticles and their physicochemical properties (shape, size and surface properties) are very important to their antimicrobial effects [36]. The size of nanoparticles is important to their antibacterial effect, so smaller nanoparticles, by binding to the surface of bacteria with high affinity, can disrupt the function of the cell membrane of bacteria compared to larger nanoparticles [37]. The interaction of nanoparticles with the bacterial membrane causes local pores in the membrane. Additionally the entry of nanoparticles into bacterial cells causes damage to DNA and proteins (especially sulfur-rich proteins). In this way, nanoparticles can disrupt the function of bacteria. Nanocarriers containing antibacterial agents can also combine their structure with the bacterial cell wall and introduce their medicinal substances into the cytoplasm [38]. TEM pictures proved the mesoporous building and the rod-shaped morphology of the prepared nanoparticles. The filled pores of mesoporous silica can also be detected by TEM imaging of drug-loaded mesoporous silica nanoparticles that show the loading of curcumin into the silica nanoparticles. Rod-shaped nanoparticles may display a longer circulation time and a slight uptake by the RES in the body compared with spherical particles [39,40]. A recent in vivo study also showed that rod-type nanoparticles exhibit a high capacity to overcome uptake through RES and show a longer presence in the blood compared with spherical nanoparticles [41]. The percentage of cytotoxicity (cell viability) of the prepared nanoparticles on dental pulp stem cells is shown in Figure 4. There was no significant reduction in the viability of the cells exposed to the nanoparticles compared to the control group (cells grown without any material). Therefore, the prepared nanoparticles were non-cytotoxic against dental pulp stem cells (Figure 4). The loading results showed that the loading percentage of curcumin in silica nanoparticles was $68\%$ ± 1.02. Currently, most nanoparticle systems have relatively low drug loading, and increasing the increase drug loading capacity remains a challenge. The reason for the high drug-loading percentage of our nanoparticles was their mesoporous structure. The prepared nanoparticles displayed a relatively fast release pattern in the first 5 days (Figure 5). The release of curcumin from silica nanoparticles continued slowly until day 45. The burst release of curcumin from the prepared nanoparticles could eradicate acute infections, and the controlled sustained release could provide the drug content for long periods. As a result, the drug remained in the pores and could be removed when required [42]. It seems that the pattern of rapid drug release from nanoparticles in the first days is related to drugs adsorbed to the surface of nanoparticles that are not inside the cavities and have a weak interaction with the outer surface of the cavities. Curcumin molecules inside the cavities that had electrostatic interactions with the nanoparticle cavity wall caused slow and continuous release on days 6 to 45. The slow-release pattern of drugs is very critical in the clinical application of drugs [43]. Memar et al. achieved similar results for meropenem-loaded silica nanoparticles [44]. They showed that, in the first two days, about 40 percent of meropenem was released from silica nanoparticles, and then slow release was sustained until the 30th day. With a conventional drug-delivery method, the drug concentration in the blood remains within a relatively large range for a short period of time, which can fall short of the lowest effective dose or exceed the maximum tolerated dose. As a result, frequent doses are necessary, which will be associated with side effects. Using the appropriate nanocarrier, the blood concentration of the drug at the site of infection can be maintained at the required effective concentration for a long time and, as a result, reduce the frequency of consumption, produce good stability, reduce patient pain, and improve patient compliance. The drug loaded in the nanocarrier has a much more prominent inhibitory effect on cell growth with long-term drug release compared to the free drug at the same concentration [45]. ## Antimicrobial Action The results of microbial tests showed that P. gingivalis is sensitive to the silica nanoparticles loaded with curcumin at concentrations of 50, 25, 12.5, and 6.25 μg/mL. The mean growth inhibition zones of curcumin-loaded silica nanoparticles concentrations and control antibiotic (metronidazole) are shown in Table 1 and Figure 6. Based on the MIC test, the nanoparticles showed inhibitory effects against P. gingivalis at 6.25 µL/mL. In addition, based on our previous study, free silica nanoparticles did not have any significant antibacterial effects [46]. One-way ANOVA (between curcumin groups) revealed that there is a significant relation in the concentration of curcumin-loaded silica nanoparticles with the size of the growth inhibition, zone and the highest inhibition zone was displayed in the concentration of 50 µg/mL (p ≤ 0.05). Tukey’s post hoc test showed that there was a significant difference between the antimicrobial effects of all concentrations of curcumin-loaded silica nanoparticles (p ≤ 0.05). Thus, the nanoparticles had dose-dependent antimicrobial effects. Other studies used P. gingivalis (ATCC33277). In a study, Shahzad et al. reported that the growth inhibition of P. gingivalis (ATCC33277) was effected by curcumin at a concentration of 7.8 μg/mL [47]. Additionally, Mandroli and Bhat showed that the MIC of curcumin against P. gingivalis (ATCC33277) was 125 μg/mL [48], while Izui et al. showed that the prevention of bacterial growth occurred with curcumin at a concentration of 20 μg/mL [49]. In another recent study, the sensitivity of P. gingivalis (ATCC33277) to curcumin was shown in a concentration of 100 μg/mL [50]. The main reason for the difference between the results of our study and the results of other studies may be that they investigated the effects of free curcumin on laboratory strains, while in our study, the effects of sustained-release nanoparticles containing curcumin on clinically isolated P. gingivalis were investigated. In our previous study, the prevalence of P. gingivalis isolated from the gingival crevicular fluid (GCF) of 15 Iranian patients with implant failure was investigated. The results showed that, out of 15 patients, eight ($53.33\%$) were positive for the presence of P. gingivalis. The antimicrobial action of curcumin nanocrystals was also investigated against P. gingivalis isolated from patients with implant failure, and the results showed that curcumin nanocrystals had an MBC of 12.5 µg/mL and a MIC of 6.25 µg/mL. Additionally curcumin nanocrystals showed the highest inhibition zone at the concentration of 50 µg/mL ($$p \leq 0.0003$$) [51]. A study showed that curcumin prevented bacterial strains by damaging the membrane of bacteria [52]. Curcumin can inhibit the proliferation of bacteria by perturbation of FtsZ assembly. Some studies have shown that curcumin deactivates bacteria by stimulating ROS generation [53,54]. Kumbar and coworkers explained the effects of curcumin on the biofilm formation and virulence factor gene expression of P. gingivalis using gene expression studies. They showed that the MBC and MIC of curcumin for both clinical strains and ATCC of P. gingivalis were 125 and 62.5 µg/mL, respectively. Curcumin inhibited attachment and biofilm formation of bacteria in a dose-dependent way. Additionally, curcumin decreased the virulence of P. gingivalis by decreasing the expression of proteinases (rgpA, rgpB, and kgp) and adhesions (fimA, hagA, and hagB) as the main genes of virulence factors. Curcumin has presented anti-biofilm and antibacterial effects against P. gingivalis. Furthermore, due to the pleiotropic actions of curcumin, it can be an inexpensive and readily available therapeutic agent in the treatment of periodontal disease [55]. Chen and coworkers investigated the anti-inflammatory effects and the mechanism of action of curcumin in macrophages stimulated by P. gingivalis lipopolysaccharide (LPS). They reported that curcumin prevented the expression of IL-1β and TNF-α genes and protein synthesis in RAW264.7 cells that were stimulated with LPS of P. gingivalis. In RAW264.7 cells, LPS of P. gingivalis stimulated NF-ĸB-dependent transcription, which was downregulated by pretreatment with curcumin [56]. ## 5. The Strengths and Limitations The results of this investigation showed that curcumin-loaded silica nanoparticles had suitable antibacterial actions against P. gingivalis. This finding could be very useful in overcoming bacterial resistance. In addition, the concentrations obtained in this study were lower compared to those obtained previous research works, advancing the hope of preparing optimal formulations based on these nanoparticles. The main limitation of this study was its use of a single isolate of P. gingivalis. A single isolate is not enough to draw conclusions regarding MIC values and accurately compare them to other studies. In addition, the possibility of human error in the sampling of bacteria, nanoparticle aggregation, and microbial contaminations with other bacterial strains can be considered other limitations. There are also other types of bacteria that act as periodontal pathogens, such as Fusobacterium nu-cleatum, Prevotella Intermedia, Aggregatibacter and Actinomicetencomitans. Curcumin-loaded silica nanoparticles should also be examined against these bacteria in future studies. This report was an in vitro study. Any possible toxicity of these nanoparticles should be tested in future studies before any animal or clinical trials. Moreover, the antimicrobial and antibiofilm mechanisms for them should be investigated to confirm their exact function. ## 6. Suggestions and Future Perspective It is suggested to investigate the effects of curcumin-loaded silica nanoparticles on P. gingivalis-related infections in vivo and then clinically. Additionally silica nanoparticles co-loaded with curcumin and other antibacterial agents can be prepared, and their antibacterial effects can be investigated in vitro, in vivo, and clinically. A limited number of clinical isolates of P. gingivalis were analyzed in this study, and they can be used in future studies to investigate the effects of curcumin-loaded silica nanoparticles on a greater number of bacteria. Nanoformulations of plant substances or phytochemicals can replace chemical antibacterial drugs in the future. This replacement can be a solution to reduce the use of antibiotics, which will reduce not only microbial resistance but also the toxicity and side effects caused by antibiotics. ## 7. Conclusions This study showed that P. gingivalis clinically isolated from the gingival crevice fluid of a patient with chronic periodontal diseases is highly sensitive to curcumin-loaded silica nanoparticles at a low concentration. In addition, the two-stage release profile of the prepared nanoparticles can provide both the burst release and the controlled sustained release of curcumin, which can be used to eradicate acute infections at first and then provide the drug content for a long time. It can be concluded that local nanocurcumin application for periodontal disease and implant-related infections can be considered as a promising method for the near future in dentistry. ## References 1. Shirmohammadi A., Babaloo A., Dizaj S.M., Lotfipour F., Sharifi S., Ghavimi M.A., Khezri K.. A View on Polymerase Chain Reaction as an Outstanding Molecular Diagnostic Technique in Periodontology. BioMed Res. Int. (2021) 2021 1-8. DOI: 10.1155/2021/9979948 2. Kumar M., Prakash S., Radha J.M., Lorenzo D., Chandran S., Dhumal A., Dey M., Senapathy N., Rais S.. Apitherapy and Periodontal Disease: Insights into. Antioxidants (2022) 11. DOI: 10.3390/antiox11050823 3. Fragkioudakis I., Riggio M.P., Apatzidou D.A.. Understanding the microbial components of periodontal diseases and periodontal treatment-induced microbiological shifts. J. Med. Microbiol. (2021) 70 jmm001247. DOI: 10.1099/jmm.0.001247 4. Development of a classification system for periodontal diseases and conditions. Ann. Periodontol. (1999) 4 1-6. DOI: 10.1902/annals.1999.4.1.1 5. Norowski P.A., Bumgardner J.D.. Biomaterial and antibiotic strategies for peri-implantitis: A review. J. Biomed. Mater. Res. Part B Appl. Biomater. Off. J. Soc. Biomater. Jpn. Soc. Biomater. Aust. Soc. Biomater. Korean Soc. Biomater. (2009) 88 530-543. DOI: 10.1002/jbm.b.31152 6. Lafzi A., Shirmohammadi A., Faramarzi M., Jabali S., Shayan A.. Clinical Comparison of Autogenous Bone Graft with and without Plasma Rich in Growth Factors in the Treatment of Grade II Furcation Involvement of Mandibular Molars. J. Dent. Res. Dent. Clin. Dent. Prospect. (2013) 7 22-29. DOI: 10.5681/joddd.2013.004 7. Yue C., Zhao B., Ren Y., Kuijer R., Van Der Mei H., Busscher H., Rochford E.. The implant infection paradox: Why do some succeed when others fail? Opinion and discussion paper. Eur. Cells Mater. (2015) 29 303-313. DOI: 10.22203/eCM.v029a23 8. Chapple C.C., Kumar R., Hunter N.. Vascular remodelling in chronic inflammatory periodontal disease. J. Oral Pathol. Med. (2000) 29 500-506. DOI: 10.1034/j.1600-0714.2000.291004.x 9. Mitwalli H., Alsahafi R., Balhaddad A.A., Weir M.D., Xu H.H.K., Melo M.A.S.. Emerging Contact-Killing Antibacterial Strategies for Developing Anti-Biofilm Dental Polymeric Restorative Materials. Bioengineering (2020) 7. DOI: 10.3390/bioengineering7030083 10. Chi M., Qi M., Wang P., Weir M.D., Melo M.A., Sun X., Dong B., Li C., Wu J.. Novel Bioactive and Therapeutic Dental Polymeric Materials to Inhibit Periodontal Pathogens and Biofilms. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20020278 11. Zhu J., Huang Y., Chen M., Hu C., Chen Y.. Functional Synergy of Antimicrobial Peptides and Chlorhexidine Acetate against Gram-Negative/Gram-Positive Bacteria and a Fungus. Infect. Drug Resist. (2019) 12 3227-3239. DOI: 10.2147/IDR.S218778 12. Saidin S., Jumat M.A., Amin N.A.A.M., Al-Hammadi A.S.S.. Organic and inorganic antibacterial approaches in combating bacterial infection for biomedical application. Mater. Sci. Eng. C (2020) 118 111382. DOI: 10.1016/j.msec.2020.111382 13. Cieplik F., Jakubovics N.S., Buchalla W., Maisch T., Hellwig E., Al-Ahmad A.. Resistance toward chlorhexidine in oral bacteria–is there cause for concern?. Front. Microbiol. (2019) 10 587. DOI: 10.3389/fmicb.2019.00587 14. Ahmed A., Khan A.K., Anwar A., Ali S.A., Shah M.R.. Biofilm inhibitory effect of chlorhexidine conjugated gold nanoparticles against Klebsiella pneumoniae. Microb. Pathog. (2016) 98 50-56. DOI: 10.1016/j.micpath.2016.06.016 15. Addy M.. Oral hygiene products: Potential for harm to oral and systemic health?. Periodontology (2008) 48 54-65. DOI: 10.1111/j.1600-0757.2008.00253.x 16. Njoroge T., Genco R.J., Sojar H.T., Hamada N., Genco A.C.. A role for fimbriae in. Infect. Immun. (1997) 65 1980-1984. DOI: 10.1128/iai.65.5.1980-1984.1997 17. Mohseni M., Samadi N., Ghanbari P., Yousefi B., Tabasinezhad M., Sharifi S., Nazemiyeh H.. Co-treatment by docetaxel and vinblastine breaks down P-glycoprotein mediated chemo-resistance. Iran. J. Basic Med. Sci. (2016) 19 300-309. DOI: 10.22038/IJBMS.2016.6649 18. El-Saadony M.T., Zabermawi N.M., Zabermawi N.M., Burollus M.A., Shafi M.E., Alagawany M., Yehia N., Askar A.M., Alsafy S.A., Noreldin A.E.. Nutritional Aspects and Health Benefits of Bioactive Plant Compounds against Infectious Diseases: A Review. Food Rev. Int. (2021) 37 1-23. DOI: 10.1080/87559129.2021.1944183 19. Hinault M.-P., Farina-Henriquez-Cuendet A., Goloubinoff P.. Molecular Chaperones and Associated Cellular Clearance Mechanisms against Toxic Protein Conformers in Parkinson’s Disease. Neurodegener. Dis. (2011) 8 397-412. DOI: 10.1159/000324514 20. Soory M.. Inflammatory Mechanisms and Redox Status in Periodontal and Cardiometabolic Diseases: Effects of Adjunctive Nutritional Antioxidants and Statins. Infect. Disord. Drug Targets (2012) 12 301-315. DOI: 10.2174/187152612801319302 21. Zambrano L.M.G., Brandao D.A., Rocha F.R.G., Marsiglio R.P., Longo I.B., Primo F.L., Tedesco A.C., Guimaraes-Stabili M.R., Junior C.R.. Local administration of curcumin-loaded nanoparticles effectively inhibits inflammation and bone resorption associated with experimental periodontal disease. Sci. Rep. (2018) 8 1-11. DOI: 10.1038/s41598-018-24866-2 22. Wahab S., Ahmad P., Hussain A., Qadir S.F.A.. Nanomateria, ls for the delivery of Herbal Bioactive Compounds. Curr. Nanosci. (2021) 18 425-441. DOI: 10.2174/1573413717666211004090341 23. Banerjee H.N., Verma M.. Application of nanotechnology in cancer. Technol. Cancer Res. Treatment (2008) 7 149-154. DOI: 10.1177/153303460800700208 24. Abhilash P.P., Nayak D.K., Sangoju B., Kumar R., Kumar V.. Effect of nano-silica in concrete; a review. Constr. Build. Mater. (2021) 278 122347. DOI: 10.1016/j.conbuildmat.2021.122347 25. Saravanakumar P., Arunkumar S., Mansingh B.B., Kumar P.M., Subbiah R., Eswarlal V.K.. Investigating the effect of thermal cycling on thermal characteristics of the nano-silica based phase changing material (PCM). Mater. Today Proc. (2021) 50 1502-1507. DOI: 10.1016/j.matpr.2021.09.095 26. Said A., Zeidan M., Bassuoni M., Tian Y.. Properties of concrete incorporating nano-silica. Constr. Build. Mater. (2012) 36 838-844. DOI: 10.1016/j.conbuildmat.2012.06.044 27. Ambrogio M.W., Thomas C.R., Zhao Y.-L., Zink J.I., Stoddart J.F.. Mechanized Silica Nanoparticles: A New Frontier in Theranostic Nanomedicine. Accounts Chem. Res. (2011) 44 903-913. DOI: 10.1021/ar200018x 28. Lam M., Migonney V., Falentin-Daudre C.. Review of silicone surface modification techniques and coatings for antibacterial/antimicrobial applications to improve breast implant surfaces. Acta Biomater. (2020) 121 68-88. DOI: 10.1016/j.actbio.2020.11.020 29. Massa M.A., Covarrubias C., Bittner M., Fuentevilla I.A., Capetillo P., Von Marttens A., Carvajal J.C.. Synthesis of new antibacterial composite coating for titanium based on highly ordered nanoporous silica and silver nanoparticles. Mater. Sci. Eng. C (2014) 45 146-153. DOI: 10.1016/j.msec.2014.08.057 30. Patra D., Karaman D., Desai D., El Khoury E., Rosenholm J.M.. Preparation of curcumin loaded mesoporous silica nanoparticles: Determining polarizability inside the mesopores. Mater. Res. Bull. (2016) 84 267-272. DOI: 10.1016/j.materresbull.2016.08.012 31. Hunt D., Jones J., Dowell V.. Selective medium for the isolation of. J. Clin. Microbiol. (1986) 23 441-445. DOI: 10.1128/jcm.23.3.441-445.1986 32. Dowding J.M., Das S., Kumar A., Dosani T., McCormack R., Gupta A., Sayle T.X.T., Sayle D.C., von Kalm L., Seal S.. Cellular Interaction and Toxicity Depend on Physicochemical Properties and Surface Modification of Redox-Active Nanomaterials. ACS Nano (2013) 7 4855-4868. DOI: 10.1021/nn305872d 33. Bakand S., Hayes A.J.. Toxicological considerations, toxicity assessment, and risk management of inhaled nanoparticles. Int. J. Mol. Sci. (2016) 17. DOI: 10.3390/ijms17060929 34. Mitchell M.J., Billingsley M.M., Haley R.M., Wechsler M.E., Peppas N.A., Langer R.. Engineering precision nanoparticles for drug delivery. Nat. Rev. Drug Discov. (2020) 20 101-124. DOI: 10.1038/s41573-020-0090-8 35. Fissan H., Ristig S., Kaminski H., Asbach C., Epple M.. Comparison of different characterization methods for nanoparticle dispersions before and after aerosolization. Anal. Methods (2014) 6 7324-7334. DOI: 10.1039/C4AY01203H 36. Marambio-Jones C., Hoek E.M.V.. A review of the antibacterial effects of silver nanomaterials and potential implications for human health and the environment. J. Nanopart. Res. (2010) 12 1531-1551. DOI: 10.1007/s11051-010-9900-y 37. Salatin S., Maleki Dizaj S., Yari Khosroushahi A.. Effect of the surface modification, size, and shape on cellular uptake of nanoparticles. Cell Biol. Int. (2015) 39 881-890. DOI: 10.1002/cbin.10459 38. Yin I.X., Zhang J., Zhao I.S., Mei M.L., Li Q., Chu C.H.. The antibacterial mechanism of silver nanoparticles and its application in dentistry. Int. J. Nanomed. (2020) 15 2555-2562. DOI: 10.2147/IJN.S246764 39. Geng Y., Dalhaimer P., Cai S., Tsai R., Tewari M., Minko T., Discher D.E.. Shape effects of filaments versus spherical particles in flow and drug delivery. Nat. Nanotechnol. (2007) 2 249-255. DOI: 10.1038/nnano.2007.70 40. Li D., Tang Z., Gao Y., Sun H., Zhou S.. A Bio-Inspired Rod-Shaped Nanoplatform for Strongly Infecting Tumor Cells and Enhancing the Delivery Efficiency of Anticancer Drugs. Adv. Funct. Mater. (2015) 26 66-79. DOI: 10.1002/adfm.201503664 41. Zhao Y., Wang Y., Ran F., Cui Y., Liu C., Zhao Q., Gao Y., Wang D., Wang S.. A comparison between sphere and rod nanoparticles regarding their. Sci. Rep. (2017) 7 4131. DOI: 10.1038/s41598-017-03834-2 42. Parani M., Lokhande G., Singh A., Gaharwar A.K.. Engineered Nanomaterials for Infection Control and Healing Acute and Chronic Wounds. ACS Appl. Mater. Interfaces (2016) 8 10049-10069. DOI: 10.1021/acsami.6b00291 43. Bhattacharyya S., Wang H., Ducheyne P.. Polymer-coated mesoporous silica nanoparticles for the controlled release of macromolecules. Acta Biomater. (2012) 8 3429-3435. DOI: 10.1016/j.actbio.2012.06.003 44. Memar M.Y., Yekani M., Ghanbari H., Nabizadeh E., Vahed S.Z., Dizaj S.M., Sharifi S.. Antimicrobial and antibiofilm activities of meropenem loaded-mesoporous silica nanoparticles against carbapenem-resistant Pseudomonas aeruginosa. J. Biomater. Appl. (2021) 36 605-612. DOI: 10.1177/08853282211003848 45. Wang L., Hu C., Shao L.. The antimicrobial activity of nanoparticles: Present situation and prospects for the future. Int. J. Nanomed. (2017) 12 1227-1249. DOI: 10.2147/IJN.S121956 46. Memar M.Y., Yekani M., Ghanbari H., Shahi S., Sharifi S., Dizaj S.M.. Biocompatibility, cytotoxicity and antibacterial effects of meropenem-loaded mesoporous silica nanoparticles against carbapenem-resistant. Artif. Cells Nanomed. Biotechnol. (2020) 48 1354-1361. DOI: 10.1080/21691401.2020.1850466 47. Shahzad M., Millhouse E., Culshaw S., Edwards C.A., Ramage G., Combet E.. Selected dietary (poly)phenols inhibit periodontal pathogen growth and biofilm formation. Food Funct. (2014) 6 719-729. DOI: 10.1039/C4FO01087F 48. Mandroli P.S., Bhat K.. An in-vitro evaluation of antibacterial activity of curcumin against common endodontic bacteria. J. Appl. Pharm. Sci. (2013) 3 16 49. Izui S., Sekine S., Maeda K., Kuboniwa M., Takada A., Amano A., Nagata H.. Antibacterial Activity of Curcumin Against Periodontopathic Bacteria. J. Periodontol. (2016) 87 83-90. DOI: 10.1902/jop.2015.150260 50. Bomdyal R.S., Shah M.U., Doshi Y.S., Shah V.A., Khirade S.P.. Antibacterial activity of curcumin (turmeric) against periopathogens-An. J. Adv. Clin. Res. Insights (2017) 4 175-180. DOI: 10.15713/ins.jcri.188 51. Maleki Dizaj S., Shokrgozar H., Yazdani J., Memar M.Y., Sharifi S., Ghavimi M.A.J.C.. Antibacterial Effects of Curcumin Nanocrystals against. Clin. Praqct. (2022) 12 809-817. DOI: 10.3390/clinpract12050085 52. Shang W., Zhao L.-J., Dong X.-L., Zhao Z.-M., Li J., Zhang B.-B., Cai H.. Curcumin inhibits osteoclastogenic potential in PBMCs from rheumatoid arthritis patients via the suppression of MAPK/RANK/c-Fos/NFATc1 signaling pathways. Mol. Med. Rep. (2016) 14 3620-3626. DOI: 10.3892/mmr.2016.5674 53. Cirano F., Pimentel S., Casati M., Corrêa M., Pino D., Messora M., Silva P., Ribeiro F.. Effect of curcumin on bone tissue in the diabetic rat: Repair of peri-implant and critical-sized defects. Int. J. Oral Maxillofac. Surg. (2018) 47 1495-1503. DOI: 10.1016/j.ijom.2018.04.018 54. Xu C.S., Ip M., Leung A.W., Wang X.N., Yang Z.R., Zhang B.T., Ip S.P.. Sonodynamic bactericidal activity of curcumin against foodborne bacteria. Hong Kong Med. J. (2018) 24 43-44. PMID: 30229738 55. Kumbar V.M., Peram M.R., Kugaji M.S., Shah T., Patil S.P., Muddapur U.M., Bhat K.G.. Effect of curcumin on growth, biofilm formation and virulence factor gene expression of. Odontology (2021) 109 18-28. DOI: 10.1007/s10266-020-00514-y 56. Chen D., Nie M., Fan M.-W., Bian Z.. Anti-Inflammatory Activity of Curcumin in Macrophages Stimulated by Lipopolysaccharides from. Pharmacology (2008) 82 264-269. DOI: 10.1159/000161127
--- title: Study of Coagulation Disorders and the Prevalence of Their Related Symptoms among COVID-19 Patients in Al-Jouf Region, Saudi Arabia during the COVID-19 Pandemic authors: - Heba Bassiony Ghanem - Abozer Y. Elderdery - Hana Nassar Alnassar - Hadeel Ali Aldandan - Wajd Hamed Alkhaldi - Kholod Saad Alfuhygy - Mjd Muharib Alruwyli - Razan Ayed Alayyaf - Shoug Khaled Alkhalef - Saud Nahar L. Alruwaili - Jeremy Mills journal: Diagnostics year: 2023 pmcid: PMC10047254 doi: 10.3390/diagnostics13061085 license: CC BY 4.0 --- # Study of Coagulation Disorders and the Prevalence of Their Related Symptoms among COVID-19 Patients in Al-Jouf Region, Saudi Arabia during the COVID-19 Pandemic ## Abstract Introduction: The coronavirus (COVID-19) has affected millions of people around the world. COVID-19 patients, particularly those with the critical illness, have coagulation abnormalities, thrombocytopenia, and a high prevalence of intravascular thrombosis. Objectives: This work aims to assess the prevalence of coagulation disorders and their related symptoms among COVID-19 patients in the Al-Jouf region of Saudi Arabia. Subjects and methods: We conducted a retrospective study on 160 COVID-19 patients. Data were collected from the medical records department of King Abdulaziz Specialist Hospital, Sakaka, Al-Jouf, Saudi Arabia. The socio-demographic data, risk factors, coagulation profile investigation results, symptom and sign data related to coagulation disorders, and disease morbidity and mortality for COVID-19 patients were extracted from medical records, and the data were stored confidentially. Results: Males represented the highest prevalence of COVID-19 infection at $65\%$; $29\%$ were aged 60 or over; $28\%$ were smokers; and $36\%$ were suffering from chronic diseases, with diabetes mellitus representing the highest prevalence. Positive D-dimer results occurred in $29\%$ of cases, with abnormal platelet counts in $26\%$. Conclusion: Our findings confirm that the dysregulation of the coagulation cascade and the subsequent occurrence of coagulation disorders are common in coronavirus infections. The results show absolute values, not increases over normal values; thus, it is hard to justify increased risk and presence based on the presented data. ## 1. Introduction In December 2019, Chinese authorities in Wuhan identified atypical pneumonia clusters of unclear cause. On January 7, 2020, they identified a new Coronaviridae virus as the cause of the respiratory infection outbreak, naming it SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) [1]. COVID-19 has been a global health disaster, with almost 497,960,492 cases of COVID-19 and 6,181,850 deaths associated with COVID-19 documented globally [2]. Due to the abundance of viruses that bind to the angiotensin-converting enzyme 2 (ACE2) receptor in the human body, COVID-19 has the ability to impact several organs and systems in the body, including the pulmonary, central nervous, cardiovascular, hematological, urogenital, and gastrointestinal systems. In addition, lung fibrosis may indirectly injure other organs by impairing oxygen delivery and triggering a cytokine storm, which leads to the malfunctioning of immune responses, impaired coagulation, and inflammatory cell infiltration [3]. The severity of COVID-19-induced pneumonia was attributed to a number of factors, such as age, chronic obstructive pulmonary disease (COPD), smoking history, and respiratory failure [4]. COVID-19 patients who experience severe symptoms, especially those who have coexisting conditions such as chronic diseases, may quickly develop acute respiratory distress syndrome (ARDS) as well as pneumonia, with a higher mortality rate days after the disease first appears, suggesting that it may lead to a multisystem condition [1]. Fever, cough, dyspnea, wheezing, and excessive mucus production were the most prevalent symptoms of COVID-19 disease [5]. Death in severe COVID-19 infection is generally caused by the development of ARDS, sepsis, and multiple organ failure, all of which are a result of dysfunctional immunological, endothelial, and coagulation responses characterized by thrombocytopenia, leukopenia, hypercoagulation, and elevated D-dimer levels [6]. Neurological clinical manifestations of COVID-19 are generally non-specific for the SARS-CoV-2 virus. However, SARS-CoV-2 has the capability to gain direct access to the nervous system, increasing the risk for even more serious neurological complications, including ischemic stroke [7]. The most common neurologic manifestations of COVID-19 are alterations in taste and smell, headache, disorders of consciousness/cognition, and neuropsychiatric manifestations; rarer manifestations include seizures, transverse myelitis, Guillain-Barre syndrome, rhabdomyolysis, and cranial nerve palsy [8]. In individuals with pre-existing neurological disorders like Parkinson’s disease or dementia, COVID-19 may raise the likelihood of exacerbating neurological complications and vice versa [9]. Severe COVID-19 seems to have a strong link with coagulopathy and platelet count abnormalities [10]. Many individuals have coagulation abnormalities that match those of other systemic coagulopathies associated with severe infections, such as disseminated intravascular coagulation (DIC), venous thromboembolism (VTE), and/or thrombotic microangiopathy [11]. Alterations caused by COVID-19 coagulopathy, such as increased blood coagulation, result in a reduction of platelets, which leads to DIC that rarely occurs [12]. The disruption in the synthesis of coagulation proteins may result in the formation of exhaustion factors, which can lead to bleeding. Even if DIC in certain individuals is uncommon, assessing sepsis-induced coagulopathy is extremely useful in predicting the severe consequences of COVID-19 [12]. The mechanisms that trigger coagulation and exacerbate coagulative disorders in association with COVID-19 infection have been connected to immune system responses, specifically the pro-inflammatory mediator release that interacts with platelets, activating tissue factor expression, and triggering stimulation of plasminogen activator inhibitor-1, which leads to the inhibition of the fibrinolytic system and leads to endothelial dysfunction, causing thrombogenesis. Elevated D-dimer levels have been identified as a reliable biomarker for poor prognosis [13]. The aim of this study is to assess the prevalence of coagulation disorders and their related symptoms among COVID-19 patients in the Al-Jouf region of Saudi Arabia. ## 2. Materials and Methods The study design in this research was a retrospective study on COVID-19 patients, conducted in the Al-Jouf region of Saudi Arabia. Study duration and sample size were conducted ($\frac{12}{2021}$–$\frac{02}{2022}$), with samples comprised of the medical records of 160 COVID-19 patients previously admitted ($\frac{01}{2020}$–$\frac{10}{2021}$) to isolation and intensive care units in King Abdulaziz Specialist Hospital, Sakaka, Al-Jouf, Saudi Arabia. Exclusion criteria were pregnancy, recurrent miscarriage, history of stroke or deep venous thrombosis (DVT), long history of birth control pill use and/or hormone replacement therapy, family history of coagulation disorders, and recent surgery with prolonged bed rest. For data collection and ethical approval, data were collected from the medical records department in King Abdulaziz Specialist Hospital, Sakaka, Al-Jouf, Saudi Arabia, with confidentiality, and stored securely in accordance with the study protocol approved by the Research Ethics Committee, Qurayyat Health Affairs, Reg. No.: H-13-S-071. Sociodemographic data and risk factor information, such as smoking and history of chronic diseases, were collected for COVID-19 patients, along with results for thrombosis risk, including platelet count, PT (prothrombin time), aPTT (activated partial thromboplastin time), INR (international normalized ratio), and D-dimer level. A control group was added to the analysis for coagulation values, platelet counts, and D-dimer using data gathered from 160 normal cases without COVID-19 infection. Symptoms and signs for the 160 COVID-19 patients, related to coagulation disorders, disease morbidity, and mortality data, were also stored. The relative risk and odds ratio at $95\%$ CI (confidence interval) for the thrombosis risk factor in COVID-19 patients were estimated. Data analyses were carried out using SPSS 23 statistical software, and results were analyzed using the Chi-square test. Relative risk, odds ratio, and Fisher’s exact test were used, and any differences were considered statistically significant at $p \leq 0.05.$ ## 3. Results A total of 160 COVID-19 patients were involved in this study. Table 1 displays sociodemographic data features in COVID-19 patients analyzed by using the Chi-Square test. A total of 104 ($65\%$) of cases were males infected with COVID-19, while 56 ($35\%$) were females (p-value < 0.001 ). Patients ranging from 19–59 years old were 113 ($71\%$), while patients 60 years old or more were 47 ($29\%$) (p-value < 0.001 ). A fraction ($79\%$) of patients were hospitalized in isolation units and $21\%$ in ICUs (p-value < 0.001 ). Table 1 also illustrates risk factors in COVID-19 patients’ prevalence. A total of 42 ($28\%$) cases were smokers, and 118 ($72\%$) were non-smokers (p-value < 0.001 ). A total of 57 ($36\%$) cases were suffering from chronic diseases and 103 ($64\%$) were not (p-value < 0.001 ). Figure 1 shows the prevalence of chronic diseases in COVID-19 patients ($36\%$ of the total number of COVID-19 patients), with the highest percentage being that of diabetes mellitus ($39\%$), followed by hypertension ($30\%$), COPD ($14\%$), chronic liver diseases ($10\%$), and chronic renal diseases ($7\%$) (p-value <0.001 ). Table 2 displays the results of investigations into thrombotic risk and coagulation disorders. Platelet count results were normal in $74\%$ of cases, low in $18\%$, and high in $8\%$ (p-value <0.001 ). For coagulation profile results, PT and INR results were normal in $55\%$ of cases, high in $34\%$, and low in $11\%$ (p-value <0.001 ). Moreover, aPTT results were normal in $64\%$ of cases, high in $28\%$, and low in $8\%$ (p-value <0.001 ). D-dimer results were negative in $71\%$ of cases and positive in $29\%$ of cases (p-value <0.001 ). Table 3 demonstrates the prevalence of symptoms and signs in COVID-19 patients’ infections with cough presented the highest prevalence ($64\%$), followed by fever ($56\%$), muscle ache ($41\%$), chest pain ($30\%$), dyspnea ($29\%$), sore throat ($19\%$), epigastric pain ($7\%$), and diarrhea ($5\%$). Table 3 displays the comparison of platelet count, coagulation profile, and D-dimer results in the control group and COVID-19 patients using an independent T test. Platelet count results were lower in COVID-19 patients than in the control group (p-value 0.001 ). For coagulation profile results, PT, aPTT, and INR results were higher in COVID-19 patients than in the control group (p-values <0.001 , <0.001 , and 0.006, respectively). D-dimer results were also higher in COVID-19 patients than in the control group (p-value 0.005 ). Table 4 demonstrates the prevalence of symptoms and signs in COVID-19 patients’ infection, with cough presenting the highest prevalence ($64\%$), followed by fever ($56\%$), muscle ache ($41\%$), chest pain ($30\%$), dyspnea ($29\%$), sore throat ($19\%$), epigastric pain ($7\%$), and diarrhea ($5\%$). Table 5 displays the prevalence of symptoms and signs related to precipitating thrombotic risk in COVID-19 patients, with pulmonary thrombosis ($29\%$) having the highest percentage of symptoms and signs, followed by DVT ($22\%$). Figure 2 shows the prevalence of disease morbidity in COVID-19 patients and the different complications that occurred in the COVID-19 patients in the studied group. Pneumonia affected $30\%$ of all cases, $12\%$ had thrombosis, $5\%$ were unconscious, $4\%$ developed bleeding disorders, $4\%$ had septic shock, $3\%$ had electrolyte disturbances, and $2.5\%$ had renal failure. COVID-19 patients who needed O2 therapy were $76\%$, and $7\%$ needed mechanical ventilators, and the mortality rate was $2\%$ (p-value <0.001 ). Figure 3 shows types of thrombosis in COVID-19 patients with a prevalence complicated by thrombosis, with the highest percentage for DVT ($53\%$), followed by pulmonary ($26\%$), cerebral ($16\%$), and cardiac thrombosis ($5\%$), with a significant p-value of 0.03 . Table 6 shows the relative risk (RR) and odds ratio (OR) for thrombosis risk factors. Patients with chronic diseases were at the greatest risk, with an OR of 18.04 (3.65–82.4) and a RR of 13.55 (3.2–57.17). Thrombosis was associated with a high death rate, with an OR of 15.4 (1.3–179) and a RR of 5.8 (2.3–14.4). Other significant risk factors are age 60 years or more, with an OR of 12.77 (3.96–41.2) and a RR of 9.02 (3.2–25.7). Positive D-dimer was an important risk factor that had an OR of 9.16 (3.07–27.3) and a RR of 6.7 (2.57–17.6), followed by unconsciousness with an OR of 8.5 (1.9–37.3) and a RR of 4.75 (2.06–10.9), high PT and high INR with an OR of 6.8 (2.3–20.2) and a RR of 5.3 (2.03–14.07), low platelet count with an OR of 5.78 (2.08–16.06) and a RR of 4.2 (1.9–9.5), and high aPTT with an OR of 5.6 (2.04–15.4) and a RR of 4.38 (1.8–10.4). Smoking is also a risk factor with an OR of 4.88 (1.8–13.18) and RR of 3.86 (1.67–8.9), in addition to respiratory failure and mechanical ventilator with an OR of 4.75 (1.25–18.02) and a RR of 3.39 (1.37–8.4). ## 4. Discussion COVID-19 surfaced in Wuhan, China, and rapidly spread across the globe, causing a pandemic that crippled life as we know it and put tremendous pressure on governments and individuals alike. The SARS-CoV-2 strain is incredibly contagious, and the hospitalization rate increases significantly in patients aged ≥50 and in those with underlying conditions such as hypertension or obesity [14]. Furthermore, hospitalized patients suffering from severe respiratory or systemic ailments are more susceptible to developing venous thromboembolism and thrombotic complications, which can be life-threatening [15]. Our study focused on the prevalence of coagulation abnormalities and symptoms associated with COVID-19 in patients in the Al-Jouf region. The first COVID-19 case in Saudi Arabia was reported on March 2, 2020 [16]. There were 507,000 confirmed COVID-19 cases and 8048 fatalities as of July 19, 2021 [17]. In May 2021, Saudi Arabia documented the first case of the delta variant, and delta became the predominant variant of concern from May through June 2021. Sequenced samples revealed that delta was the most prevalent variant (at $40.9\%$), which, compared to the original strain, is more contagious and causes more serious symptoms [18]. Additionally, the second most abundant strains are beta ($15.9\%$) and alpha ($11.6\%$); this is consistent with transmission rates reported from other nations, such as the UK and France, the latter of which in June 2021 reported a rapid spread of delta [19]. Since our data were collected from January 2020 to October 2021, we speculate the aforementioned strains were circulating from May 2021 until the end of the study, in addition to the original strain that had been circulating in Saudi Arabia since the beginning of 2020. This research found that the rate of COVID-19 infection in the Al-Jouf region was significantly higher in males ($65\%$) than females ($35\%$). Similarly, a study conducted in Saudi Arabia revealed that men constituted $71\%$ of all cases [20]. On the other hand, Chinese research demonstrated that male and female patients exhibited similar susceptibility to the virus [21]. In addition, the gap in infection rates between sexes in Saudi Arabia does not comport with several studies from China, the USA, and Europe that showed a comparable number of cases in both men and women [22,23,24]. Moreover, during COVID-19 outbreaks, women in Saudi Arabia had better knowledge of infection control practices and were more compliant with the WHO prevention protocols [25]. This may explain the difference in gender infection rates in Saudi Arabia when compared with the rest of the world. Another possible explanation is that most women in Saudi Arabia tend to wear a cloth niqab, which works roughly as a mask and may prevent face touching and limit the spread of droplets. Coughing is the most common symptom in our study, followed by fever, muscle ache, chest pain, dyspnea, sore throat, and epigastric pain. The highest prevalence of symptoms associated with (and possibly precipitating) thrombotic risk was found in pulmonary thrombosis ($29\%$), followed by DVT, cerebral, and urinary tract thrombosis. Monitoring symptoms may assist clinicians in identifying individuals with bad prognoses, especially in high-risk patients. By promptly identifying risk factors, we can focus on high-risk groups to hopefully reduce the severity and mortality of the disease through timely diagnosis, isolation, and treatment. With this in mind, our study demonstrated that the most prevalent risk factor is diabetes mellitus ($39\%$), followed by hypertension ($30\%$), COPD ($14\%$), chronic liver diseases, and chronic renal disorders. Additionally, $28\%$ of all cases involved smokers, and $36\%$ had chronic diseases. Other risk factors include unconsciousness, septic shock, susceptibility to bleeding, electrolyte imbalance, and renal failure. Finally, $76\%$ of patients in this study required O2 therapy. These risk factors are associated with the progression of pneumonia in COVID-19 patients and may increase the severity of the disease [4]. We found that the most common morbidity associated with COVID-19 was pneumonia ($30\%$), followed by thrombosis ($12\%$). Similar findings have linked several morbidities to COVID-19, such as pneumonia with ARDS, thrombosis, renal failure, anorexia, and digestive issues [26]. Furthermore, we found that DVT was the most common type of thrombosis in COVID-19 patients, accounting for $53\%$ of all cases, followed by pulmonary ($26\%$), cerebral ($16\%$), and cardiac thrombosis ($5\%$). A study by Hanff and co-workers indicated that the most common thrombotic events in COVID-19 are pulmonary and DVT, followed by cerebral and cardiac thrombosis, which result in ischemic stroke [27]. These findings support our results that COVID-19 is highly prothrombotic. Finally, since COVID-19 infection is commonly associated with hypercoagulability, the chance of developing DVT or a potentially fatal pulmonary embolism is significantly higher as a result of small thrombus migration [28,29]. In this study, about one out of five patients ($21\%$) was admitted to the ICU. Of these, $7\%$ required artificial ventilation for respiratory system failure. Additional complications found in the ICU included shock and organ failure. Although ICU admissions are at $21\%$ in Al-Jouf, the mortality rate is only $2\%$. This finding is in line with the global mortality rate of around $2\%$ [30]. Numerous variables, such as age, smoking, and illness, have a significant impact on the death rate [31]. The fatality rate raises significantly in adults above the age of 50; patients above 80 years of age have a mortality rate of $14.8\%$; and those between ages of 70 and 79 have a mortality rate of $8\%$. Finally, the mortality rate at 60–69 years of age is $3.6\%$ [32]. In our study, the risk of thrombosis substantially increased in the presence of chronic diseases, old age, positive D-dimer, thrombocytopenia, and smoking. Our coagulation panel revealed that COVID-19 patients exhibited elevated D-dimer levels in $29\%$ of cases, elevated PT and INR in $34\%$ of cases, and $28\%$ had high aPTT. Both low and high platelet counts were noted. These findings are in line with the findings of other studies, which reported that coagulopathy was a common symptom of SARS-CoV-2 infection, with an increase in D-dimer, fibrinogen, PT, aPTT, and moderate thrombocytopenia [28,33]. As we have previously demonstrated, the levels of PT were significantly higher in COVID-19 patients, and this can be used as a key coagulation marker that may help us better predict the outcome of the disease, as higher PT results have been associated with an increased ICU admission rate as well as increased mortality [5]. Additionally, Zhang et al. discovered that elevated D-dimer levels above 2.0 mg/L might predict mortality with a sensitivity of $92.3\%$ and a specificity of $83.3\%$ [34]. Similarly, Poudel et al. found that D-dimer levels exceeding 1.5 μg/mL predict mortality in COVID-19 patients with high specificity and sensitivity [35]. Tang et al. demonstrated that elevated D-dimer, thrombocytopenia, and delayed prothrombin time have been linked to poor prognosis in patients with COVID-19 [28]. Yet, if a substantial thrombus develops in the body but is not cleared (a far more severe situation for the body), the spike in D-dimer could be moderate. In particular, even in extreme situations such as death, the rise in D-dimer is relatively moderate in suppressed-fibrinolytic-type DIC induced by sepsis [36]. Thus, to properly identify and forecast the outcome of COVID-19-associated coagulopathies, we should not depend exclusively on D-dimer as a coagulation indicator; other coagulation makers should be included. Our analysis of COVID-19 patient data showed significantly higher ($8\%$) and, in some cases, lower ($18\%$) levels of platelet count. This finding may suggest that platelet abnormalities may be predictive of poor prognosis since low platelet count is associated with a fivefold increase in the risk of disease severity, which could be attributed to secondary infections [37,38]. Furthermore, only $8\%$ of ICU patients had platelet counts of less than 100 × 109/L when they were admitted, according to Wu et al. [ 5]. Other studies have observed the reduction and elevation of platelet count in COVID-19 patients; this phenomenon may point to an increased inflammatory state, with a low platelet count suggesting excessive platelet consumption via thrombi formation but an increase in platelet count suggesting a cytokine storm [39]. Different mechanisms can dictate viral-platelet interactions depending on the virus type; these interactions may change platelet number and function. For example, suppressing platelet production or enhancing platelet destruction, anti-viral antibodies that cross-react with platelet surface integrins, inducing systemic inflammation, and clearing of activated platelets via splenic/liver macrophages and/or phagocytosis by neutrophils may increase platelet counts in COVID-19 patients, and activated platelets may, in turn, contribute to lung injury [40]. Long-term COVID is a chronic, frequently disabling illness that affects many organ systems and is present in at least $10\%$ of severe COVID-19 infections. It has more than 200 known manifestations. Furthermore, numerous pathogenic theories have been put forth, such as SARS-CoV-2 reservoirs remaining in tissues, immunological dysregulation, effects of SARS-CoV-2 on the microbiota (i.e., the virome, autoimmune responses, and immune system priming caused by molecular mimicry), microvascular blood clotting with endothelial dysfunction, and dysfunctional signaling in the brainstem and/or vagus nerve. Its symptoms include postural orthostatic tachycardia syndrome, dysautonomia, and myalgic encephalomyelitis/chronic fatigue syndrome [41]. The contradictory presence of both lymphopaenia and an autoimmune state in long-term COVID is puzzling; autoantibodies are directed against ACE2, ACE1, beta-adrenergic receptors, muscarinic cholinergic receptors, and a variety of self-antigens. The presence of CD4+ T-cell lymphopenia was found to be a reliable predictor of severity and hospitalization in COVID-19 individuals. This phenomenon suggests that the effects of SARS-CoV-2 may be selectively targeting CD8-expressing T lymphocytes while leaving the B lymphocytic system unaffected [42]. Coagulopathies resulting from a COVID-19 infection are possibly caused by numerous underlying mechanisms. Firstly, viral infections induce a systemic inflammatory response that alters the homeostatic balance between procoagulant and anticoagulant activities through various mechanisms, for example, endothelial dysfunction, increased von Willebrand factor, TLR (Toll-like receptor) activation, and tissue-factor pathway activation. These pathways can lead to excessive activation of the coagulation cascade, increasing D-dimer levels [43]. Additionally, during infection, levels of pro-inflammatory cytokines such as IL-1 (interleukin) and IL-6 increase as a result of the virus binding to TLR, which eventually leads to inflammation, fever, and lung fibrosis [44]. Furthermore, because of the strong expression of the virus-binding ACE2 receptor, it is widely established that the gastrointestinal system is actively involved in COVID-19 pathogenesis. The intestinal invasion of SARS-CoV-2 could disturb intestinal homeostasis and host immunological homeostasis, which were responsible for COVID-19′s adverse consequences [45,46]. The dispersion of the gut microbiota and bacteria, which is associated with socioeconomic status and could correlate with viral invasion and the severity of COVID-19 illness. During a COVID-19 infection, antibiotics are administered to treat secondary infections; this could potentially worsen the prognosis due to the decline in human microbiota that promotes the host’s immune system as a result of excessive antibiotic consumption [47]. The capacity of microbial pathogenic components to translocate via the leaky gut into the circulatory system allows inflammatory cytokines to be secreted by triggering pattern recognition receptor-like TLRs and NOD-like receptors, resulting in systemic inflammation. Moreover, the amounts of various microbial species varied in tandem with the course of COVID-19 and were linked to biomarkers of host immunology and inflammation [48]. An additional possibility for the increased inflammation in COVID-19 patients is the glycolysis pathway enrichment, which has been linked to greater SRAS-Cov-2 activity. Due to the TLR involvement and subsequent activation of the PI3K/Akt pathway, the primary energy metabolism shifts from lipid to glycolysis to create ATP amidst viral and bacterial infections, particularly following macrophage polarization and dendritic cell activation [49]. Cytokines such as IL-6, TNF, and IL-1 have a vital role in the down-regulation of crucial physiological coagulant pathways, as well as the hyper-activation of platelets that can result in an aberrant clot formation [50]. TLR activation mediated by a COVID infection is responsible for the initiation of extrinsic coagulation pathways and inflammatory signaling in endothelial cells, leukocyte infiltration, neutrophil granule leakage, platelet aggregation, and fibrin deposition that cause thrombosis [51]. Thrombosis caused by a cytokine storm, antiphospholipid antibody syndrome, macrophage activation syndrome, complement cascade, and RAS dysregulation are only a few of the thrombogenic processes that may be involved in the COVID-19 thrombosis. Although cytokine storm and IL-6, in particular, are substantially increased in COVID-19 patients compared to other septic etiologies, and these are mechanistically upstream of numerous thrombogenic pathways, it is currently unclear which set of pathways are prominent in COVID-19. Further studies are needed to better understand how these pathways could potentially help us develop novel, pathway-guided treatment strategies for the disease [52]. ## 5. Conclusions Our findings confirm that coagulation cascade dysregulation and subsequent disorders are common in COVID-19 infections and that thrombosis risk increases with chronic disease, old age, a positive D-dimer, unconsciousness, an increased coagulation profile, thrombocytopenia, and smoking. In addition, coagulation disorders cause intravascular clots in deep veins and in the pulmonary, cerebral, and cardiac circulations. Considering the high prevalence of coagulation disorders in COVID-19 patients in Al-Jouf, a national strategy for prevention and management is urgently needed and could provide public education programs for awareness focused on prevention of transmission modes, spread risk factors, signs, and symptoms. ## References 1. Zhu N., Zhang D., Wang W., Li X., Yang B., Song J., Zhao X., Huang B., Shi W., Lu R.. A novel coronavirus from patients with pneumonia in China, 2019. N. Engl. J. Med. (2020) 382 727-733. DOI: 10.1056/NEJMoa2001017 2. Suthar A.B., Wang J., Seffren V., Wiegand R.E., Griffing S., Zell E.. Public health impact of covid-19 vaccines in the US: Observational study. BMJ (2022) 377 e069317. DOI: 10.1136/bmj-2021-069317 3. Rabaan A.A., Bakhrebah M.A., Mutair A.A., Alhumaid S., Al-Jishi J.M., AlSihati J., Albayat H., Alsheheri A., Aljeldah M., Garout M.. Systematic review on pathophysiological complications in severe COVID-19 among the non-vaccinated and vaccinated population. Vaccines (2022) 10. DOI: 10.3390/vaccines10070985 4. Liu W., Tao Z.W., Wang L., Yuan M.L., Liu K., Zhou L., Wei S., Deng Y., Liu J., Liu H.G.. Analysis of factors associated with disease outcomes in hospitalized patients with 2019 novel coronavirus disease. Chin. Med. J. (2020) 133 1032-1038. DOI: 10.1097/CM9.0000000000000775 5. Huang C., Wang Y., Li X., Ren L., Zhao J., Hu Y., Zhang L., Fan G., Xu J., Gu X.. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet (2020) 395 497-506. DOI: 10.1016/S0140-6736(20)30183-5 6. Alsrhani A., Alshomar A., Elderdery A.Y., Rasheed Z., Farhana A.. Diagnosis and Stratification of COVID-19 Infections Using Differential Plasma Levels of D-Dimer: A Two-Center Study from Saudi Arabia. Microbiol. Res. (2023) 14 67-76. DOI: 10.3390/microbiolres14010006 7. Chen X., Laurent S., Onur O.A., Kleineberg N.N., Fink G.R., Schweitzer F., Warnke C.. A systematic review of neurological symptoms and complications of COVID-19. J. Neurol. (2021) 268 392-402. DOI: 10.1007/s00415-020-10067-3 8. Hensley M.K., Markantone D., Prescott H.C.. Neurologic manifestations and complications of COVID-19. Annu. Rev. Med. (2022) 73 113-127. DOI: 10.1146/annurev-med-042320-010427 9. Kubota T., Kuroda N.. Exacerbation of neurological symptoms and COVID-19 severity in patients with preexisting neurological disorders and COVID-19: A systematic review. Clin. Neurol. Neurosurg. (2021) 200 106349. DOI: 10.1016/j.clineuro.2020.106349 10. Madani M., Goldstein D., Stefanescu R., Woodman S.E., Rojas-Hernandez C.M.. Characterization of coagulopathy and outcomes in cancer patients with severe COVID-19 illness: Longitudinal changes in hospitalized cancer patients. Cancer Med. (2022) 11 3771-3785. DOI: 10.1002/cam4.4753 11. Connors J.M., Levy J.H.. Thromboinflammation and the hypercoagulability of COVID-19. J. Thromb. Haemost. (2020) 18 1559-1561. DOI: 10.1111/jth.14849 12. Teimury A., Khameneh M.T., Khaledi E.M.. Major coagulation disorders and parameters in COVID-19 patients. Eur. J. Med. Res. (2022) 27 25. DOI: 10.1186/s40001-022-00655-6 13. Terpos E., Ntanasis-Stathopoulos I., Elalamy I., Kastritis E., Sergentanis T.N., Politou M., Psaltopoulou T., Gerotziafas G., Dimopoulos M.A.. Hematological findings and complications of COVID-19. Am. J. Hematol. (2020) 95 834-847. DOI: 10.1002/ajh.25829 14. Miyazawa D.. Why obesity, hypertension, diabetes, and ethnicities are common risk factors for COVID-19 and H1N1 influenza infections. J. Med. Virol. (2021) 93 127-128. DOI: 10.1002/jmv.26220 15. Giannis D., Ziogas I.A., Gianni P.J.J.. Coagulation disorders in coronavirus infected patients: COVID-19, SARS-CoV-1, MERS-CoV and lessons from the past. J. Clin. Virol. (2020) 127 104362. DOI: 10.1016/j.jcv.2020.104362 16. Obied D.A., Alhamlan F.S., Al-Qahtani A.A., Al-Ahdal M.N.. Containment of COVID-19: The unprecedented response of Saudi Arabia. J. Infect. Dev. Ctries (2020) 14 699-706. DOI: 10.3855/jidc.13203 17. Alsayer R.M., Alsharif H.M., Al Baadani A.M., Kalam K.A.. Clinical and epidemiological characteristics of COVID-19 mortality in Saudi Arabia. Saudi Med. J. (2021) 42 1083-1094. DOI: 10.15537/smj.2021.42.10.20210396 18. Gunadi M.S., Hakim H., Wibawa V., Marcellus I., Setiawaty E., Slamet R., Trisnawati K.. Is the Infection of the SARS-CoV-2 Delta Variant Associated With the Outcomes of COVID-19 Patients?. Front. Med. (2021) 8 780611. DOI: 10.3389/fmed.2021.780611 19. Alhasan K., Aljamaan F., Temsah M.H., Alshahrani F., Bassrawi R., Alhaboob A., Assiri R., Alenezi S., Alaraj A., Alhomoudi R.I.. COVID-19 Delta Variant: Perceptions, Worries, and Vaccine-Booster Acceptability among Healthcare Workers. Healthcare (2021) 9. DOI: 10.3390/healthcare9111566 20. Alahmari A.A., Khan A.A., Elganainy A., Almohammadi E.L., Hakawi A.M., Assiri A.M., Jokhdar H.A.. Epidemiological and clinical features of COVID-19 patients in Saudi Arabia. J. Infect. Public Health (2021) 14 437-443. DOI: 10.1016/j.jiph.2021.01.003 21. Mukherjee S., Pahan K.. Is COVID-19 Gender-sensitive?. J. Neuroimmune Pharm. (2021) 16 38-47. DOI: 10.1007/s11481-020-09974-z 22. Stokes E.K., Zambrano L.D., Anderson K.N., Marder E.P., Raz K.M., El Burai Felix S., Tie Y., Fullerton K.E.. Coronavirus Disease 2019 Case Surveillance—United States, January 22-May 30, 2020. MMWR Morb. Mortal. Wkly. Rep. (2020) 69 759-765. DOI: 10.15585/mmwr.mm6924e2 23. Garg S., Kim L., Whitaker M., O’Halloran A., Cummings C., Holstein R., Prill M., Chai S.J., Kirley P.D., Alden N.B.. Hospitalization Rates and Characteristics of Patients Hospitalized with Laboratory-Confirmed Coronavirus Disease 2019—COVID-NET, 14 States, March 1-30, 2020. MMWR Morb. Mortal. Wkly. Rep. (2020) 69 458-464. DOI: 10.15585/mmwr.mm6915e3 24. Gebhard C., Regitz-Zagrosek V., Neuhauser H.K., Morgan R., Klein S.L.. Impact of sex and gender on COVID-19 outcomes in Europe. Biol. Sex Differ. (2020) 11 29. DOI: 10.1186/s13293-020-00304-9 25. Alshammary F., Siddiqui A.A., Amin J., Ilyas M., Rathore H.A., Hassan I., Alam M.K., Kamal M.A.. Prevention Knowledge and Its Practice Towards COVID-19 Among General Population of Saudi Arabia: A Gender-based Perspective. Curr. Pharm. Des. (2021) 27 1642-1648. DOI: 10.2174/1381612826666200818213558 26. Wang D., Hu B., Hu C., Zhu F., Liu X., Zhang J., Wang B., Xiang H., Cheng Z., Xiong Y.J.J.. Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus–infected pneumonia in Wuhan, China. JAMA (2020) 323 1061-1069. DOI: 10.1001/jama.2020.1585 27. Hanff T.C., Mohareb A.M., Giri J., Cohen J.B., Chirinos J.A.. Thrombosis in COVID-19. Am. J. Hematol. (2020) 95 1578-1589. DOI: 10.1002/ajh.25982 28. Tang N., Li D., Wang X., Sun Z.. Abnormal coagulation parameters are associated with poor prognosis in patients with novel coronavirus pneumonia. J. Thromb. Haemost. (2020) 18 844-847. DOI: 10.1111/jth.14768 29. Chen N., Zhou M., Dong X., Qu J., Gong F., Han Y., Qiu Y., Wang J., Liu Y., Wei Y.. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: A descriptive study. Lancet (2020) 395 507-513. DOI: 10.1016/S0140-6736(20)30211-7 30. Abate S.M., Checkol Y.A., Mantedafro B., Basu B., Ethiopia D.. Prevalence and risk factors of mortality among hospitalized patients with COVID-19: A systematic review and meta-analysis. Bull. World Health Organ. (2020) 10 6 31. Alwafi H., Naser A.Y., Qanash S., Brinji A.S., Ghazawi M.A., Alotaibi B., Alghamdi A., Alrhmani A., Fatehaldin R., Alelyani A.. Predictors of Length of Hospital Stay, Mortality, and Outcomes Among Hospitalised COVID-19 Patients in Saudi Arabia: A Cross-Sectional Study. J. Multidiscip. Heal. (2021) 14 839-852. DOI: 10.2147/JMDH.S304788 32. Hafeez A., Ahmad S., Siddqui S.A., Ahmad M., Mishra S.J.E.. A review of COVID-19 (Coronavirus Disease-2019) diagnosis, treatments and prevention. EJMO (2020) 4 116-125 33. Guan W.J., Ni Z.Y., Hu Y., Liang W.H., Ou C.Q., He J.X., Liu L., Shan H., Lei C.L., Hui D.S.C.. Clinical Characteristics of Coronavirus Disease 2019 in China. N. Engl. J. Med. (2020) 382 1708-1720. DOI: 10.1056/NEJMoa2002032 34. Zhang Q., Xu Q., Chen Y.Y., Lou L.X., Che L.H., Li X.H., Sun L.Y., Bao W.G., Du N.. Clinical characteristics of 41 patients with pneumonia due to 2019 novel coronavirus disease (COVID-19) in Jilin, China. BMC Infect. Dis. (2020) 20. DOI: 10.1186/s12879-020-05677-1 35. Poudel A., Poudel Y., Adhikari A., Aryal B.B., Dangol D., Bajracharya T., Maharjan A., Gautam R.. D-dimer as a biomarker for assessment of COVID-19 prognosis: D-dimer levels on admission and its role in predicting disease outcome in hospitalized patients with COVID-19. PLoS ONE (2021) 16. DOI: 10.1371/journal.pone.0256744 36. Asakura H., Ogawa H.. COVID-19-associated coagulopathy and disseminated intravascular coagulation. Int. J. Hematol. (2021) 113 45-57. DOI: 10.1007/s12185-020-03029-y 37. Yang X., Yu Y., Xu J., Shu H., Liu H., Wu Y., Zhang L., Yu Z., Fang M., Yu T.. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: A single-centered, retrospective, observational study. Lancet Respir. Med. (2020) 8 475-481. DOI: 10.1016/S2213-2600(20)30079-5 38. Elkhalifa A.M.E., Elderdery A.Y., Al Bataj I.A., Tamomh A.G., Alyami M.M., Almakrami H.A., Alofair M.A., Almorish M.A., Bashanfer S., Tabash M.I.. Hematological Findings among COVID-19 Patients Attending King Khalid Hospital at Najran, Kingdom of Saudi Arabia. BioMed Res. Int. (2022) 2022 4620037. DOI: 10.1155/2022/4620037 39. Amgalan A., Othman M.. Hemostatic laboratory derangements in COVID-19 with a focus on platelet count. Platelets (2020) 31 740-745. DOI: 10.1080/09537104.2020.1768523 40. Amgalan A., Othman M.. Exploring possible mechanisms for COVID-19 induced thrombocytopenia: Unanswered questions. J. Thromb. Haemost. (2020) 18 1514-1516. DOI: 10.1111/jth.14832 41. Davis H.E., McCorkell L., Vogel J.M., Topol E.J.. Long COVID: Major findings, mechanisms and recommendations. Nat. Rev. Microbiol. (2023) 21 133-146. DOI: 10.1038/s41579-022-00846-2 42. Gerlach J., Baig A.M., Fabrowski M., Viduto V.. The immune paradox of SARS-CoV-2: Lymphocytopenia and autoimmunity evoking features in COVID-19 and possible treatment modalities. Rev. Med. Virol. (2023) e2423. DOI: 10.1002/rmv.2423 43. Subramaniam S., Scharrer I.J.F.B.. Procoagulant activity during viral infections. Front. Biosci.-Landmark (2018) 23 1060-1081 44. Conti P., Ronconi G., Caraffa A.L., Gallenga C.E., Ross R., Frydas I., Kritas S.K.. Induction of pro-inflammatory cytokines (IL-1 and IL-6) and lung inflammation by Coronavirus-19 (COVI-19 or SARS-CoV-2): Anti-inflammatory strategies. J. Biol. Regul. Homeost. Agents (2020) 34 1 45. Hamming I., Timens W., Bulthuis M., Lely A., Navis G., van Goor H.. Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis. J. Pathol. A J. Pathol. Soc. Great Br. Irel. (2004) 203 631-637. DOI: 10.1002/path.1570 46. Liu Q., Mak J.W.Y., Su Q., Yeoh Y.K., Lui G.C.-Y., Ng S.S.S., Zhang F., Li A.Y., Lu W., Hui D.S.-C.. Gut microbiota dynamics in a prospective cohort of patients with post-acute COVID-19 syndrome. Gut (2022) 71 544-552. DOI: 10.1136/gutjnl-2021-325989 47. Brogna C., Cristoni S., Brogna B., Bisaccia D.R., Marino G., Viduto V., Montano L., Piscopo M.. Toxin-like Peptides from the Bacterial Cultures Derived from Gut Microbiome Infected by SARS-CoV-2—New Data for a Possible Role in the Long COVID Pattern. Biomedicines (2023) 11. DOI: 10.3390/biomedicines11010087 48. Sun Z., Song Z.-G., Liu C., Tan S., Lin S., Zhu J., Dai F.-H., Gao J., She J.-L., Mei Z.. Gut microbiome alterations and gut barrier dysfunction are associated with host immune homeostasis in COVID-19 patients. BMC Med. (2022) 20. DOI: 10.1186/s12916-021-02212-0 49. Krawczyk C.M., Holowka T., Sun J., Blagih J., Amiel E., DeBerardinis R.J., Cross J.R., Jung E., Thompson C.B., Jones R.G.. Toll-like receptor–induced changes in glycolytic metabolism regulate dendritic cell activation. Blood J. Am. Soc. Hematol. (2010) 115 4742-4749. DOI: 10.1182/blood-2009-10-249540 50. Miesbach W., Makris M.. COVID-19: Coagulopathy, Risk of Thrombosis, and the Rationale for Anticoagulation. Clin. Appl. Thromb. Hemost. (2020) 26 1076029620938149. DOI: 10.1177/1076029620938149 51. Biswas I., Khan G.A.. Coagulation disorders in COVID-19: Role of toll-like receptors. J. Inflamm. Res. (2020) 13 823. DOI: 10.2147/JIR.S271768 52. Cai C., Guo Y., You Y., Hu K., Cai F., Xie M., Yang L., Ling K., Ye D., Misra S.. Deep Venous Thrombosis in COVID-19 Patients: A Cohort Analysis. Clin. Appl. Thromb. Hemost. (2020) 26 1076029620982669. DOI: 10.1177/1076029620982669
--- title: Improvement of Self-Esteem in Children with Specific Learning Disorders after Donkey-Assisted Therapy authors: - Francesco Corallo - Lilla Bonanno - Davide Cardile - Francesca Luvarà - Silvia Giliberto - Marcella Di Cara - Simona Leonardi - Angelo Quartarone - Giuseppe Rao - Alessandra Pidalà journal: Children year: 2023 pmcid: PMC10047257 doi: 10.3390/children10030425 license: CC BY 4.0 --- # Improvement of Self-Esteem in Children with Specific Learning Disorders after Donkey-Assisted Therapy ## Abstract Dyslexia is a learning disorder related to receptive language characterized by difficulties with decoding, fluent word recognition, automatic naming skills and/or reading comprehension skills. It usually leads to severe functional impairment and the permanent need for support and interventions. Since animal-assisted interventions (AAIs) have been found to improve physical, emotional, cognitive and/or social functioning in humans, the aim of this study is to demonstrate the effectiveness of onotherapy on children with SLD by improving self-esteem and school performance. Sixteen patients with a diagnosis of dyslexia were randomly assigned to two treatment groups: the first was a conventional neuropsychological group therapy without onotherapy, and the second was a neuropsychological group therapy incorporating AAIs with therapy donkeys. The neuropsychological assessment included the WISC-IV, DDE and the TMA test, which were administered before and after the treatment in both groups. The results of the experimental group show significant improvement in word reading test correctness ($$p \leq 0.03$$) and speed ($$p \leq 0.03$$), non-word reading test speed ($$p \leq 0.01$$), reading text test correctness ($$p \leq 0.05$$) and speed ($$p \leq 0.03$$), word writing test correctness ($$p \leq 0.01$$), non-word writing test correctness ($$p \leq 0.02$$), writing sentences with homophonic words correctness ($$p \leq 0.01$$), interpersonal TMA ($$p \leq 0.04$$) and the total TMA ($$p \leq 0.04$$), which were significative. On the other hand, in the control group, significant differences were found in word reading test speed ($$p \leq 0.01$$), non-word reading test speed ($$p \leq 0.04$$), reading text test speed ($$p \leq 0.02$$), writing word test correctness ($$p \leq 0.01$$), writing non-word test correctness ($$p \leq 0.01$$) and writing sentences with homophonic words ($$p \leq 0.01$$). However, in this group, we observed no significant difference in the esteem of children. Training associated with the donkeys determined improved scholastic performances as far as reading is concerned and a change in self-esteem. Therefore, we can state that AAIs for dyslexia could be a viable and effective option to enhance the rehabilitation process, increase self-esteem and improve cognitive functions and language skills recovery. ## 1. Introduction Specific learning disorders affect 1–$2.5\%$ of the general population in the Western world, and they include a diverse group of ailments in which children with preserved intellectual capacities have problems with processing information or generating output. Learning difficulties often lead to alterations in neurocognitive processes that can manifest themselves as a deficient ability to read, speak, write, organize information, solve mathematical problems, spell, listen or concentrate [1]. Approximately $80\%$ of people affected by learning disabilities have dyslexia, which is the most frequent learning disability [2,3,4,5,6,7]. Their etiologies are multifactorial and reflect both genetic influences [3,4] and dysfunctions in the verbal systems. Learning Disorders (SLDs) occur in an almost constant association with other disorders (comorbidities); this determines the marked heterogeneity of the profiles and the expressiveness with which SLDs occur, and it has significant repercussions on diagnostic investigations. Reading disability, or dyslexia, is a learning disorder based on receptive language characterized by difficulties with decoding, fluent word recognition, automatic naming skills and/or reading comprehension skills [8,9]. It generally involves severe functional impairment and the permanent need for support and interventions. Early rehabilitation was found to be a remedy in many children with dyslexia [9,10]. Learning disorders have generated several vision-based diagnostic methods and therapeutic procedures [11], sometimes without scientific support [1]. Rapid progress is being made in understanding the etiopathogenesis of many learning disorder syndromes [12,13,14], but aspects such as quality of life and the impact of the disease on families of people with learning difficulties need further investigation [15]. A very important aspect that should not be underestimated is the psychological state of the child for the situation they are living in, and this can have serious repercussions both on their work at school and their personal lives. Often the greatest difficulty is given by the onset of low self-esteem in subjects with learning disabilities. Dyslexia is a disability that evolves over time with results that are not easily predictable. Each dyslexic child can present with a different evolution, which depends on the multiple concomitant factors that can improve or worsen the evolution. Recent studies have examined some factors that can improve or worsen reading performance. Among these factors, Jakovljevic et al. found that gender [16] and color modifications [17] in text background and overlay influence their reading, with turquoise background/overlays and yellow backgrounds improving their performance. Moreover, some authors [18] have described the emotional and motivational aspects that characterize a child with an SLD, stating that a child with an SLD often presents a more negative self-concept, lower self-esteem, is more anxious and feels less emotional support. The construct of well-being is very complex, and various factors intervene, such as physical and psychological health, the quality of family and friends relationships, the sense of self-efficacy and self-esteem and emotional experiences. It has been seen that among children and adolescents, a good level of self-esteem is an indication of a healthy lifestyle [19]. Some data in the literature [20] demonstrate that self-esteem in children with neuropsychiatric problems, including learning disabilities, must be assessed carefully, especially in the female gender, and there is a need for measures to prevent a trajectory toward psychopathology. Other studies have shown that therapy involving animals with people affected by intellectual impairments reports positive improvement for all psychosocial outcomes [21]. Therapies using animals to improve cognitive, emotional, physical and social functioning are called animal-assisted interventions (AAIs) [22]. The rehabilitation of children with disabilities can include the inclusion of animal-assisted therapy, as it is positively accepted by both children and families [23]. This type of therapy is used with children with different diagnoses: ADHD [24], communication disorders [25,26], Down’s Syndrome [27], ASD [28,29,30] and Cerebral Palsy [31,32]. Although the results show positive effects in relation to the treatment of children with autism, the diversity of scales used to measure the outcomes makes it occasionally difficult to establish its efficacy [33]. Moreover, these programs are used by several education and health professionals, and the results may vary depending on the participants and the animal used in the therapy [34]. Recently, some studies have shown the efficacy of equine-assisted therapy in improving the main symptoms of children with attention deficit hyperactivity disorder [35,36]. Many studies have shown that equine therapy is effective in improving self-esteem in children [37,38,39,40]; however, no studies have highlighted the effects of onotherapy on self-esteem and, consequently, on academic performance in children with learning disabilities. Donkey-assisted therapy for patients with SLDs could represent a valid treatment strategy as part of a multimodal therapy for children with SLDs and not just a playful-recreational activity and could lead to improved self-esteem in children with SLDs. The donkey should act as a facilitator of communication between the caregiver and the patient. Children benefit greatly from contact with the animal because of its sensitivity and the natural empathy that is established between them [41]. Moreover, during onotherapy sessions, children could improve self-regulation [42] and self-care [43] and learn new skills that can be generalized to everyday life by promoting their autonomy, openness to others and self-esteem. Our study was aimed at demonstrating the effects of onotherapy on reading, writing and self-esteem in children with dyslexia. This constitutes an innovative aspect since there are currently no studies in the literature showing the effect of onotherapy on self-esteem and school performance in children with SLDs. ## 2.1. Study Design and Population This study was conducted according to the Declaration of Helsinki. Our Institutional Ethics Committee approved the study protocol on 24 February 2021. Parents provided informed consent in both verbal and written forms. Sixteen patients with a diagnosis of dyslexia (8 males and 8 females, a range of 7–12 years, with a median age of 9 years) with a range of 3–6 years of education that had been admitted to the Scientific Research and Care Institute “IRCCS Centro Neurolesi Bonino Pulejo” of Messina, were enrolled in this study. All of the participants were assigned randomly to one of two treatment groups: (a) a conventional neuropsychological group therapy without onotherapy (Control Group: CG $$n = 8$$); (b) a neuropsychological group therapy incorporating an AAI with therapy donkeys (Experimental Group: EG $$n = 8$$). Randomization minimizes the selection effect, and the two comparison groups allow us to determine the possible effects of the treatment compared to the group without treatment (control), while the other variables were kept constant. The inclusion criteria for this study were as follows: [1] a diagnosis of dyslexia according to the Diagnostic Statistical Manual, Fifth Edition (DSM-5); [2] aged between 8 and 11 years; [3] Q.I. normal (WISC score less than 80); and [4] the presence of low self-esteem (TMA score less than 86). The exclusion criteria were as follows: [1] the presence of other neuropsychiatric pathologies; [2] the presence of comorbidities; and [3] the presence of family conflicts that can affect self-esteem. ## 2.2. Outcome Measures The neuropsychological assessment in this study included the Wechsler Intelligence Scale for Children-IV (WISC-IV) [43], the Italian Battery for Evaluation of Dyslexia and Dysorthography (DDE) [44] and the TMA test (Multidimensional Self-Esteem Test) for the evaluation of self-esteem age of development in its many dimensions. The WISC-IV is a standardized clinical tool used for the cognitive assessment of children and young people aged between 6 and 16 years. The four cognitive areas assessed by WISC IV correspond to specific indices, as follows: the Verbal Comprehension Index (VCI), the Visual–Perceptual Reasoning Index (PRI), the Working Memory Index (WML) and the Processing Speed Index (PSI). Moreover, WISC-IV allows, from the sum of these indices, the elaboration of three composite ones, as follows: one index for Global Intellectual Quotient (IQ), one index for General Ability (IAG) and at least one for Cognitive Competence (ICC). WISC-IV is mainly used to support diagnostic hypotheses in the assessment of children with specific learning disorders (SLDs). In Italy, in fact, to diagnose an SLD, the subject must have an IQ of at least 85 and a significant discrepancy between their scholastic performance and their IQ [45]. The DDE is a battery used in the assessment of the level of competence acquired in writing and reading, and it is very useful for monitoring progress during and after treatment. The DDE is divided into 8 tests. Of these tests, 5 aim to assess and analyze the reading process, requiring the subject to name graphemes, read words and non-words, understand sentences with homophones and correct homophones. On the other hand, the remaining 3 tests of the battery aim to assess and analyze the reading process by asking the subject to write words and non-words under dictation and write sentences with homophonic words under dictation. This battery has been recognized by the Italian Dyslexia Association as part of the basic diagnostic protocol for the evaluation of disorders in terms of writing, reading and calculation. It also makes it possible to compare pre- and post-treatment performance and promote communication between rehabilitation centers and operators. The TMA [46] assesses self-esteem by deepening six areas, as follows: the interpersonal area (evaluating a subject’s perception of their social relationships with peers and adults), the school area (a subject’s sense of success or failure in the classroom), the emotional area (delving into the emotional sphere and assessing the ability to control negative emotions), the family area (a subject’s perceptions of family relationships and the degree to how important and loved they feel, etc.), the body area (a subject’s perception of their appearance, physical and sporting skills, etc.) and the area of mastery over the environment (a subject’s feeling of being able to dominate the events of one’s life, etc.). ## 2.3. Procedures One group (EG) underwent bi-weekly individual neuropsychological training associated with donkey-assisted therapy training once a week, and a control group (CG) underwent traditional individual neuropsychological training. Children with SLDs were recruited and diagnosed at our Child Neuropsychiatry clinic with the following battery: detailed medical history, neuropsychiatric visit and WISC-IV for an intelligence assessment. School performance was assessed using the DDE battery for the assessment of reading, writing and arithmetic. The level of self-esteem was assessed through the Italian version of the Multidimensional Self-Esteem Test—TMA. The training lasted for 6 months, at the end of which the children were re-checked again with the administration of school tests and TMA. ## 2.4. Donkey-Assisted Training Each training session lasted 45 min and was divided into 5 phases of 10 min each, unlike the last one, which lasted 5 min. The first phase of donkey-assisted training consists of interaction with the donkey and an introduction to the environment in which they live. This was followed by a phase of grooming, cleaning and physical contact to teach the children basic safety rules, animal anatomy and etiology and animal management issues. The third phase involved the children conducting the donkey along the path used for exercise, and the fourth phase was dedicated to saddling, riding and leading the children on the donkey in the field while invited to perform specific exercises. From these two phases, the children could learn basic aspects of riding (such as guiding the donkey around objects, mounting, dismounting, positioning, walking and trotting). This allows the children to increase motor abilities and coordination, self-esteem and sensor perception. During the last phase, the children were dismounted and encouraged to socialize and relate to the animal for greetings (saying “goodbye” and hugging the donkey). ## 3. Statistical Analysis With regard to statistical analysis, nonparametric analysis was performed because the results of the Shapiro normality test highlighted a non-normal distribution of most of the target variables. The numerical data are presented as the median and the first-third quartile as a non-normal distribution. The Wilcoxon signed-rank test and the Mann–Whitney U test were utilized for intra and inter-group analysis, respectively. An interaction effect analysis (improved time) was performed by submitting the T1–T0 differences in the variables scores to correlation and regression analyses. Spearman correlation was used to evaluate whether there was a relationship between the DDE battery and the TMA sub-test. The analyses were performed using an open source R3.0 software package. The confidence level was set to $95\%$ with a $5\%$ alpha error. Statistical significance was set at $p \leq 0.05.$ ## 4. Results The Wilcoxon signed-rank test showed a significant difference in the experimental group between T0 and T1 (Table 1). In particular, word reading test correctness ($$p \leq 0.03$$) and speed ($$p \leq 0.03$$), non-word reading test speed ($$p \leq 0.01$$), reading text test correctness ($$p \leq 0.05$$) and speed ($$p \leq 0.03$$), word writing test correctness ($$p \leq 0.01$$), non-word writing test correctness ($$p \leq 0.02$$), writing sentences with homophonic words correctness ($$p \leq 0.01$$), interpersonal TMA ($$p \leq 0.04$$) and total TMA ($$p \leq 0.04$$) were significant. Moreover, a trend was present between non-word reading test correctness ($$p \leq 0.07$$), emotional TMA ($$p \leq 0.08$$) and scholastic TMA ($$p \leq 0.08$$) between T0 and T1. In the control group, we found significant differences in word reading test speed ($$p \leq 0.01$$), non-word reading test speed ($$p \leq 0.04$$), reading text test speed ($$p \leq 0.02$$), writing word test correctness ($$p \leq 0.01$$), writing non-word test correctness ($$p \leq 0.01$$) and writing sentences with homophonic words ($$p \leq 0.01$$). In this group, we have no heightened significant difference between T0 and T1 in the TMA sub-test. In the inter-group, we found a significant difference in the word writing test ($$p \leq 0.02$$) at T1 (Table 1). Spearman correlation showed a trend between word reading test speed and scholastic TMA ($$p \leq 0.07$$) in the experimental group (Figure 1). People with specific learning disorders may have lower levels of self-esteem and present more difficulties in terms of emotional and behavioral fields than those without dyslexia. However, the nature of the relationship between self-esteem and psychopathology remains unknown [47]. This study reports results from a small sample of patients with dyslexia and low self-esteem involved in a 6-month AAI intervention compared to an activity control group that had no interaction with a donkey. The results show significant post-intervention improvements in the AAI group compared to the group control. In fact, the results obtained confirm that onotherapy associated with traditional therapy can help improve school performance and self-esteem in children with dyslexia. We considered only children between the ages of 8 and 11 to avoid the presence of subjects who might have adolescent problems that could result in lowered self-esteem. We also excluded children with socio-familial problems, considering that self-esteem can be damaged by situations other than dyslexia. There were significant improvements in both the control group and the experimental group. While in the control group, the improvements concern only word reading test speed, non-word reading test speed, reading passage speed test, writing word test correctness, writing non-word test correctness and writing homophone word test correctness, but there were no improvements in the esteem of the children, in the group who underwent onotherapy associated with traditional training, in addition to the improvements already recorded in the control group, there is also an improvement in the correctness of reading the passage but above all in both global and interpersonal self-esteem. It is evident that the positive change in dyslexia is the result of the traditional cognitive training of reinforcement of the deficient areas, but the training associated with the donkey determined improved scholastic performances as far as reading is concerned and a positive change in the self-esteem values, which were not present in the control group. There is a lack of high-quality evidence on which to base any decision about the use of animal-assisted therapy for dyslexic children. Previous studies have demonstrated that children with reading difficulties presented decreased emotional and executive function abilities [48]. There is a consistent association between academic self-esteem, emotional symptoms and internalizing difficulties in dyslexia. There are multiple treatments and interventions that ameliorate dyslexia symptoms in children [49]. It is extremely important to evaluate the etiology of the disturbances to adequately plan the intervention. In fact, any intervention must also take into consideration any comorbidities to be as complete as possible [50]. The various intervention methods used to improve the literacy, cognitive function and reading ability of children with dyslexia include the use of colored backgrounds or layouts [16,17], multisensory and multimedia methods [51] and virtual reality [52,53]. The combination of cognitive functions, linguistic literacy deficits and self-esteem interventions is not provided in any study in the literature. Children with dyslexia present with deficits in higher-order processing or executive control processes. They also have peculiar eye movement tendencies [54], deficits in visual attention span, processing speed, verbal working memory [55,56] and difficulties with memorizing. Animal-assisted interventions represent an innovative rehabilitation approach that can improve the social, emotional and physical aspects of several diseases. Few data are available regarding donkey therapy. The results support that the use of donkey therapy for children with SLDs can improve self-esteem and obtain improved learning performance. The outcomes of this study suggest that there is an important function in the child–donkey interaction that can affect positive changes in terms of self-esteem and learning. Our results generate hypotheses regarding the role of the child–donkey interaction requiring further investigation. One hypothesis is that working together with the donkey involves the stimulation of nonverbal joint attention, shared attention, executive attention and memory that serve as a platform for improving linguistic abilities. Additionally, motor training with the donkey also has positive effects on several cognitive domains, including attention, memory, processing speed and inhibition beyond mood and mobility. Another hypothesis is that the child’s experience with the equine (i.e., the warmth of the donkey’s body) creates a calming context, which could have a calming effect on children with dyslexia and, consequently, improve their performance in school tests due to an increased ability to concentrate. In addition, motor training with the donkey, featured in our protocol, could bring about positive effects on several cognitive domains, including memory, attention, inhibition and processing speed, in addition to acting on mood and mobility. All of this would, in our opinion, promote the improvement of self-esteem in the experimental group, as evidenced by the results obtained, reporting an increase in the total score on the total TMA score and in the interpersonal TMA score. Self-esteem could be improved thanks to the human–donkey interaction defined as “total grooming”, as it is linked to the idea of parental care, where the patient is rocked and pampered by the parents [57]. The relationship and the emotional attachments that children establish with a donkey has a positive influence on several fronts. Social functioning seems to benefit from the children–animal contact as well as executive abilities [58]. Specific reference is made to animal contact because, occasionally, this in itself has been shown to improve the individual’s behavior, executive function, sustained attention, working memory and probably even emotional control [59]. Another aspect of the equines that can elicit a positive change in patients is its movement. In fact, the animal, when moving, stimulates several systems simultaneously, such as the limbic, sensory, skeletal, muscular and vestibular systems [60]. Moreover, a recent meta-analysis [61] specifies that Equine-Assisted Therapies reduce reaction time in problem-solving situations and maladaptive behaviors while improving socialization and engagement in ASD patients. A diagnostic category that is often not referred to when discussing equine-assisted therapy is mental illness. A study conducted by Punzo et al. found that children and adolescents suffering from mental illness find relief from everyday anxieties and fears and improve their self-esteem thanks to the friendship established with the animal, strengthening their self-confidence [62]. Similar results were found by Alfonso et al. [ 63], who noted in their study that an equine-assisted intervention was highly effective in reducing symptoms and signs of social anxiety in young women aged 18–29. These results might make one consider the idea of combining equine-assisted therapy with psychiatric treatments to decrease the prescription of psychiatric drugs. Recent studies have investigated the effects of equine-assisted therapy in fields of application that are still partially unexplored, namely in the treatment of obesity, substance use disorders, post-traumatic stress disorder (PTSD) and work-related stress. An exploratory case study of Schroeder et al. [ 64] shows that equine therapy is perceived as acceptable and enjoyable by the participants, who benefit from it as it increases their physical activity and their self-efficacy for physical exercise. Similar studies had already been carried out on at-risk adolescents and their body image, self-control, confidence and satisfaction [65], but not in patients with obesity. The intervention was also found to have a positive effect on the treatment of patients suffering from substance use disorders [66]. The latter perceived the contact with the animal as a break from traditional treatments, and positive results were recorded both in the retention and completion of treatment. As regards post-traumatic stress disorder, a study conducted on veterans expressed the potential of AAI in reducing PTSD symptoms [67]. Finally, the treatment also seems to find use in work environments. Healthcare workers are among the categories most affected by stress and burnout, and this is why a very recent study investigated the safety, feasibility and perception of participants with a resiliency intervention that included AAIs [68]. The results showed that the participants found it enjoyable, and there was a significant increase in psychological flexibility. While conventional wisdom has always affirmed the value of animals in promoting human well-being, only recently has their therapeutic role in medicine become the focus of dedicated research [22]. We believe that a donkey AAI has led to better outcomes through specific animal training that encourages active communication, enhances engagement, brings enjoyment and motivation and increases treatment compliance. The latter strongly relates to another fundamental aspect that is often put in the background or even neglected: satisfaction. In their review, Siewertsen et al. [ 69] have shown that pet therapy (with both dogs and equines) has a high satisfaction index not only among participants but also and especially among their families. When satisfaction is high, it tends to increase the commitment and frequency with which parents are willing to place their resources into treatment, thus limiting the risk of drop-out. However, although the benefits of equine-assisted activities and therapies have been amply demonstrated by many studies, the implementation of this type of treatment remains limited due to the lack of standardized treatments and systematic theory-based knowledge. This is the first study evaluating the effects of an AAI in children with dyslexia and low self-esteem. Our study also showed a trend of improvement after the AAI in terms of emotional and school TMA. These data could be confirmed with the significance of the results by increasing the sample size. Effectively, the main limitations of this study consist of the relatively small sample size, which limits the generalizability of our data and the lack of an adequate long-term follow-up. Therefore, we can state that the use of AAIs for dyslexia could be feasible and effective as it allows us to enhance the rehabilitation process, improving self-esteem and increasing the recovery of language skills as well as cognitive functions. ## 5. Conclusions In conclusion, this study demonstrates that an AAI using donkeys in a rehabilitation program could be a motivational and effective tool for enhancing language skills, favoring cognitive functions, promoting psychological well-being and improving self-esteem. Further larger sample studies with long-term follow-up periods are needed to confirm the effect of this interesting approach. The conclusions are preliminary, and the limitations of the study include the small number of subjects and a lack of a longer follow-up, and that it is not randomized. The value of the current small study only points in one direction, increasing self-esteem in children with dyslexia, and no firm conclusions can be drawn yet. ## References 1. Handler S.M., Fierson W.M.. Section on Ophthalmology; Council on Children with Disabilities; American Academy of Ophthalmology; American Association for Pediatric Ophthalmology and Strabismus; American Association of Certified Orthoptists. Learning disabilities, dyslexia, and vision. Pediatrics (2011) 127 e818-e856. DOI: 10.1542/peds.2010-3670 2. Shaywitz S.E.. Dyslexia. N. Engl. J. Med. (1998) 338 307-312. DOI: 10.1056/NEJM199801293380507 3. Shaywitz S.E., Shaywitz B.A.. The science of reading and dyslexia. J. AAPOS (2003) 7 158-166. DOI: 10.1016/S1091-8531(03)00002-8 4. Lyon G.R.. Learning disabilities. Futur. Child. (1996) 6 54-76. DOI: 10.2307/1602494 5. Shaywitz S.E., Shaywitz B.A.. The Neurobiology of Reading and Dyslexia. ASHA Lead. (2007) 12 20-21. DOI: 10.1044/leader.FTR7.12122007.20 6. Lyon G.R.. Report on Learning Disabilities Research. (2010) 7. Shaywitz S.E.. Overcoming Dyslexia: A New and Complete Science Based Program for Overcoming Reading Problems at Any Level (2003) 8. Lyon G.R., Shaywitz S.E., Shaywitz B.A.. Una definizione di dislessia. Ann. Dyslexia. (2003) 53 1-14. DOI: 10.1007/s11881-003-0001-9 9. Pennington B.F.. Diagnosi dei Disturbi Dell’apprendimento. Un Quadro Neuropsicologico (2009) 10. Fletcher J.M., Lione G.R., Fuchs L.S., Barnes M.A.. Disturbi Dell’Apprendimento: Dall’Identificazione All’Intervento (2007) 11. Anis M.Y.N., Normah C.D., Mahadir A., Norhayati I., Rogayah A.R., Dzalani H.. Interventions for children with dyslexia: A review on current intervention methods. Med. J. Malays. (2018) 73 311-320 12. Şahin İ., Say G.N., Avcı B., Kesim N.. Low serum allopregnanolone levels in children with attention deficit hyperactivity disorder. Psychoneuroendocrinology (2022) 146 105923. DOI: 10.1016/j.psyneuen.2022.105923 13. Caylak E.. Neurobiological approaches on brains of children with dyslexia. Acad. Radiol. (2009) 16 1003-1024. DOI: 10.1016/j.acra.2009.02.012 14. Menghini D., Hagberg G.E., Petrosini L., Bozzali M., Macaluso E., Caltagirone C., Vicari S.. Structural correlates of implicit learning deficits in subjects with developmental dyslexia. Ann. N. Y. Acad. Sci. (2008) 1145 212-221. DOI: 10.1196/annals.1416.010 15. Gilberg C., Soderotom H.. Learning Disability. Lancet (2003) 362 811-821. DOI: 10.1016/S0140-6736(03)14275-4 16. Jakovljević T., Janković M.M., Savić A.M., Soldatović I., Mačužić I., Jakulin T.J., Papa G., Ković V.. The effect of colour on reading performance in children, measured by a sensor hub: From the perspective of gender. PLoS ONE (2021) 16. DOI: 10.1371/journal.pone.0252622 17. Jakovljević T., Janković M.M., Savić A.M., Soldatović I., Čolić G., Jakulin T.J., Papa G., Ković V.. The relation between physiological parameters and colour modifications in text background and overlay during reading in children with and without dyslexia. Brain Sci. (2021) 11. DOI: 10.3390/brainsci11050539 18. Moè A., De Beni R., Cornoldi C., Comoldi C.. Difficoltà dell’apprendimento: Aspetti emotivo- motivazionali. Difficoltá e Disturbi Dell’apprendimento (2007) 253-271 19. Zangirolami F., Iemmi D., Vighi V., Pellai A.. Evaluating self-esteem modifications after a Life Skills-Based Education intervention. Minerva Pediatr. (2018) 70 340-344. DOI: 10.23736/S0026-4946.16.04639-9 20. Ek U., Westerlund J., Holmberg K., Fernell E.. Self-esteem in children with attention and/or learning deficits: The importance of gender. Acta Paediatr. (2008) 97 1125-1130. DOI: 10.1111/j.1651-2227.2008.00894.x 21. Maber-Aleksandrowicz S., Avent C., Hassiotis A.. A Systematic Review of Animal-Assisted Therapy on Psychosocial Outcomes in People with Intellectual Disability. Res. Dev. Disabil. (2016) 49–50 322-338. DOI: 10.1016/j.ridd.2015.12.005 22. Muñoz Lasa S., Máximo Bocanegra N., Valero Alcaide R., Atín Arratibel M.A., Varela Donoso E., Ferriero G.. Animal assisted interventions in neurorehabilitation: A review of the most recent literature. Neurología (2015) 30 1-7. DOI: 10.1016/j.nrl.2013.01.012 23. Yap E., Scheinberg A., Williams K.. Attitudes to and beliefs about animal assisted therapy for children with disabilities. Complement. Ther. Clin. Pract. (2017) 26 47-52. DOI: 10.1016/j.ctcp.2016.11.009 24. Gilboa Y., Helmer A.. Self-Management Intervention for Attention and Executive Functions Using Equine-Assisted Occupational Therapy among Children Aged 6–14 Diagnosed with Attention Deficit/Hyperactivity Disorder. J. Altern. Complement. Med. (2020) 26 239-246. DOI: 10.1089/acm.2019.0374 25. Oliveira G.R., Ichitani T., Cunha M.C.. Animal Assisted Activity: Effects in communication and interpersonal relations inside the school environment. Distúrb Comun. (2016) 28 759-763 26. Machová K., Kejdanová P., Bajtlerová I., Bajtlerová R., Svobodová I., Mezian K.. Canine-assisted Speech Therapy for Children with Communication Impairments: A Randomized Controlled Trial. Anthrozoos (2018) 31 587-598. DOI: 10.1080/08927936.2018.1505339 27. Satiansukpong N., Pongsaksri M., Sasat D.. Thai Elephant-Assisted therapy programme in children with down syndrome. Occup. Ther. Int. (2016) 23 121-131. DOI: 10.1002/oti.1417 28. Ward S.C., Whalon K., Rusnak K., Wendell K., Paschall N.. The association between therapeutic horseback riding and the social communication and sensory reactions of children with autism. J. Autism. Dev. Disord. (2013) 43 2190-2198. DOI: 10.1007/s10803-013-1773-3 29. Tan V.X.L., Simmonds J.G.. Parent Perceptions of Psychosocial Outcomes of Equine-Assisted Interventions for Children with Autism Spectrum Disorder. J. Autism. Dev. Disord. (2018) 48 759-769. DOI: 10.1007/s10803-017-3399-3 30. O’Haire M.E., McKenzie S.J., McCune S., Slaughter V.. Effects of classroom animal-assisted activities on social functioning in children with autism spectrum disorder. J. Altern. Complement. Med. (2014) 20 162-168. DOI: 10.1089/acm.2013.0165 31. Lucena-Antón D., Rosety-Rodríguez I., Moral-Munoz J.A.. Effects of a hippotherapy intervention on muscle spasticity in children with cerebral palsy: A randomized controlled trial. Complement. Ther. Clin. Pract. (2018) 31 188-192. DOI: 10.1016/j.ctcp.2018.02.013 32. Hsieh Y.L., Yang C.C., Sun S.H., Chan S.Y., Wang T.H., Luo H.J.. Effects of hippotherapy on body functions, activities and participation in children with cerebral palsy based on ICF-CY assessments. Disabil Rehabil. (2017) 39 1703-1713. DOI: 10.1080/09638288.2016.1207108 33. Charry-Sánchez J.D., Pradilla I., Talero-Gutiérrez C.. Efficacia della terapia assistita da animali nella popolazione pediatrica: Revisione sistematica e meta-analisi di studi controllati. J. Dev. Behav. Pediatr. (2018) 39 580-590. DOI: 10.1097/DBP.0000000000000594 34. Mandrá P.P., Moretti T.C.D.F., Avezum L.A., Kuroishi R.C.S.. Animal assisted therapy: Systematic review of literature. Terapia assistida por animais: Revisão sistemática da literature. CoDAS (2019) 31 e20180243. DOI: 10.1590/2317-1782/20182018243 35. Jang B., Song J., Kim J., Kim S., Lee J., Shin H.Y., Kwon J.Y., Kim Y.H., Joung Y.S.. Equine-Assisted Activities and Therapy for Treating Children with Attention-Deficit/Hyperactivity Disorder. J. Altern. Complement. Med. (2015) 21 546-553. DOI: 10.1089/acm.2015.0067 36. Yoo J.H., Oh Y., Jang B., Song J., Kim J., Kim S., Lee J., Shin H.Y., Kwon J.Y., Kim Y.H.. The Effects of Equine-assisted Activities and Therapy on Resting-state Brain Function in Attention-deficit/Hyperactivity Disorder: A Pilot Study. Clin. Psychopharmacol. Neurosci. Off. Sci. J. Korean Coll. Neuropsychopharmacol. (2016) 14 357-364. DOI: 10.9758/cpn.2016.14.4.357 37. Mayberry R.P.. The mystique of the horse is strong medicine: Riding as therapeutic recreation. Rehabil. Lit. (1978) 39 192-196. PMID: 694220 38. Oh Y., Joung Y.S., Jang B., Yoo J.H., Song J., Kim J., Kim K., Kim S., Lee J., Shin H.-Y.. Efficacy of Hippotherapy versus pharmacotherapy in Attention-Deficit/Hyperactivity Disorder: A randomized clinical trial. J. Altern. Complement. Med. (2018) 24 463-471. DOI: 10.1089/acm.2017.0358 39. Mendizábal Alonso P.. Intervenciones fisioterápicas mediante hipoterapia en el tratamiento de la parálisis cerebral infantil. Revisión bibliográfica [Physiotherapy interventions through hippotherapy in the treatment of cerebral palsy. A literature review]. Rehabilitacion (2020) 54 96-106. DOI: 10.1016/j.rh.2019.11.003 40. Wilson K., Buultjens M., Monfries M., Karimi L.. Equine-Assisted Psychotherapy for adolescents experiencing depression and/ or anxiety: A therapist’s perspective. Clin. Child Psychol. Psychiatry (2017) 22 16-33. DOI: 10.1177/1359104515572379 41. Maresca G., Portaro S., Naro A., Crisafulli R., Raffa A., Scarcella I., Aliberti B., Gemelli G., Calabrò R.S.. Hippotherapy in neurodevelopmental disorders: A narrative review focusing on cognitive and behavioral outcomes. Appl. Neuropsychol. Child (2020) 11 553-560. DOI: 10.1080/21622965.2020.1852084 42. Gabriels R.L., Pan Z., Dechant B., Agnew J.A., Brim N., Mesibov G.. Randomized Controlled Trial of Therapeutic Horseback Riding in Children and Adolescents with Autism Spectrum Disorder. J. Am. Acad. Child Adolesc. Psychiatry (2015) 54 541-549. DOI: 10.1016/j.jaac.2015.04.007 43. McDaniel Peters B.C., Wood W.. Autism and Equine-Assisted Interventions: A Systematic Mapping Review. J. Autism Dev. Disord. (2017) 47 3220-3242. DOI: 10.1007/s10803-017-3219-9 44. Orsini A., Pezzuti L., Picone L.. WISC-IV: Contributo Alla Taratura Italiana (WISC-IV Italian) (2012) 45. Sartori G., Job R.. DDE-2: Batteria per la Valutazione della Dislessia e della Disortografia Evolutiva-2. Assessment Battery for Developmental Reading and Spelling Disorders (2007) 46. Istituto Superiore di Sanità Consensus Conference on Learning Disabilities. (2011) 47. Bracken B.A.. Test TMA, Valutazione Multidimensionale Dell’autostima (1993) 48. Terras M.M., Thompson L.C., Minnis H.. Dyslexia and psychosocial functioning: An exploratory study of the role of self-esteem and understanding. Dyslexia (2009) 15 304-327. DOI: 10.1002/dys.386 49. Nachshon O., Horowitz-Krauss T.. Cognitive and emotional challenges in children with reading difficulties. Acta Pediatr. (2019) 108 1110-1114. DOI: 10.1111/apa.14672 50. Duff F.J., Clarke P.J.J.. Reading disorders: What are the effective interventions and how should they be implemented and evaluated?. Child Psychol Psychiatry (2011) 52 3-12. DOI: 10.1111/j.1469-7610.2010.02310.x 51. Snowling M.J.. Early identification and interventions for dyslexia: A contemporary view. J. Res. Spec. Educ. Needs (2013) 13 7-14. DOI: 10.1111/j.1471-3802.2012.01262.x 52. Pedroli E., Padula P., Guala A., Meardi M.T., Riva G., Albani G.. A Psychometric Tool for a Virtual Reality Rehabilitation Approach for Dyslexia. Comput. Math. Methods Med. (2017) 2017 7048676. DOI: 10.1155/2017/7048676 53. Kiper P., Szczudlik A., Mirek E., Nowobilski R., Opara J., Agostini M., Turolla A.. The application of virtual reality in neuro-rehabilitation: Motor re-learning supported by innovative technologies. Med. Rehabil. (2013) 17 29-36. DOI: 10.5604/01.3001.0009.3087 54. Vajs I., Ković V., Papić T., Savić A.M., Janković M.M.. Spatiotemporal eye-tracking feature set for improved recognition of dyslexic reading patterns in children. Sensors (2022) 22. DOI: 10.3390/s22134900 55. Maresca G., Leonardi S., De Cola M.C., Giliberto S., Di Cara M., Corallo F., Quartarone A., Pidalà A.. Use of Virtual Reality in Children with Dyslexia. Children (2022) 9. DOI: 10.3390/children9111621 56. Poblano A., Valadéz-Tepec T., Arias M.D.L., García-Pedroza F.. Phonological and visuo-spatial working memory alterations in dyslexic children. Arch. Med. Res. (2000) 31 493-496. DOI: 10.1016/S0188-4409(00)00096-5 57. Portaro S., Maresca G., Raffa A., Gemelli G., Aliberti B., Calabrò R.S.. Donkey Therapy and Hippo-therapy: Two Faces of the Same Coin?. Innov. Clin. Neurosci. (2020) 17 20-21. PMID: 32547842 58. Borgi M., Loliva D., Cerino S., Chiarotti F., Venerosi A., Bramini M., Nonnis E., Marcelli M., Vinti C., De Santis C.. Effectiveness of a Standardized Equine-Assisted Therapy Program for Children with Autism Spectrum Disorder. J. Autism Dev. Disord. (2016) 46 1-9. DOI: 10.1007/s10803-015-2530-6 59. Norwood M.F., Lakhani A., Maujean A., Downes M., Fullagar S., Barber B.L., Kendall E.. The Horse as a Therapist: Effects of an Equine Program without “Therapy” on the Attention and Behavior of Youth Disengaged from Traditional School. J. Altern. Complement. Med. (2021) 27 678-687. DOI: 10.1089/acm.2020.0500 60. Granados A.C., Agís I.F.. Why children with special needs feel better with hippotherapy sessions: A conceptual review. J. Altern. Complement. Med. (2011) 17 191-197. DOI: 10.1089/acm.2009.0229 61. Trzmiel T., Purandare B., Michalak M., Zasadzka E., Pawlaczyk M.. Equine assisted activities and therapies in children with autism spectrum disorder: A systematic review and a meta-analysis. Complement. Ther. Med. (2019) 42 104-113. DOI: 10.1016/j.ctim.2018.11.004 62. Punzo K., Skoglund M., Carlsson I.M., Jormfeldt H.. Experiences of an Equine-Assisted Therapy Intervention among Children and Adolescents with Mental Illness in Sweden—A Nursing Perspective. Issues Ment. Health Nurs. (2022) 43 1080-1092. DOI: 10.1080/01612840.2022.2126571 63. Alfonso S.V., Alfonso L.A., Llabre M.M., Fernandez M.I.. Project Stride: An Equine-Assisted Intervention to Reduce Symptoms of Social Anxiety in Young Women. Explore (2015) 11 461-467. DOI: 10.1016/j.explore.2015.08.003 64. Schroeder K., Van Allen J., Dhurandhar E., Lancaster B., Heidari Z., Cazenave K., Boone D., Erdman P.. Riding into Health: A Case Study on an Equine-Assisted Childhood Obesity Intervention. Int. J. Environ. Res. Public Health (2019) 16. DOI: 10.3390/ijerph16234835 65. Bachi K., Terkel J., Teichman M.. Equine-facilitated psychotherapy for at-risk adolescents: The influence on self-image, self-control and trust. Clin. Child Psychol. Psychiatry (2012) 17 298-312. DOI: 10.1177/1359104511404177 66. Diaz L., Gormley M.A., Coleman A., Sepanski A., Corley H., Perez A., Litwin A.H.. Equine-assisted services for individuals with substance use disorders: A scoping review. Subst. Abus. Treat. Prev. Policy (2022) 17 81. DOI: 10.1186/s13011-022-00506-x 67. Fisher P.W., Lazarov A., Lowell A., Arnon S., Turner J.B., Bergman M., Ryba M., Such S., Marohasy C., Zhu X.. Equine-Assisted Therapy for Posttraumatic Stress Disorder Among Military Veterans: An Open Trial. J. Clin. Psychiatry (2021) 82 21m14005. DOI: 10.4088/JCP.21m14005 68. Marchand W.R., Sullivan-Sakaeda L.. A pilot observational study of a psychotherapy incorporating equines resiliency intervention for staff at a large medical center. Complement. Ther. Clin. Pract. (2022) 49 101660. DOI: 10.1016/j.ctcp.2022.101660 69. Siewertsen C.M., French E.D., Teramoto M.. Autism spectrum disorder and pet therapy. Adv. Mind-Body Med. (2015) 29 22-25. PMID: 25831431
--- title: 'Association between Family and School Pressures, Consumption of Ultra-Processed Beverages, and Obesity in Preadolescents: A School-Based Epidemiological Study' authors: - Ioannis Gketsios - Thomas Tsiampalis - Alexandra Foscolou - Ioanna Panagiota Kalafati - Tonia Vassilakou - Aikaterini Kanellopoulou - Venetia Notara - George Antonogeorgos - Andrea Paola Rojas-Gil - Odysseas Androutsos - Ekaterina N. Kornilaki - Areti Lagiou - Demosthenes B. Panagiotakos - Rena I. Kosti journal: Children year: 2023 pmcid: PMC10047258 doi: 10.3390/children10030500 license: CC BY 4.0 --- # Association between Family and School Pressures, Consumption of Ultra-Processed Beverages, and Obesity in Preadolescents: A School-Based Epidemiological Study ## Abstract The aim of the present work was two-fold. Firstly, to evaluate the association between the consumption of ultra-processed beverages (UPB) on preadolescents’ likelihood of being obese. Secondly, to investigate the potential impact of family and school environmental stressors on this unhealthy lifestyle habit. A cross-sectional study was conducted among 1718 Greek preadolescents and their parents, during the school years 2014 to 2016. Parental and child characteristics were collected anonymously, through self-administered and validated questionnaires. Among others, UPB consumption (soft and flavored drinks) was recorded, classifying children as low, moderate, or high consumers, while anthropometric characteristics [height, body weight, Body Mass Index (BMI)] were also recorded. Almost seven out of ten preadolescents were classified as at least moderate UPB consumers, while approximately three out of ten were classified as high UPB consumers. Higher UPB consumption was associated with significantly higher levels of BMI, while preadolescents living in a more stressful family and school environment were found to consume significantly higher amounts of UPB. Stakeholders should implement programs that raise awareness among parents and teachers about the sources of stress in preadolescence as a potential “triggering factor” of unhealthy dietary preferences. ## 1. Introduction Childhood and preadolescence obesity remains a serious global burden, despite public health initiatives [1]. Due to the multifactorial nature of obesity, its prevention and treatment are still unsolved [2]. As several researchers have already demonstrated, a sedentary lifestyle and the consumption of “energy-dense” foods and soft drinks with added sugar, rather than purely endocrinological or genetic factors, are the main environmental causes of the obesity epidemic, even in early childhood [3,4]. It is worth mentioning that the consequences of childhood obesity are tracking into adulthood and increasing the risk of obesity, cardiovascular diseases, type II diabetes mellitus, and premature mortality in later life [1,2]. Soft drinks, i.e., nonalcoholic beverages, frequently carbonated, that contain natural or artificial sweeteners and flavors, and sometimes juice [5], have been linked to the obesity pandemic [6]. However, it is not only soft drinks with added sugars that have been considered potentially harmful. Accordingly, even though there are no clear findings about the effects of concentrated fruit juices and chocolate milk on Body Mass Index (BMI) status [7,8,9], both of them are considered concentrated sources of sugar for children and have been accused of contributing extra calories due to the added sugars [10]. Indeed, in accordance with the recently published Position Paper of the European Academy of Paediatrics and the European Childhood Obesity Group, the consumption of sugar-sweetened beverages (SSB) has to be limited among children and adolescents [11] in an attempt to contain the burden of childhood obesity. Adolescents’ and children’s soft drink intake appears to be significantly influenced by factors such as gender, educational goals, dietary restrictions, ease of access, parental role models, adolescents’ attitudes, and preferences [12,13]. At the same time, stress has recently been revealed to be associated with children’s weight status [14]. Adolescence and preadolescence may be a vulnerable time for social stress [15], which has been related to functional and structural changes in brain systems, crucial for social-affective processing and cognitive regulation [16]. Potential environmental sources of stress among preadolescents may stem from anxiety due to conflict with parents, and peer relationships [17,18] as well as from school-related stress as a result of educational attainment [19]. The findings reported in the literature suggest a bidirectional association between SSB consumption and emotional distress. In particular, drinking SSB on a daily basis has been linked to increased anxiety and depression among youngsters [20], while it has also been suggested that children and adolescents may use soft drink consumption as a dysfunctional strategy for coping with mental health difficulties [21]. It is also known that the consumption of foods that are rich in sugar or fat has been found to distract from stress [22], and thus, alleviating perceived emotional stress [23]. It seems that there may be an interplay between stress, consumption of ultra-processed beverages (UPB), and the likelihood of obesity in preadolescence. However, there is scarce evidence regarding the dominant stressors in preadolescence and their association with UPB consumption. In this context, the hypothesis of the present study was that environmental stress affects UPB consumption in preadolescents, influencing the likelihood of being obese. Thus, the aim of the present work was double-fold: [1] to evaluate the association between the consumption of UPB on preadolescents’ likelihood of being obese; [2] to investigate the potential impact of family and school environmental stressors on this unhealthy lifestyle habit. ## 2.1. Design and Setting This is a cross-sectional, school-based, observational study. The study took place in the metropolitan Athens area, in the Heraklion city area (the capital city of the island of Crete), and in three counties of the Peloponnese peninsula (Sparta, Kalamata, and Pyrgos). The regions represent large urban and rural municipalities from southern Greece. The enrollment procedure was carried out during the school years 2014 to 2016. During the two consecutive school years, a gradual enrollment of schools was conducted. The schools that participated in this study were selected using random sampling from a list provided by the Greek Ministry of Education; all children aged 10–12 were then asked to participate. In total, 47 primary schools (32 from the Athens area, 5 from Heraklion, 3 from Pyrgos, 2 from Kalamata, and 5 from Sparta) were included. More information can be found elsewhere [24]. ## 2.2. Bioethics Before starting the study, approval was requested from the appropriate department of the Ministry of Education and Religious Affairs (code of approval F$\frac{15}{396}$/72005/C1 by the Institute of Educational Policy) and the study was carried out following the principles of the Declaration of Helsinki. The investigators informed all people who were involved about the aims and procedures of the research. The students participated in the study after the written consent of their parents. ## 2.3. Sample Students and their parents were recruited through school registries. Children’s questionnaires were filled out in school settings, whereas parents’ questionnaires were filled out at home and returned to school. In total, 1728 students (795 males; $46\%$) aged 10–12 years old, enrolled in the study. The participation rate of children ranged from 95 to $100\%$ between schools, without any significant differences between the studied areas. After checking the questionnaires’ completeness for the needs of the present work, the final working sample for the analyses was $$n = 1716$$ children. All children’s parents were also invited to participate, with a $68.9\%$ response rate being achieved ($$n = 1190$$). The working sample was adequate to evaluate the effect size measures’ differences of $20\%$ at <$5\%$ level of significance achieving $85\%$ statistical power. ## 2.4.1. Children’s Characteristics Each child completed a questionnaire specially developed for the study. To avoid errors and discrepancies, the study’s investigators assisted children by giving practical examples when it was necessary. Each child was provided with a personal code by the school principal for the questionnaires to be cross-referenced to those of their parents. The questionnaire retrieved information about socio-demographics (age, sex), anthropometric characteristics as well as dietary habits, and lifestyle parameters. ## 2.4.2. Anthropometric Characteristics Specially trained health scientists/investigators (i.e., dietitians, registered nurses, physicians) took the necessary anthropometric measurements of children (height and body weight in cm and kg, respectively) using a tape measure and a scale (with skin-tight clothing, to minimize measurement errors) and performed a face-to-face interview with them, which lasted a maximum of 10–15 min. Each child’s body weight (kg) was measured to the nearest 100 g using a digital scale (Tanita), and height (cm) was measured to the nearest 0.1 cm using a portable stadiometer (Leicester Height Measure). Children’s BMI was calculated as the ratio of kg/height in m2. Children’s body weight status was evaluated through the age- and the sex-specific International Obesity Task Force (IOTF) Body Mass Index cut-off criteria [25]. ## 2.4.3. Physical Activity Status The standardized, validated, and reliable questionnaire Physical Activity and Lifestyle Questionnaire (PALQ) [26] was used to measure children’s physical activity status. The latter was defined as their participation in out-of-school activities such as sports club participation, playing with others, running, and swimming, on a daily or weekly basis. ## 2.4.4. Stress Children’s stress was assessed through self-report questions, which reflected five sources of stress: Parental expectations (yes/no); Teachers’ expectations (yes/no); School performance (yes/no); Busy schedule of extracurricular activities (yes/no); Pressure from classmates and friends (yes/no). These were based on the Adolescent Stress Questionnaire (ASQ) [27] translated and evaluated in several countries, including Greece [28]. ## 2.4.5. Dietary Habits The level of adherence to the Mediterranean Diet was evaluated using a Mediterranean Diet quality index for children and adolescents (i.e., KIDMED) [29]. Dietary habits with a positive aspect to this dietary pattern scored +1; dietary habits with a negative association scored −1; and, dietary habits with a neutral association scored 0. The theoretical total score ranges from −4 to 12. Lower scores indicated low adherence to the Mediterranean Diet while higher scores indicated high adherence to the Mediterranean Diet. To acquire information on children’s eating habits, we used a validated semi-quantitative food frequency questionnaire (FFQ) [30] which contained all foods and beverages (either sugar-sweetened, or with no added sugar, or sugar-free beverages), including soft drinks, (i.e., carbonated soft drinks, concentrated fruit drinks) and flavored drinks (i.e., chocolate milk) which are commonly consumed by the general child population. Henceforth, for ease of reference, soft drinks and flavored drinks, all of them either sugar-sweetened or sugar-free or with no added sugar, will be referred to as UPB. For the purpose of this study, 330 mL of drink consumption was considered as the standard portion size. More specifically, information regarding UPB consumption was collected, based on a daily (1 time or more than 2 times per day), weekly (i.e., 1, 2–6 times per week), or monthly (i.e., 1–3 times per month, less than 1 time per month) basis. Thus, following the methodological approach of Naomi et al. [ 2022] [31], each drink item was assigned a score of 0–5 depending on the frequency of drink consumption (0 being never/rarely and 5 being 1 or more times per day) so that the UPB consumption measure ranged from 0 to 15 (0 being no UPB consumption). Particularly, UPB consumption was summarized into three categories for the analyses in the present study: “Low” when the frequency consumption was equal or less than 1 time per week; “Moderate” for 2–6 times per week; “High” for at least 1 time per day. The same categorization was also applied to the frequency of chocolate milk consumption. ## 2.5. Parental Characteristics Several parental sociodemographic characteristics (age), anthropometric characteristics [body weight (kg) and height (m)], educational level (i.e., primary, secondary, higher), and financial characteristics (income status under or over 18.000 €/year), as well as lifestyle characteristics [smoking status (yes/no), physical activity status (not at all or at least 1–2 times per week)], were recorded by the children’s parents. ## 2.6. Dietary Characteristics Similar to children’s dietary habits evaluation, parental dietary habits were assessed on a semi-quantitative, validated, and reproducible FFQ. Overall assessment of dietary habits was evaluated through a special diet score (MedDietScore, range 0–55), which assesses adherence to the Mediterranean dietary pattern [32]. Higher values on the score indicate greater adherence to this pattern and, consequently, healthier dietary habits. Parents whose score was ≤25 units were classified as being away from the Mediterranean Diet, while parents whose score was >25 units were classified as being close/very close to the Mediterranean Diet. ## 2.7. Statistical Analysis Continuous characteristics are presented as mean ± standard deviation (SD), and categorical characteristics as relative frequencies (%). The One-way Analysis of Variance (ANOVA) was used in order to investigate the association between the continuous characteristics and the frequency of UPB consumption (Low, Moderate, High) (dependent variable), while Pearson’s Chi-square test was used in the case of the categorical characteristics. Normality of the continuous variables’ distribution was tested through graphical (histograms, PP- plots, QQ- plots) and statistical means (Shapiro–Wilk test). Multivariable linear regression analysis was implemented in order to evaluate the association between preadolescents’ UPB consumption and their BMI (dependent variable). In addition, multivariable logistic regression analysis was also performed in order to investigate the individual effect, as well as the combined effect, of different sources of stress on UPB and chocolate milk consumption. In the case of linear regression analysis, the results are presented as unstandardized beta-coefficients and standard errors, while in the case of logistic regression analysis, the results are presented as Odds Ratios (OR) and $95\%$ Confidence Intervals (CI). Both the beta-coefficients and the ORs, compare all the UPB and chocolate milk consumption categories, with each other. All the results are adjusted for various participants’ characteristics (i.e., children’s age, sex, KIDMED score, physical activity status as well as parents’ age, BMI, smoking status, physical activity status, educational level, income and MedDietScore). All statistical analyses were performed using IBM SPSS Statistics, version 29 (IBM Corp., Armonk, NY, USA) and significance level was set at $a = 0.05.$ ## 3.1. Profile of High UPB Consuming Preadolescents Table 1 presents the characteristics of the preadolescents and their parents, both in the total sample, as well as separately, according to the frequency of UPB consumption. Almost seven out of 10 preadolescents ($67.1\%$) were classified as at least moderate UPB consumers, while $27.6\%$ of the total sample were classified as high UPB consumers. As depicted, boys vs. girls ($$p \leq 0.001$$), as well as preadolescents with higher body mass index ($$p \leq 0.001$$) and unhealthier lifestyle habits, both in terms of lower physical activity levels ($$p \leq 0.002$$), as well as in terms of lower diet quality ($p \leq 0.001$), were more likely to be high UPB consumers. As far as the parents’ characteristics are concerned, high UPB consuming preadolescents were more likely to have parents characterized by lower educational levels ($p \leq 0.05$ both for fathers and mothers) and income ($$p \leq 0.040$$), while at the same time they seemed to have parents with both unhealthier nutritional habits ($p \leq 0.001$), as well as lower physical activity levels ($$p \leq 0.030$$). Finally, regarding the influence of the stress levels on UPB consumption, preadolescents being influenced by more stressors were more likely to be high UPB consumers ($$p \leq 0.015$$), and in particular, a significantly higher percentage of the high UPB consumers were found to be stressed due to parental expectations ($$p \leq 0.023$$), teachers’ expectations ($$p \leq 0.021$$), as well as by the pressure coming from their classmates and friends ($$p \leq 0.038$$). ## 3.2. Impact of High UPB Consumption on Preadolescents’ Body Mass Index After adjusting for several preadolescents and their parents’ characteristics, as presented in Table 2, higher UPB consumption was associated with significantly higher BMI among preadolescents. More specifically, based on the fully adjusted model (Model 7), when compared to low consumers, preadolescents with high UPB consumption were found to have 0.89 kg/m2 higher BMI ($b = 0.89$; se = 0.37; $$p \leq 0.020$$). In addition, high UPB consuming preadolescents were also found to have significantly higher BMI levels, even when compared to moderate UPB consumers ($b = 0.69$; se = 0.32; $$p \leq 0.030$$). Finally, it should be noted that there was no significant difference between low and moderate UPB consuming preadolescents. ## 3.3. Association of UPB Consumption with Stress Levels Moreover, multivariable logistic regression analysis was also performed in order to evaluate the association between the preadolescents’ UPB consumption and their stress levels. As depicted in Table 3, preadolescents who report being affected by a greater number of stressors are significantly more likely to be high UPB consumers than low UPB consumers (per 1 stressor increment: OR = 1.11; $95\%$ CI = 1.01–1.24), while at the same time, they were found to have higher odds of being high UPB consumers than moderate UPB consumers (per 1 stressor increment: OR = 1.09; $95\%$ CI = 0.99–1.20), with the difference being of borderline significance ($p \leq 0.10$). More specifically, preadolescents reporting that they are being stressed by their parents’ expectations, had $32\%$ higher odds of being high UPB consumers than low consumers (OR = 1.32; $95\%$ CI= 1.01–1.73), while those being stressed by their teachers’ expectations also had $33\%$ higher odds of being high UPB consumers than low UPB consumers (OR = 1.33; $95\%$ CI = 1.03–1.73). As for the pressure from their classmates and friends, preadolescents reporting this source of stress, were found to have $43\%$ higher odds of being high UPB consumers than low UPB consumers (OR = 1.43; $95\%$ CI = 1.02–2.02), while they were also found to have approximately two times higher odds of being high UPB consumers than moderate UPB consumers (OR = 1.91; $95\%$ CI = 1.36–2.69). In addition, the combination of the sources of stress seemed to increase the preadolescents’ odds of being high rather than low UPB consumers, with those reporting being stressed both by their parents’ expectations and their classmates/friends, as well as those being stressed by their parents, teachers, and classmates, having approximately two times higher odds of being high rather than low UPB consumers. Furthermore, it was also found that preadolescents being stressed by the pressure from classmates/friends, in combination with the parental and the teacher’s expectations, had at least two times higher odds of being higher rather than moderate UPB consumers. Finally, when compared to preadolescents reporting no source of stress, those reporting being stressed by all factors (parents, teachers, and classmates/friends), had 2.25 times and approximately three times higher odds of being high rather than low UPB consumers (OR = 2.25; $95\%$ CI = 0.97–5.19) and high rather than moderate UPB consumers (OR = 2.84; $95\%$ CI = 1.25–6.46), respectively. Regarding the association between the frequency of chocolate milk consumption and the sources of stress, as presented in Table 4, there was an indication ($p \leq 0.10$) that a higher number of stressors is related to more frequent consumption of chocolate milk, while it should also be noted that preadolescents reporting being stressed by their parents’ expectations, as well as those being stressed by their classmates and friends’ pressure, were found to have approximately two times higher odds of being high rather than low chocolate milk consumers, as well as high rather than moderate chocolate milk consumers (all p-values < 0.05). When combining the effect of the stressors, it is worth noting the fact that, compared to preadolescents being not stressed at all, those being stressed by all factors were found to have at least four times higher odds of being high rather than low chocolate milk consumers ($p \leq 0.05$), as well as high rather than moderate chocolate milk consumers ($p \leq 0.05$). ## 4. Discussion As it was hypothesised, environmental stress affects UPB consumption in preadolescents, influencing the likelihood of being obese. To the best of our knowledge, the role of the family and school environmental stressors on the frequency of soft and flavored drink consumption (either sugary sweetened, or with no added sugar) has been less investigated. The main study findings showed that nearly three out of 10 Greek preadolescents consumed at least one carbonated soft drink, chocolate milk and/or concentrated fruit drink on a daily basis. Boys, those having unhealthier lifestyle habits and higher BMI, as well as preadolescents belonging in families with lower socioeconomic level (in terms of educational and income level) and unhealthier lifestyle, were more likely to consume UPB on a more frequent basis. Moreover, it was revealed that the more frequent the UPB consumption, the higher the BMI for preadolescents, with family, teachers and peers being recognized as the stressors with the greater impact on UPB consumption. In particular, it was found that a higher number of stressors is significantly related to higher odds of being a high UPB consumer. From the public health perspective, this observation is of utmost importance, revealing the role of stress in preadolescents as a potential “triggering factor” of unhealthy dietary habits with its subsequent effects on childhood obesity. In line with other studies, Greek preadolescent boys were found to be more susceptible to soft drinks and other less healthy beverages compared to girls [33,34]. It is supported that young girls tend to have healthier dietary habits than boys [35], and this could be partly attributed to the fact that boys are more exposed to food advertising, and their tastes are more influenced by this exposure, which coincides with male-dominant advertisement content [36]. Moreover, the literature findings suggest that physiological, psychological and sociocultural factors determine the associations between gender and dietary behaviors. In particular, research has shown that males are prone to preferring strong-tasting, sweet foods and are driven by the pleasure of consumption [37]. In line with the literature, the current work revealed that after adjusting for several confounding factors, higher UPB consumption is related to higher preadolescents’ BMI [38,39], while it is also worth noting the fact that the level of preadolescents’ adherence to the Mediterranean Diet did not counterbalance the impact of high UPB consumption on BMI. Despite the fact that the Mediterranean Diet has been shown to have great benefits, both on children’s weight [40,41] and on their future weight as adults [42], it has also been demonstrated that excessive and regular intake of soft drinks might negate this impact, as it has been claimed that only substituting soft drinks with water could make youngsters lose weight [43]. The positive association between the consumption of UPB and weight status has been attributed to the high added sugar content, low satiety, and insufficient compensation for overall energy [44,45]. Recent findings also reveal the potential role of high-fructose corn syrup drinks on metabolic dysregulation [46], and on the alteration of gut microbiota, inducing obesity in animal studies [47]. In accordance with other studies supporting the premise that increased soft drinks, energy drinks, and generally sweet drinks or sweet food intake, are associated with increased stress levels and vice versa [48,49], it was found that the higher the number of stressors, the higher the preadolescents’ likelihood of being high UPB consumers, highlighting the effect of stress on sugar cravings. It is also worth mentioning that the role of chocolate milk has been investigated separately in our attempt to differentiate its consumption from the other undoubtedly unhealthy beverages. Albeit the association of flavored milk consumption with obesity has been proven to be rather protective regarding weight status [24,50], one could speculate that the higher the stressors, the higher the flavored milk consumption, implying the desire of preadolescents for the taste of sugar in stressful situations. Chronic stress has been found to be correlated with a preference for high-sugar foods [51]. Stress from school is often linked to stress from parents and/or teacher’s demands as a whole. The literature suggests that the parent-child relationship, referring to the unique and enduring bond between a caregiver and his or her child, is one of the crucial factors which may influence students’ academic pressure [52]. Moreover, teachers’ behaviors and demands may affect students’ academic achievement and the level of school stress [53]. Preadolescents appraise their family and school environment as important for their well-being and struggle to cope with their excessive academic expectations, pleasing teachers and parents, and keeping up with their peers, which are all interrelated factors [54]. Evidence from the literature demonstrates that peer pressure is an influential force during adolescence, a period of special vulnerability, shaping both adaptive and maladaptive behaviors [55]. Moreover, parental expectations of their children’s academic performance compared to their peers, as well as the emphasis on family sacrifices to support their studies can serve as sources of stress and depression [56]. This is also observed in the present study implying that the demanding expectations of parents on their children’s academic performance are part of the Greek culture. Evidence suggests that the impact of stress on unhealthy eating may begin as early as in preadolescence [57]. A plausible speculation for the observed associations could be that overall, sugar-sweetened drinks increase weight gain by adding liquid calories to the diet, inducing hyperinsulinemia, as well as through a possible dopaminergic reward activation [6]. Additionally, cortisol is produced in response to stress [58] which in turn may induce cravings for sugary, fatty, and salty foods [59]. Stress also induces secretion of both glucocorticoids, and insulin which in turn increases motivation for food, and promotes food intake and obesity, respectively [60]. It is worth mentioning that the emerging concept of “depreobesity”, which incorporates the co-existence of obesity and depression, has been suggested to be the future epidemic [61]. From a public health perspective, numerous public health regulations such as taxation, marketing regulation, nutrition labeling, consumers’ education, and healthier food environments [62,63] have been designed to discourage youth and children from drinking unhealthy beverages. However, it seems that the roles of the family and the school environment have probably been underestimated regarding their influence on UPB consumption through the modulation of induced stress. To the best of our knowledge, this is the first study in Greece, evaluating the effect of preadolescents’ sources of stress on the association between soft and flavored drink consumption with the BMI of Greek preadolescents. The novel findings, the large sample size with the implemented stratified random sampling scheme as well as the use of validated questionnaires, are considered the main strengths of the current work. However, given our study’s cross-sectional design, numerous limitations should be addressed when interpreting the data. No temporal relationship and causal inferences can be established in observational studies, while self-reported questionnaires by schoolchildren may introduce reporting bias. Moreover, residual confounding may also exist. To eliminate this form of bias and maximize the validity of the replies, trained investigators were present during the entire process of completing the questionnaire in schools in order to clarify any potential misunderstandings. ## 5. Conclusions Family, through parents’ expectations, and school stressors, either derived from teacher’s expectations or pressure from peers, seem to encourage UPB consumption which in turn contributes to an increased BMI in Greek preadolescents. It is thus urgent to identify and manage the interplay between stress, depression, unhealthy dietary habits and obesity so as to battle the epidemic of “depreobesity” in preadolescence. Stakeholders should implement programs that raise awareness among parents and teachers about the sources of stress in preadolescence as a potential “triggering factor” of unhealthy dietary preferences. ## References 1. Di Cesare M., Sorić M., Bovet P., Miranda J.J., Bhutta Z., Stevens G.A., Laxmaiah A., Kengne A.-P., Bentham J.. The Epidemiological Burden of Obesity in Childhood: A Worldwide Epidemic Requiring Urgent Action. BMC Med. (2019.0) 17. DOI: 10.1186/s12916-019-1449-8 2. Sahoo K., Sahoo B., Choudhury A.K., Sofi N.Y., Kumar R., Bhadoria A.S.. Childhood Obesity: Causes and Consequences. J. Fam. Med. Prim. Care (2015.0) 4 187-192. DOI: 10.4103/2249-4863.154628 3. Faruque S., Tong J., Lacmanovic V., Agbonghae C., Minaya D.M., Czaja K.. The Dose Makes the Poison: Sugar and Obesity in the United States—A Review. Pol. J. Food Nutr. Sci. (2019.0) 69 219-233. DOI: 10.31883/pjfns/110735 4. Schwartz M.W., Seeley R.J., Zeltser L.M., Drewnowski A., Ravussin E., Redman L.M., Leibel R.L.. Obesity Pathogenesis: An Endocrine Society Scientific Statement. Endocr. Rev. (2017.0) 38 267-296. DOI: 10.1210/er.2017-00111 5. Pietka M.J., Korab H.E.. Harry Edward Soft Drink. Encyclopedia Britannica 6. Malik V.S., Hu F.B.. The role of sugar-sweetened beverages in the global epidemics of obesity and chronic diseases. Nat. Rev. Endocrinol. (2022.0) 18 205-218. DOI: 10.1038/s41574-021-00627-6 7. Auerbach B.J., Wolf F.M., Hikida A., Vallila-Buchman P., Littman A., Thompson D., Louden D., Taber D.R., Krieger J.. Fruit Juice and Change in BMI: A Meta-Analysis. Pediatrics (2017.0) 139 e20162454. DOI: 10.1542/peds.2016-2454 8. Fayet-Moore F.. Effect of Flavored Milk vs Plain Milk on Total Milk Intake and Nutrient Provision in Children. Nutr. Rev. (2016.0) 74 1-17. DOI: 10.1093/nutrit/nuv031 9. Sakaki J.R., Li J., Gao S., Ha K., Tamimi R.M., Chavarro J.E., Chen M.-H., Sun Q., Hart J.E., Chun O.K.. Associations between Fruit Juice and Milk Consumption and Change in BMI in a Large Prospective Cohort of U.S. Adolescents and Preadolescents. Pediatr. Obes. (2021.0) 16 e12781. DOI: 10.1111/ijpo.12781 10. Peckham J.G., Kropp J.D., Mroz T.A., Haley-Zitlin V., Granberg E.. Students Choosing Fat-Free Chocolate Milk during School Lunch Consume More Calories, Total Sugar, Protein, Minerals and Vitamins at Lunch. Public Health Nutr. (2021.0) 24 1818-1827. DOI: 10.1017/S1368980021000161 11. Dereń K., Weghuber D., Caroli M., Koletzko B., Thivel D., Frelut M.L., Socha P., Grossman Z., Hadjipanayis A., Wyszyńska J.. Consumption of Sugar-Sweetened Beverages in Paediatric Age: A Position Paper of the European Academy of Paediatrics and the European Childhood Obesity Group. Ann. Nutr. Metab. (2019.0) 74 296-302. DOI: 10.1159/000499828 12. Bere E., Glomnes E.S., te Velde S.J., Klepp K.-I.. Determinants of Adolescents’ Soft Drink Consumption. Public Health Nutr. (2008.0) 11 49-56. DOI: 10.1017/S1368980007000122 13. van de Gaar V.M., van Grieken A., Jansen W., Raat H.. Children’s Sugar-Sweetened Beverages Consumption: Associations with Family and Home-Related Factors, Differences within Ethnic Groups Explored. BMC Public Health (2017.0) 17. DOI: 10.1186/s12889-017-4095-0 14. Miller A.L., Lumeng J.C.. Pathways of Association from Stress to Obesity in Early Childhood. Obesity (2018.0) 26 1117-1124. DOI: 10.1002/oby.22155 15. Sisk L.M., Gee D.G.. Stress and Adolescence: Vulnerability and Opportunity during a Sensitive Window of Development. Curr. Opin. Psychol. (2022.0) 44 286-292. DOI: 10.1016/j.copsyc.2021.10.005 16. Miller J.G., Chahal R., Gotlib I.H.. Early Life Stress and Neurodevelopment in Adolescence: Implications for Risk and Adaptation. Curr. Top. Behav. Neurosci. (2022.0) 54 313-339. DOI: 10.1007/7854_2022_302 17. Lewis C.E., Siegel J.M., Lewis M.A.. Feeling bad: Exploring sources of distress among pre-adolescent children. Am. J. Public Health (1984.0) 74 117-122. DOI: 10.2105/AJPH.74.2.117 18. Anniko M.K., Boersma Κ., Tillfors Μ.. Sources of stress and worry in the development of stress-related mental health problems: A longitudinal investigation from early- to mid-adolescence. Anxiety Stress Coping (2019.0) 32 155-167. DOI: 10.1080/10615806.2018.1549657 19. Kaplan D.S., Liu R.X., Kaplan H.B.. School related stress in early adolescence and academic performance three years later: The conditional influence of self expectations. Soc. Psychol. Educ. (2005.0) 8 3-17. DOI: 10.1007/s11218-004-3129-5 20. Zhang X., Huang X., Xiao Y., Jing D., Huang Y., Chen L., Luo D., Chen X., Shen M.. Daily Intake of Soft Drinks Is Associated with Symptoms of Anxiety and Depression in Chinese Adolescents. Public Health Nutr. (2019.0) 22 2553-2560. DOI: 10.1017/S1368980019001009 21. Kadel P., Schneider S., Mata J.. Soft Drink Consumption and Mental Health Problems: Longitudinal Relations in Children and Adolescents. Soc. Sci. Med. (2020.0) 258 113123. DOI: 10.1016/j.socscimed.2020.113123 22. Desmet P.M.A., Schifferstein H.N.J.. Sources of Positive and Negative Emotions in Food Experience. Appetite (2008.0) 50 290-301. DOI: 10.1016/j.appet.2007.08.003 23. Macht M., Haupt C., Ellgring H.. The Perceived Function of Eating Is Changed during Examination Stress: A Field Study. Eat Behav. (2005.0) 6 109-112. DOI: 10.1016/j.eatbeh.2004.09.001 24. Kanellopoulou A., Notara V., Antonogeorgos G., Chrissini M., Rojas-Gil A.P., Kornilaki E.N., Lagiou A., Panagiotakos D.B.. Inverse Association Between Health Literacy and Obesity Among Children in Greece: A School-Based, Cross-Sectional Epidemiological Study. Health Educ. Behav. (2022.0) 49 54-65. DOI: 10.1177/1090198120982944 25. Cole T.J.. Establishing a Standard Definition for Child Overweight and Obesity Worldwide: International Survey. BMJ (2000.0) 320 1240. DOI: 10.1136/bmj.320.7244.1240 26. Argiropoulou E.C., Michalopoulou M., Aggeloussis N., Avgerinos A.. Validity and Reliability of Physical Activity Measures in Greek High School Age Children. J. Sports Sci. Med. (2004.0) 3 147-159. PMID: 24482592 27. Byrne D.G., Davenport S.C., Mazanov J.. Profiles of Adolescent Stress: The Development of the Adolescent Stress Questionnaire (ASQ). J. Adolesc. (2007.0) 30 393-416. DOI: 10.1016/j.adolescence.2006.04.004 28. Darviri C., Legaki P., Chatzioannidou P., Gnardellis C., Kraniotou C., Tigani X., Alexopoulos E.C.. Adolescent Stress Questionnaire: Reliability and Validity of the Greek Version and Its Description in a Sample of High School (Lyceum) Students. J. Adolesc. (2014.0) 37 1373-1377. DOI: 10.1016/j.adolescence.2014.10.003 29. Serra-Majem L., Ribas L., Ngo J., Ortega R.M., García A., Pérez-Rodrigo C., Aranceta J.. Food, Youth and the Mediterranean Diet in Spain. Development of KIDMED, Mediterranean Diet Quality Index in Children and Adolescents. Public Health Nutr. (2004.0) 7 931-935. DOI: 10.1079/PHN2004556 30. Antonogeorgos G., Grigoropoulou D., Papadimitriou A., Priftis K.N., Anthracopoulos M., Nikolaidou P., Panagiotakos D.B.. Validation of a Food Frequency Questionnaire Designed for Children 10-12 Years: The Panacea-FFQ. J. Perioper Nurs. (2013.0) 2 40-54 31. Naomi N.D., Brouwer-Brolsma E.M., Buso M.E.C., Soedamah-Muthu S.S., Harrold J.A., Halford J.C.G., Raben A., Geleijnse J.M., Feskens E.J.M.. Association of Sweetened Beverages Consumption with All-Cause Mortality Risk among Dutch Adults: The Lifelines Cohort Study (the SWEET Project). Eur. J. Nutr. (2022.0) 62 797-806. DOI: 10.1007/s00394-022-03023-6 32. Panagiotakos D.B., Pitsavos C., Stefanadis C.. Dietary Patterns: A Mediterranean Diet Score and Its Relation to Clinical and Biological Markers of Cardiovascular Disease Risk. Nutr. Metab. Cardiovasc. Dis. (2006.0) 16 559-568. DOI: 10.1016/j.numecd.2005.08.006 33. Ogden C.L., Kit B.K., Carroll M.D., Park S.. Consumption of Sugar Drinks in the United States, 2005–2008. NCHS Data Brief (2011.0) 71 1-8 34. West D.S., Bursac Z., Quimby D., Prewitt T.E., Spatz T., Nash C., Mays G., Eddings K.. Self-Reported Sugar-Sweetened Beverage Intake among College Students. Obesity (2006.0) 14 1825-1831. DOI: 10.1038/oby.2006.210 35. Askovic B., Kirchengast S.. Gender Differences in Nutritional Behavior and Weight Status during Early and Late Adolescence. Anthropol. Anz. (2012.0) 69 289-304. DOI: 10.1127/0003-5548/2012/0212 36. Castronuovo L., Guarnieri L., Tiscornia M.V., Allemandi L.. Food Marketing and Gender among Children and Adolescents: A Scoping Review. Nutr. J. (2021.0) 20 52. DOI: 10.1186/s12937-021-00706-4 37. Grzymisławska M., Puch E.A., Zawada A., Grzymisławski M.. Do nutritional behaviors depend on biological sex and cultural gender?. Adv. Clin. Exp. Med. (2020.0) 29 165-172. DOI: 10.17219/acem/111817 38. Luger M., Lafontan M., Bes-Rastrollo M., Winzer E., Yumuk V., Farpour-Lambert N.. Sugar-Sweetened Beverages and Weight Gain in Children and Adults: A Systematic Review from 2013 to 2015 and a Comparison with Previous Studies. Obes. Facts (2017.0) 10 674-693. DOI: 10.1159/000484566 39. Marshall T.A., Curtis A.M., Cavanaugh J.E., Warren J.J., Levy S.M.. Child and Adolescent Sugar-Sweetened Beverage Intakes Are Longitudinally Associated with Higher Body Mass Index z Scores in a Birth Cohort Followed 17 Years. J. Acad. Nutr. Diet. (2019.0) 119 425-434. DOI: 10.1016/j.jand.2018.11.003 40. Kosti R.I., Kanellopoulou A., Fragkedaki E., Notara V., Giannakopoulou S.-P., Antonogeorgos G., Rojas-Gil A.P., Kornilaki E.N., Lagiou A., Panagiotakos D.B.. The Influence of Adherence to the Mediterranean Diet among Children and Their Parents in Relation to Childhood Overweight/Obesity: A Cross-Sectional Study in Greece. Child. Obes. (2020.0) 16 571-578. DOI: 10.1089/chi.2020.0228 41. Lassale C., Fitó M., Morales-Suárez-Varela M., Moya A., Gómez S.F., Schröder H.. Mediterranean Diet and Adiposity in Children and Adolescents: A Systematic Review. Obes. Rev. (2022.0) 23 e13381. DOI: 10.1111/obr.13381 42. D’Innocenzo S., Biagi C., Lanari M.. Obesity and the Mediterranean Diet: A Review of Evidence of the Role and Sustainability of the Mediterranean Diet. Nutrients (2019.0) 11. DOI: 10.3390/nu11061306 43. Wang Y.C., Ludwig D.S., Sonneville K., Gortmaker S.L.. Impact of Change in Sweetened Caloric Beverage Consumption on Energy Intake among Children and Adolescents. Arch. Pediatr. Adolesc. Med. (2009.0) 163 336-343. DOI: 10.1001/archpediatrics.2009.23 44. Libuda L., Kersting M.. Soft Drinks and Body Weight Development in Childhood: Is There a Relationship?. Curr. Opin. Clin. Nutr. Metab. Care (2009.0) 12 596-600. DOI: 10.1097/MCO.0b013e32833189f6 45. Malik V.S., Schulze M.B., Hu F.B.. Intake of Sugar-Sweetened Beverages and Weight Gain: A Systematic Review. Am. J. Clin. Nutr. (2006.0) 84 274-288. DOI: 10.1093/ajcn/84.2.274 46. Hattori H., Hanai Y., Oshima Y., Kataoka H., Eto N.. Excessive Intake of High-Fructose Corn Syrup Drinks Induces Impaired Glucose Tolerance. Biomedicines (2021.0) 9. DOI: 10.3390/biomedicines9050541 47. Wang X., Zhu L., Li X., Wang X., Hao R., Li J.. Effects of high fructose corn syrup on intestinal microbiota structure and obesity in mice. npj Sci. Food (2022.0) 6 17. DOI: 10.1038/s41538-022-00133-7 48. Errisuriz V.L., Pasch K.E., Perry C.L.. Perceived Stress and Dietary Choices: The Moderating Role of Stress Management. Eat Behav. (2016.0) 22 211-216. DOI: 10.1016/j.eatbeh.2016.06.008 49. Ling J., Zahry N.R.. Relationships among Perceived Stress, Emotional Eating, and Dietary Intake in College Students: Eating Self-Regulation as a Mediator. Appetite (2021.0) 163 105215. DOI: 10.1016/j.appet.2021.105215 50. Patel A.I., Moghadam S.D., Freedman M., Hazari A., Fang M.-L., Allen I.E.. The Association of Flavored Milk Consumption with Milk and Energy Intake, and Obesity: A Systematic Review. Prev. Med. (2018.0) 111 151-162. DOI: 10.1016/j.ypmed.2018.02.031 51. Jacques A., Chaaya N., Beecher K., Ali S.A., Belmer A., Bartlett S.. The Impact of Sugar Consumption on Stress Driven, Emotional and Addictive Behaviors. Neurosci. Biobehav. Rev. (2019.0) 103 178-199. DOI: 10.1016/j.neubiorev.2019.05.021 52. Liao M., Zhou S., Ying R.. The Influence of Parent-Child Relationship on the Academic Pressure of Elementary Students: A Moderated Mediation Model-Based on the Survey and Analysis of 38,069 Elementary Students. SSRN Electron. J. (2021.0) 9 1161-1182 53. Piekarska A.. School Stress, Teachers’ Abusive Behaviors, and Children’s Coping Strategies. Child Abus. Negl. (2000.0) 24 1443-1449. DOI: 10.1016/S0145-2134(00)00201-5 54. Deng Y., Cherian J., Khan N.U.N., Kumari K., Sial M.S., Comite U., Gavurova B., Popp J.. Family and Academic Stress and Their Impact on Students’ Depression Level and Academic Performance. Front. Psychiatry (2022.0) 13 869337. DOI: 10.3389/fpsyt.2022.869337 55. Laursen B., Veenstra R.. Toward Understanding the Functions of Peer Influence: A Summary and Synthesis of Recent Empirical Research. J. Res. Adolesc. (2021.0) 31 889-907. DOI: 10.1111/jora.12606 56. Dundes L., Cho E., Kwak S.. The duty to succeed: Honor versus happiness in college and career choices of East Asian students in the United States. Pastor. Care Educ. (2009.0) 27 135-156. DOI: 10.1080/02643940902898960 57. Hill D.C., Moss R.H., Sykes-Muskett B., Conner M., O’Connor D.B.. Stress and Eating Behaviors in Children and Adolescents: Systematic Review and Meta-Analysis. Appetite (2018.0) 123 14-22. DOI: 10.1016/j.appet.2017.11.109 58. Russell G., Lightman S.. The Human Stress Response. Nat. Rev. Endocrinol. (2019.0) 15 525-534. DOI: 10.1038/s41574-019-0228-0 59. Geiker N.R.W., Astrup A., Hjorth M.F., Sjödin A., Pijls L., Markus C.R.. Does Stress Influence Sleep Patterns, Food Intake, Weight Gain, Abdominal Obesity and Weight Loss Interventions and Vice Versa?. Obes. Rev. (2018.0) 19 81-97. DOI: 10.1111/obr.12603 60. Dallman M.F.. Stress-induced obesity and the emotional nervous system. Trends Endocrinol. Metab. (2010.0) 21 159-165. DOI: 10.1016/j.tem.2009.10.004 61. Mediouni M., Kaczor-Urbanowicz K.E., Madiouni R.. Future epidemic: Depreobesity. Obes. Med. (2020.0) 19 100240. DOI: 10.1016/j.obmed.2020.100240 62. Costa D., Warkentin S., Oliveira A.. Sugar-Sweetened Beverages, Effects on Appetite and Public Health Strategies to Reduce the Consumption among Children: A Review. Porto Biomed. J. (2022.0) 7 e172. DOI: 10.1097/j.pbj.0000000000000172 63. Muth N.D., Dietz W.H., Magge S.N., Johnson R.K.. American academy of pediatrics; section on obesity; committee on nutrition; american heart association public policies to reduce sugary drink consumption in children and adolescents. Pediatrics (2019.0) 143 e20190282. DOI: 10.1542/peds.2019-0282
--- title: 'Contributors to Preterm Birth: Data from a Single Polish Perinatal Center' authors: - Iwona Jańczewska - Monika Cichoń-Kotek - Małgorzata Glińska - Katarzyna Deptulska-Hurko - Krzysztof Basiński - Mateusz Woźniak - Marek Wiergowski - Marek Biziuk - Anna Szablewska - Mikołaj Cichoń - Jolanta Wierzba journal: Children year: 2023 pmcid: PMC10047259 doi: 10.3390/children10030447 license: CC BY 4.0 --- # Contributors to Preterm Birth: Data from a Single Polish Perinatal Center ## Abstract Preterm birth may result from overlapping causes including maternal age, health, previous obstetric history and a variety of social factors. We aimed to identify factors contributing to preterm birth in respect to new social and environmental changes in the reproductive patterns. Our cross-sectional study included 495 mother–infant pairs and was based on maternal self-reporting in an originally developed questionnaire. Neonates were divided into two groups: 72 premature babies (study group) and 423 full-term babies (control group). We analyzed maternal, sociodemographic and economic characteristics, habits, chronic diseases, previous obstetric history and pregnancy complications. For statistical analysis, Pearson’s Chi-squared independence test was used with a statistical significance level of 0.05. Preterm births were more common among mothers living in villages ($p \leq 0.001$) and with lower education level ($$p \leq 0.01$$). Premature births were also positively associated with mothers who were running their own businesses ($$p \leq 0.031$$). Mothers with a history of previous miscarriages gave birth at a significantly older age ($p \leq 0.001$). The most frequent pregnancy complications were hypothyroidism ($41.4\%$), pregestational and gestational diabetes mellitus (DM; $17.8\%$) and hypertension ($8.1\%$). Pregestational DM significantly influenced the occurrence of prematurity ($p \leq 0.05$). Pregestational DM, being professionally active, a lower education level and living outside cities are important risk factors of prematurity. ## 1. Introduction Preterm labor is defined as a birth before 37 completed weeks of gestation (less than 259 days) [1]. Preterm birth is currently one of the most important problems in maternal-child health worldwide and is especially severe in low-income countries. In Europe in 2014, preterm births constituted $8.7\%$ of all births [2]. Countries with the lowest preterm birth rates are Finland and Lithuania (ranging from 5.5–$5.9\%$), whereas the highest rate of preterm births is observed in the Czech Republic ($8.1\%$), Cyprus ($10.4\%$) and Austria ($11\%$). In Poland, the preterm birth rate has increased from $6.0\%$ in 2000 up to $7.0\%$ in 2019 [2,3,4,5]. Despite the introduction of many public health projects and medical interventions to prevent preterm births, even in high-income countries preterm birth is the second most common cause of death after congenital anomalies in neonates and children under five, and it has lifelong effects on neurodevelopmental outcomes in affected individuals. Preterm babies are at increased risk of neurological impairment and disability and chronic disease in adulthood [6]. Multiple factors are associated with preterm birth, including maternal age, race, previous obstetric history, infections, and exposure to cigarettes and drugs. Comorbidities such as obesity, hypertension, hypothyroidism, pregestational and gestational diabetes mellitus (DM) in mothers may contribute to preterm birth [7,8,9,10]. Psychosocial factors that may contribute to preterm birth include marital status, professional activity, living conditions and the level of social security [11]. Genetic, environmental, social and economic factors overlap, and all have remarkable impacts on the health of pregnant women (and the rate of preterm births), especially in developing countries. Notably, in the developed countries the data is not sufficient. Despite the ongoing enhancement of perinatal care in Poland, in recent years we have observed a slightly increasing trend of preterm births. Social norms and significant changes within it have impacts on the lifestyles of individuals. Moreover, numerous women live in informal relationships and/or raise children alone [10,12,13,14]. Bearing in mind that many factors (sociodemographic, economic, lifestyle, work status) can contribute towards the increasing numbers of preterm births in the developed countries, an accurate investigation is needed. Such assessment could be a starting point to improve the health care of pregnant women by implementing new health programs aiming to provide even better health care and education about the aforementioned risk factors of preterm birth, hence decreasing its incidence. Perhaps, some of the risk factors are modifiable and could be easily avoided or changed. The aim of this preliminary study is to evaluate whether sociodemographic, economic and lifestyle factors as well as work status or chronic diseases contribute to the preterm birth in a cohort of women from the Pomeranian region. ## 2. Materials and Methods This preliminary study was conducted among mothers who gave birth between February 2019 and February 2020 in the University Clinical Center (UCC) in Gdańsk. Our center is a tertiary perinatal care facility associated with the Medical University of Gdansk (MUG), located in Gdansk, Pomeranian region, north Poland. The Department of Neonatology provides a full range of neonatal care services: intensive, intermediate, continuing care, and well-baby nursery for patients from the entire region. The Obstetrics Department provides the highest level of care for complicated pregnancies and deliveries. During the study, 2800 births took place. This Department provides diagnostic approaches to and treatment of multiple pregnancies, diabetes in pregnancy, birth defects, preterm deliveries, and other adverse pregnancy complications. It also supports physiological deliveries. Maternal and neonatal patients from lower-level hospitals in the region are referred for diagnostics and treatment to the UCC. The study was based on maternal self-reporting through an original questionnaire in which women described their state of health and any disorders occurring before and during pregnancy, as well as any medications that were prescribed during pregnancy. We analyzed maternal sociodemographic and economic characteristics, maternal habits, previous obstetric history, maternal chronic disease, and pregnancy complications. The survey was anonymous and in the Polish language. It consisted of 33 closed single- and multiple-choice questions. All questionnaires were gathered in written, “paper form”. Only completely and correctly filled questionaries were enrolled in the statistical analysis. Approximately $10\%$ of them were incomplete and, therefore, were not included in the study. Five percent of patients refused to take part in the survey, mostly due to unsatisfying health conditions postpartum. In total, we enrolled 495 mother–infant dyads. Neonates were divided into two groups: 72 newborn infants of premature birth (<37 weeks; case group) and 423 newborn infants of term delivery (≥37 weeks; control group). All maternal participants provided their written informed consent. The inclusion criteria were singleton pregnancies, enrolling Polish-speaking women who have just given birth and obtaining a written consent to take part in the study. The exclusion criteria were multiple pregnancies (since they statistically more often lead to preterm birth), inappropriately fulfilled questionnaires and refusal to take part in the study. This is a preliminary study, which we plan to extend in the future to analyze the spontaneous and iatrogenic causes of preterm birth in a larger group of patients. ## Statistical Analysis Statistical significance of differences between categorical variables was calculated with Pearson’s Chi-squared independence test. For age and gestational age, means (M) were compared with Welch’s t-test, correcting for unequal variances. All analyses were conducted in Python (version 3.8) using packages Pandas (version 1.4) and Pingouin (version 0.5). Statistical significance level was 0.05. ## 3. Results Fifty percent ($50\%$) of women in our study had a child between the ages of 28 and 35, ($M = 31.25$ years, SD = 4.93); the youngest woman was 14, and the oldest was 45 years old. The mean age of primiparous was $M = 29.4$ (SD = 4.7), while the mean age of women who had given birth previously was $M = 33$ (SD = 4.5). Of 495 babies, 72 were born prematurely, which is $14.5\%$. There was no statistically significant difference in the age of mothers giving birth to their babies at term and prematurely ($p \leq 0.05$). Nearly half the mothers ($47\%$) were residents of towns with more than 400,000 inhabitants, $31\%$ were residents of smaller towns, and $22\%$ lived in rural areas. There were statistically significant differences in the proportion of premature to full-term neonates depending on the place of residence ($p \leq 0.001$). The percentage of mothers residing in villages was higher in the group of premature neonates in comparison to the full-term neonates group. Women with a university education made up $70.5\%$ [349] of the women participating in the study and $83\%$ [409] worked during pregnancy. There were statistically significant differences depending on the mother’s level of education. The rate of prematurity was higher among mothers with lower education levels than among mothers with higher education ($$p \leq 0.01$$). We also found the relationship between prematurity and professional activity of women. Proportionally women running their own business were more frequent in the group of preterm birth than in the group of women giving birth at term ($$p \leq 0.031$$). Married and cohabiting mothers constituted $74.3\%$ [368] and $15\%$ [74] of the study group, respectively. There was a correlation between marital status and gestational age at birth: in the group of full-term infants there were more cohabiting than married mothers, while in the premature group married women were more frequent. Most of our respondents declared their socioeconomic status to be either good or very good. This factor did not affect the week of delivery ($p \leq 0.05$). There were no statistically significant effects of smoking during pregnancy on preterm labor ($p \leq 0.05$). The general characteristics of the study group are given in Table 1. A previous history of miscarriage was reported in $21.2\%$ [105]. Women who had previously had a miscarriage had a significantly higher maternal age ($M = 32.9$, SD = 4.93) in comparison to women who had not had a miscarriage ($M = 30.8$, SD = 4.84; t = −3.90, $p \leq 0.001$). The most frequent pregnancy complications were hypothyroidism in $41.4\%$ of our cohort, gestational diabetes (GDM) in $15.8\%$ and hypertension in $8.1\%$. There were no statistically significant differences in hypertension developing before and during pregnancy, hypothyroidism, GDM, chronic kidney disease, asthma, allergies, depression, anaemia, or cardiovascular diseases (all p’s > 0.05). Before pregnancy, 10 women ($2.0\%$) suffered from diabetes mellitus (DM) and this group gave birth to premature babies significantly more often than women without DM ($p \leq 0.05$). Of the entire group, $61.4\%$ [304] required some form of prescription medication, the most often being levothyroxine. Women receiving oral or intravaginal progesterone during pregnancy accounted for $14.5\%$ (45 cases). Of these, 13 had a history of spontaneous abortion, while 32 did not. In three cases there had been spontaneous preterm deliveries in prior pregnancies. In the group of women taking progesterone, the mean gestational age (GA) at birth was significantly lower, on average 37.0 weeks (SD = 2.69), compared to women not taking progesterone ($M = 38.7$ weeks, SD = 2.2); $t = 3.65$, $p \leq 001.$ Adverse pregnancy outcomes are given in Table 2. ## 4. Discussion The overall rates of preterm delivery in the recent decades in European countries such as Cyprus, Greece, Germany, and Poland have increased, while in others such as Estonia, Croatia, and Netherlands the overall rates of prematurity have either stabilized or decreased [7,15]. The regionalization of obstetric and neonatal care and transfer of pregnant women at risk of preterm labor to the tertiary care center leads to decreased mortality and morbidity rates among preterm infants. However, our study confirmed that this practice also results in a much higher frequency of preterm deliveries in these settings compared to the lower level of perinatal care hospitals. Our study population was drawn from the third level of reference care, where more complicated pregnancies and deliveries are carried out. Numerous recently published studies have shown there are new social and environmental risk factors which contribute to an increased risk of preterm birth for certain women. The determinants of change in reproductive patterns can be explained by the cultural, social, and economic changes occurring in societies of developed countries. In European countries, the decision whether to have children is not without controversy. Many women want to balance motherhood with completing their education, getting a job and maintaining a liberated lifestyle [16,17]. This trend may lead to delayed childbearing and lower fertility rates [16,17,18,19]. According to EUROPERISTAT reports, in the European Union (EU) in 2016 the average age of the mother during her first childbirth was 29.0 years, while in 2019, it was 29.4 years. A similar trend was observed in Poland; although the mean age at which women had their first child was still below 30 years, there was a slight increase from 27.2 to 27.6 between 2016 and 2019 [2,20,21]. Data from our study corresponds with these European statistics. The higher mean age of first delivery among our mothers compared to the overall mean age of primipara women in Poland may be related to the fact that nearly half lived in a large city and had higher education levels. Over past decades, many highly educated and qualified women have entered the workforce and have started their families while they were working [15,16,17,18]. Many women, rather than entering the formal workforce, create a new business in association with their family environment. These women, referred to as “mumpreneurs”, wish to find a work–life balance as a business owner [22]. However, running their own business may be challenging and stressful. Pregnant women may experience fear of failure, job-related anxiety, and discrimination related to their pregnancy [23,24,25]. Stress during pregnancy may have far-reaching implications, including lower GA at birth in women affected. Moreover, business owners tend to work longer hours, which may also contribute to preterm birth [16,24,26,27]. We found a relationship between prematurity and the professional activity of women. We showed that being a business owner positively correlated with prematurity. An increased risk of preterm birth associated with both cohabitation and single motherhood among women in EU has been reported [9]. However, when extramarital births became more common in a community, marital status ceased to be a risk factor for prematurity [28,29]. Moreover, some authors have reported that only single mothers living alone have a higher prevalence of preterm births [30,31]. We did not find extramarital relationship to be a risk factor for preterm birth in our cohort. A lower level of education and living in villages have been shown to be an important determinant of preterm births [10,20,21]. Although the health insurance system in Poland is based on principles of equal treatment and access to healthcare services and pregnant women are eligible for free healthcare benefits, women living in rural areas may have less access to health care as compared with women from urban areas, which can increase the incidence of preterm birth among rural women [10,25]. It seems that education does not constitute a direct, independent risk factor of prematurity, but may be indirectly linked to lower levels of health awareness and thus risk-related behavior and lifestyle choices. Smokers were found to be more likely to give birth to preterm babies in several studies [7,31,32,33,34,35]. We also indicated a relationship between low education levels and prematurity, but we did not find any link between prematurity and smoking. Nevertheless, the promotion of smoking abstinence at childbearing age remains important for the health of mothers and their children, as maternal smoking is a well-known risk factor for perinatal complication. Postponing the age of motherhood is also related to pregnancy complications, reproductive failures such as miscarriages, and preterm birth [18,36,37]. Medical motives were reported as frequent reasons for delayed motherhood [38], and our study confirmed that mothers with a history of previous miscarriages gave birth at an older age. The prevalence of chronic disease in childbearing women has increased dramatically during the past decades from less than $5\%$ in the late 20th century to almost $16\%$ in the early 21st century [39]. Among the disorders, both GDM and DM have become increasingly common and women with these disorders have an increased risk of a range of complications of pregnancy [40,41,42]. The adverse pregnancy outcomes associated with diabetes include preterm birth. Results from our study confirmed previous reports. DM and GDM were the second highest group of chronic diseases among women of our cohort, but only pregestational DM significantly influenced the occurrence of preterm births. Prior studies have also shown that preterm deliveries remain high in women with DM [41,42,43,44]. The limited literature data did not indicate an association between premature delivery and GDM [45]. Our results revealed that there was no link. X It has been shown that satisfactory maternal glycemic control, particularly in the periconceptional period and in the first trimester of pregnancy, is associated with reduced preterm delivery and neonatal morbidity [44,46]. Therefore, effective education of childbearing women suffering from pregestational DM who are planning to become pregnant appears to be crucial to achieve better obstetric and neonatal outcomes. Other chronic maternal conditions such as hypothyroidism and hypertensive disorders associated with preterm birth were also described as important contributors of prematurity [47,48]. According to published data, overt hypothyroidism occurs in 0.3–$0.5\%$ of pregnancies, and subclinical hypothyroidism (SCH) in 2–$3\%$. Hypothyroidism during pregnancy increases the risk of spontaneous abortion, anemia, pregnancy-induced hypertension (PIH), and placental abruption. It is also associated with neonatal complications such as prematurity, LBW, and reduced intelligence in the offspring of these women. Even women with SCH are more likely to have preterm deliveries. Numerous professional associations advise evaluation of thyroid function at the first prenatal clinical appointment to avoid preterm births and other obstetric complications. Polish guidelines suggest testing of pregnant women to identify thyroid dysfunction and recommend the treatment of thyroid disorders, including SCH, during pregnancy [48,49,50,51]. In our cohort, hypothyroidism was the most common disorder and, consequently, levothyroxine was the most frequently used medicine. We found that neither hypothyroidism nor Levothyroxine intake were associated with an increased incidence of preterm labor. This observation leads us to presume that euthyroxinemia in early pregnancy is crucial for maintaining normal placental development and function and to avoid preterm deliveries. The results of our own research suggest that in the cases in which thyroid diseases were diagnosed early and properly treated, it was possible to indirectly prolong the duration of pregnancy. In our study, hypertension was a common complication of pregnancy. While this is often mentioned as the cause of prematurity [47], we did not find a similar correlation in our cohort. Methyldopa remains the first-line treatment for hypertension in pregnancy. Significant efforts have been made to avoid preterm births. This includes the prophylactic administration of progesterone in pregnant women with a history of at least one spontaneous preterm delivery, and in pregnant women without this history with a short cervix in ultrasonographic measurement of cervical length in midgestation [52,53,54]. We found that women taking progesterone were more likely to give birth slightly earlier in comparison to mothers without progesterone. It is probable that this therapeutic intervention in women at risk of preterm labor improved obstetric outcomes and the mean duration of pregnancy was extended near term. Otherwise, the consequent mortality and handicap of infants born too soon could be more serious. ## Limitations of the Study Limitations of the study are not considering the type of delivery (vaginal birth versus cesarean section) as well as and not considering labor from Assisted Reproductive Technology (ART), such as in vitro fertilization (IVF). The goal of the survey was to collect data about the women’s lifestyle and environmental, social and economic conditions. Therefore, apart from certain concomitant diseases (hypertension, diabetes mellitus or hypothyroidism), our study focuses deeply on determining the influence of health behavior of pregnant women on the risk of preterm birth. Since it is a preliminary study, we did not include information on either ART treatments or type of delivery, as these questions were not included in the survey. For the same reason we did not ask whether the pregnancy ended with preterm birth due to iatrogenic reasons (i.e., elective cesarean section). Nonetheless, due to the increasing number of preterm births in developed countries (including Poland), it is crucial to investigate the iatrogenic reasons responsible for them; this we aim to accomplish in the next study. ## 5. Conclusions Pregnant women in the workforce are exposed to higher levels of stress, which contributes to preterm births. Having a job, especially running your own business, positively correlated with prematurity, while advanced maternal age positively correlated with miscarriages. Pregestational DM is an important risk factor for preterm delivery. Improving the health condition of diabetic mothers and minimizing the adverse outcomes of diabetes on pregnancy should be prioritized. ## References 1. WHO: Recommended Definitions, Terminology and Format for Statistical Tables Related to the Perinatal Period and Use of A New Certificate for Cause of Perinatal Deaths. Acta Obstet. Gynecol. Scand. (1977.0) 56 247-253. DOI: 10.3109/00016347709162009 2. Zeitlin J., Szamotulska K., Drewniak N., Mohangoo A., Chalmers J., Sakkeus L., Irgens L., Gatt M., Gissler M., Blondel B.. Preterm birth time trends in Europe: A study of 19 countries. BJOG (2013.0) 120 1356-1365. DOI: 10.1111/1471-0528.12281 3. Blencowe H., Cousens S., Oestergaard M.Z., Chou D., Moller A.-B., Narwal R., Adler A., Vera Garcia C., Rohde S., Say L.. National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since 1990 for selected countries: A systematic analysis and implications. Lancet (2012.0) 379 2162-2172. DOI: 10.1016/S0140-6736(12)60820-4 4. Simmons L.E., Darmstadt G.L., Gravett M.G.. Preventing Preterm Birth and Neonatal Mortality: Exploring the Epidemiology, Causes, and Interventions. Semin. Perinatol. (2010.0) 34 408-415. DOI: 10.1053/j.semperi.2010.09.005 5. Gravett M.G., Rubens C.E., Nunes T.M.. Global report on preterm birth and stillbirth (2 of 7): Discovery science. BMC Pregnancy Childbirth (2010.0) 10. DOI: 10.1186/1471-2393-10-S1-S2 6. Liu L., Oza S., Hogan D., Chu Y., Perin J., Zhu J., Lawn J.E., Cousens S., Mathers C., Black R.E.. Global, regional, and national causes of under-5 mortality in 2000–15: An updated systematic analysis with implications for the Sustainable Development Goals. Lancet (2016.0) 388 3027-3035. DOI: 10.1016/S0140-6736(16)31593-8 7. Chang H.H., Larson J., Blencowe H., Spong C.Y., Howson C.P., Cairns-Smith S., Lackritz E.M., Lee S.K., Mason E., Serazin A.C.. Preventing preterm births: Analysis of trends and potential reductions with interventions in 39 countries with very high human development index. Lancet (2013.0) 381 223-234. DOI: 10.1016/S0140-6736(12)61856-X 8. Sibai B.M., Caritis S.N., Hauth J.C., MacPherson C., VanDorsten J.P., Klebanoff M., Landon M., Paul P.H., Meis P.J., Miodovnik M.. Preterm delivery in women with pregestational diabetes mellitus or chronic hypertension relative to women with uncomplicated pregnancies. Am. J. Obstet. Gynecol. (2000.0) 183 1520-1524. DOI: 10.1067/mob.2000.107621 9. Murphy D.J.. Epidemiology and environmental factors in preterm labour. Best Pract. Res. Clin. Obstet. Gynaecol. (2007.0) 21 773-789. DOI: 10.1016/j.bpobgyn.2007.03.001 10. Hidalgo-Lopezosa P., Jiménez-Ruz A., Carmona-Torres J.M., Hidalgo-Maestre M., Rodríguez-Borrego M.A., López-Soto P.J.. Sociodemographic factors associated with preterm birth and low birth weight: A cross-sectional study. Women Birth (2019.0) 32 538-543. DOI: 10.1016/j.wombi.2019.03.014 11. Goldenberg R.L., Culhane J.F., Iams J.D., Romero R.. Epidemiology and causes of preterm birth. Lancet (2008.0) 371 75-84. DOI: 10.1016/S0140-6736(08)60074-4 12. Euro-Peristat Project, “European Perinatal Health Report. Health and Care of Pregnant Women and Babies in Europe in 2010”. (2013.0) 252 13. Kumar S., Sharma S., Thaker R.. Occupational, Environmental, and Lifestyle Factors and their Contribution to Preterm Birth—An Overview. Indian J. Occup. Envrion. Med. (2017.0) 21 9-17. DOI: 10.4103/ijoem.IJOEM_155_16 14. Londero A.P., Rossetti E., Pittini C., Cagnacci A., Driul L.. Maternal age and the risk of adverse pregnancy outcomes: A retrospective cohort study. BMC Pregnancy Childbirth (2019.0) 19. DOI: 10.1186/s12884-019-2400-x 15. Stylianou-Riga P., Kouis P., Kinni P., Rigas A., Papadouri T., Yiallouros P.K., Theodorou M.. Maternal socioeconomic factors and the risk of premature birth and low birth weight in Cyprus: A case-control study 11 Medical and Health Sciences 1117 Public Health and Health Services Paediatrics and Reproductive Medi. Reprod. Health (2018.0) 15 157. DOI: 10.1186/s12978-018-0603-7 16. Sharma R., Biedenharn K.R., Fedor J.M., Agarwal A.. Lifestyle factors and reproductive health: Taking control of your fertility. Reprod. Biol. Endocrinol. RBE (2013.0) 11 66. DOI: 10.1186/1477-7827-11-66 17. Auger N., Hansen A.V., Mortensen L.. Contribution of Maternal Age to Preterm Birth Rates in Denmark and Quebec, 1981–2008. Am. J. Public Health (2013.0) 103 33-38. DOI: 10.2105/AJPH.2013.301523 18. Glick I., Kadish E., Rottenstreich M.. Management of Pregnancy in Women of Advanced Maternal Age: Improving Outcomes for Mother and Baby. Int. J. Womens Health (2021.0) 13 751-759. DOI: 10.2147/IJWH.S283216 19. Auger N., Abrahamowicz M., Wynant W., Lo E.. Gestational age-dependent risk factors for preterm birth: Associations with maternal education and age early in gestation. Eur. J. Obstet. Gynecol. Reprod. Biol. (2014.0) 176 132-136. DOI: 10.1016/j.ejogrb.2014.02.035 20. Fertility statistics Eurostat, April 2022 in 2016 21. Live births in Poland from 1990 to 2021, by Age of Mother 22. Nel P., Maritz A., Thongprovati O.. Motherhood and Entrepreneurship: The Mumpreneur Phenomenon. Int. J. Organ. Innov. (2010.0) 3 6-34 23. Suzumori N., Ebara T., Matsuki T., Yamada Y., Kato S., Omori T., Saitoh S., Kamijima M., Sugiura-Ogasawara M.. Effects of long working hours and shift work during pregnancy on obstetric and perinatal outcomes: A large prospective cohort study—Japan Environment and Children’s Study. Birth (2020.0) 47 67-79. DOI: 10.1111/birt.12463 24. McDonald S.W., Kingston D., Bayrampour H., Dolan S.M., Tough S.C.. Cumulative psychosocial stress, coping resources, and preterm birth. Arch. Women’s Ment. Health (2014.0) 17 559-568. DOI: 10.1007/s00737-014-0436-5 25. Hackney K.J., Daniels S.R., Paustian-Underdahl S.C., Perrewé P.L., Mandeville A., Eaton A.A.. Examining the effects of perceived pregnancy discrimination on mother and baby health. J. Appl. Psychol. (2021.0) 106 774-783. DOI: 10.1037/apl0000788 26. Cai C., Vandermeer B., Khurana R., Nerenberg K., Featherstone R., Sebastianski M., Davenport M.H.. The impact of occupational shift work and working hours during pregnancy on health outcomes: A systematic review and meta-analysis. Am. J. Obstet. Gynecol. (2019.0) 221 563-576. DOI: 10.1016/j.ajog.2019.06.051 27. Mutsaerts MA Q., Groen H., Huiting H.G., Kuchenbecker WK H., Sauer PJ J., Land J.A., Stolk R.P., Hoek A.. The influence of maternal and paternal factors on time to pregnancy—A Dutch population-based birth-cohort study: The GECKO Drenthe study. Hum. Reprod. (2020.0) 27 583-593. DOI: 10.1093/humrep/der429 28. Zeitlin J.A., Saurel-Cubizolles M.-J., Ancel P.-Y.. Marital status, cohabitation, and the risk of preterm birth in Europe: Where births outside marriage are common and uncommon. Paediatr. Perinat. Epidemiol. (2002.0) 16 124-130. DOI: 10.1046/j.1365-3016.2002.00396.x 29. Raab R., Hoffmann J., Spies M., Geyer K., Meyer D., Günther J., Hauner H.. Are pre- and early pregnancy lifestyle factors associated with the risk of preterm birth? A secondary cohort analysis of the cluster-randomised GeliS trial. BMC Pregnancy Childbirth (2022.0) 22. DOI: 10.1186/s12884-022-04513-5 30. Adu-Bonsaffoh K., Gyamfi-Bannerman C., Oppong S.A., Seffah J.D.. Determinants and outcomes of preterm births at a tertiary hospital in Ghana. Placenta (2019.0) 79 62-67. DOI: 10.1016/j.placenta.2019.01.007 31. Farbu J., Haugen M., Meltzer H.M., Brantsæter A.L.. Impact of singlehood during pregnancy on dietary intake and birth outcomes- a study in the Norwegian Mother and Child Cohort Study. BMC Pregnancy Childbirth (2014.0) 14. DOI: 10.1186/s12884-014-0396-9 32. Izzatul K., Ardelia A., Hardianto G., Nuswantoro D.. Passive smoker during pregnancy is a risk factor of low birth weight. Maj Obstet. Ginekol. (2019.0) 27 12-16 33. Ko T.-J., Tsai L.-Y., Chu L.-C., Yeh S.-J., Leung C., Chen C.-Y., Chou H.-C., Tsao P.-N., Chen P.-C., Hsieh W.-S.. Parental Smoking During Pregnancy and Its Association with Low Birth Weight, Small for Gestational Age, and Preterm Birth Offspring: A Birth Cohort Study. Pediatr Neonatol. (2014.0) 55 20-27. DOI: 10.1016/j.pedneo.2013.05.005 34. Holloway A.C., Salomon A., Soares M.J., Garnier V., Raha S., Sergent F., Nicholson C.J., Feige J.-J., Benharouga M., Alfaidy N.. Characterization of the adverse effects of nicotine on placental development: In vivo and in vitro studies. Am. J. Physiol. Metab. (2014.0) 306 443-456. DOI: 10.1152/ajpendo.00478.2013 35. Sochaczewska D., Czeszynska M.B., Konefal H., Garanty-Bogacka B.. Maternal active or passive smoking in relation to some neonatal morphological parameters and complications. Ginekol. Pol. (2010.0) 81 687-692. PMID: 20973206 36. Oliver-Williams C., Fleming M., Wood A., Smith G.. Previous miscarriage and the subsequent risk of preterm birth in Scotland, 1980–2008: A historical cohort study. BJOG Int. J. Obstet. Gynaecol. (2015.0) 122 1525-1534. DOI: 10.1111/1471-0528.13276 37. Magnus M.C., Wilcox A.J., Morken N.-H., Weinberg C.R., Håberg S.E.. Role of maternal age and pregnancy history in risk of miscarriage: Prospective register based study. BMJ (2019.0) 364 1869 38. Molina-García L., Hidalgo-Ruiz M., Cocera-Ruíz E.M., Conde-Puertas E., Delgado-Rodríguez M., Martínez-Galiano J.M.. The delay of motherhood: Reasons, determinants, time used to achieve pregnancy, and maternal anxiety level. PLoS ONE (2019.0) 14. DOI: 10.1371/journal.pone.0227063 39. Jølving L.R., Nielsen J., Kesmodel U.S., Nielsen R.G., Beck-Nielsen S.S., Nørgård B.M.. Prevalence of maternal chronic diseases during pregnancy—A nationwide population based study from 1989 to 2013. Acta Obstet. Et Gynecol. Scand. (2006.0) 95 1295-1304. DOI: 10.1111/aogs.13007 40. Guariguata L., Linnenkamp U., Beagley J., Whiting D.R., Cho N.H.. Global estimates of the prevalence of hyperglycaemia in pregnancy. Diabetes Res. Clin. Pract. (2014.0) 103 176-185. DOI: 10.1016/j.diabres.2013.11.003 41. Ye W., Luo C., Huang J., Li C., Liu Z., Liu F.. Gestational diabetes mellitus and adverse pregnancy outcomes: Systematic review and meta-analysis. BMJ (2022.0) 377 e067946. DOI: 10.1136/bmj-2021-067946 42. Shub A., Lappas M.. Pregestational diabetes in pregnancy: Complications, management, surveillance, and mechanisms of disease—A review. Prenat. Diagn. (2020.0) 40 1092-1098. DOI: 10.1002/pd.5718 43. Lepercq J., Coste J., Theau A., Dubois-Laforgue D., Timsit J.. Factors associated with preterm delivery in women with type 1 diabetes: A cohort study. Diabetes Care (2004.0) 27 2824-2828. DOI: 10.2337/diacare.27.12.2824 44. Kiefer M.K., Finneran M.M., Ware C.A., Fareed N., Joseph J., Thung S.F., Costantine M.M., Landon M.B., Gabbe S.G., Venkatesh K.K.. Association of change in haemoglobin A1c with adverse perinatal outcomes in women with pregestational diabetes. Diabet. Med. (2022.0) 39 e14822. DOI: 10.1111/dme.14822 45. Zhang Y., Xiao C.-M., Zhang Y., Chen Q., Zhang X.-Q., Li X.-F., Shao R.-Y., Gao Y.-M.. Factors Associated with Gestational Diabetes Mellitus: A Meta-Analysis. J. Diabetes Res. (2021.0) 2021 1-18. DOI: 10.1155/2021/6692695 46. Davidson AJ F., Park A.L., Berger H., Aoyama K., Harel Z., Cohen E., Cook J.L., Ray J.G.. Association of Improved Periconception Hemoglobin A1cWith Pregnancy Outcomes in Women with Diabetes. JAMA Netw. Open (2020.0) 3 2030207. DOI: 10.1001/jamanetworkopen.2020.30207 47. Delker E., Bandoli G., LaCoursiere Y., Ferran K., Gallo L., Oren E., Gahagan S., Ramos G.A., Allison M.. Chronic hypertension and risk of preterm delivery: National Longitudinal Study of Adolescents to Adult Health. Paediatr. Perinat. Epidemiol. (2022.0) 36 370-379. DOI: 10.1111/ppe.12858 48. Andersen S.L., Olsen J., Wu C.S., Laurberg P.. Low Birth Weight in Children Born to Mothers with Hyperthyroidism and High Birth Weight in Hypothyroidism, whereas Preterm Birth Is Common in Both Conditions: A Danish National Hospital Register Study. Eur. Thyroid. J. (2013.0) 2 135-144. DOI: 10.1159/000350513 49. Sheehan P.M., Nankervis A., Araujo E., Da Silva Costa F.. Maternal Thyroid Disease and Preterm Birth: Systematic Review and Meta-Analysis. J. Clin. Endocrinol. Metab. (2015.0) 100 4325-4331. DOI: 10.1210/jc.2015-3074 50. Hubalewska-Dydejczyk A., Lewiński A., Milewicz A., Radowicki S., Poręba R.. Management of thyroid diseases during pregnancy. Endokrynol. Pol. (2011.0) 62 362-381. PMID: 21879479 51. Rao M., Zeng Z., Zhou F., Wang H., Liu J., Wang R., Wen Y., Yang Z., Su C., Su Z.. Effect of levothyroxine supplementation on pregnancy loss and preterm birth in women with subclinical hypothyroidism and thyroid autoimmunity: A systematic review and meta-analysis. Hum. Reprod. Update (2019.0) 25 344-361. DOI: 10.1093/humupd/dmz003 52. Norman J.E.. Progesterone and preterm birth. Int. J. Gynecol. Obstet. (2020.0) 150 24-30. DOI: 10.1002/ijgo.13187 53. Mesiano S.A., Peters G.A., Amini P., Wilson R.A., Tochtrop G.P., van Den Akker F.. Progestin therapy to prevent preterm birth: History and effectiveness of current strategies and development of novel approaches. Placenta (2019.0) 79 46-52. DOI: 10.1016/j.placenta.2019.01.018 54. Jarde A., Lutsiv O., Beyene J., McDonald S.. Vaginal progesterone, oral progesterone, 17-OHPC, cerclage, and pessary for preventing preterm birth in at-risk singleton pregnancies: An updated systematic review and network meta-analysis. BJOG Int. J. Obstet. Gynaecol. (2019.0) 126 556-567. DOI: 10.1111/1471-0528.15566
--- title: 'Understanding the Combined Effects of High Glucose Induced Hyper-Osmotic Stress and Oxygen Tension in the Progression of Tumourigenesis: From Mechanism to Anti-Cancer Therapeutics' authors: - Gayathri K. G. - Puja Laxmanrao Shinde - Sebastian John - Sivakumar K. C. - Rashmi Mishra journal: Cells year: 2023 pmcid: PMC10047272 doi: 10.3390/cells12060825 license: CC BY 4.0 --- # Understanding the Combined Effects of High Glucose Induced Hyper-Osmotic Stress and Oxygen Tension in the Progression of Tumourigenesis: From Mechanism to Anti-Cancer Therapeutics ## Abstract High glucose (HG), a hallmark of the tumour microenvironment, is also a biomechanical stressor, as it exerts hyper-osmotic stress (HG-HO), but not much is known regarding how tumour cells mechanoadapt to HG-HO. Therefore, this study aimed to delineate the novel molecular mechanisms by which tumour cells mechanoadapt to HG/HG-HO and whether phytochemical-based interference in these mechanisms can generate tumour-cell-selective vulnerability to cell death. Mannitol and L-glucose were used as hyper-osmotic equivalents of high glucose. The results revealed that the tumour cells can efficiently mechanoadapt to HG-HO only in the normoxic microenvironment. Under normoxic HG/HG-HO stress, tumour cells polySUMOylate a higher pool of mitotic driver pH3(Ser10), which translocates to the nucleus and promotes faster cell divisions. On the contrary, acute hypoxia dampens HG/HG-HO-associated excessive proliferation by upregulating sentrin protease SENP7. SENP7 promotes abnormal SUMOylation of pH3(Ser10), thereby restricting its nuclear entry and promoting the M-phase arrest and cell loss. However, the hypoxia-arrested cells that managed to survive showed relapse upon reversal to normoxia as well as upregulation of pro-survival-associated SENP1, and players in tumour growth signalling, autophagy, glycolytic pathways etc. Depletion of SENP1 in both normoxia and hypoxia caused significant loss of tumour cells vs undepleted controls. SENP1 was ascertained to restrict the abnormal SUMOylation of pH3(Ser10) in both normoxia and hypoxia, although not so efficiently in hypoxia, due to the opposing activity of SENP7. Co-treatment with Momordin Ic (MC), a natural SENP1 inhibitor, and Gallic Acid (GA), an inhibitor of identified major pro-tumourigenic signalling (both enriched in Momordica charantia), eliminated surviving tumour cells in normal glucose, HG and HG-HO normoxic and hypoxic microenvironments, suggesting that appropriate and enhanced polySUMOylation of pH3(Ser10) in response to HG/HG-HO stress was attenuated by this treatment along with further dampening of other key tumourigenic signalling, due to which tumour cells could no longer proliferate and grow. ## 1. Introduction Global Cancer Statistics (GLOBOCAN, 2020) shows an alarming rise in the cancer burden [1]. The primary reasons for treatment failure are due to rapidly expanding tumour heterogeneity and recurrence, chemo- and radio-resistance, oncogenic mutations, alternative splice variants, gene polymorphisms and rapid chromatin remodelling, all of which have posed enormous difficulties and challenges in identifying the specific cellular targets for rational anti-tumour therapeutics and prophylactic vaccine design [2,3,4,5]. Therefore, deeper insights and newer arenas that expose tumour cell vulnerabilities need to be rapidly explored to discover tumour-selective drugs with high therapeutic efficacy, sustainable costs and successful clinical outcomes. Tumour cells display robust adaptation to the microenvironment (TME)-associated biomechanical challenges. Therefore, therapeutic targeting of TME-associated mechanoadaptation is emerging as a ‘hot spot’ of tumour cell vulnerability and a promising avenue in cancer eradication [6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]. High glucose (HG) is not only a hallmark of cancer but also acts as a hyper-osmotic stressor in TME. Therefore, the major objectives of this study were to excavate the mechanisms by which tumour cells osmo-adapt to high glucose associated hyper-osmotic stress in normoxic and hypoxic microenvironments and how this adaptation can be disabled. Irrespective of diabetes, hyperglycaemia is a prominent causative factor for cancer cell proliferation and promotes poor patient survival. Hyperglycaemia is often encountered in cancer patients due to several factors such as high carbohydrate uptake, altered metabolism, hormonal disorders, chronic stress, obesity, anti-cancer drugs’ side effects, etc. [ 21,22]. The blood glucose level in a healthy adult is between 90 to 100 mg/dL, and concentrations higher than 200 mg/dL (11.1 mM) are hyperglycaemic. Glioblastoma patients have been evidenced to have as high as 459 mg/dL (approx. 25.5 mM) of blood glucose levels, and hyperglycaemic concentrations are equally well correlated with cervical cancer progression in patients [23,24]. Although there are also glucose-restricted zones in tumours with 0.45 g/L (2.5 mM) of glucose, PET scan studies have confirmed the existence of hyperglycaemia in tumours vs other body organs, probably due to interstitial fluid flow [24]. In addition, NOD (non-obese diabetic) mice, which are the chosen model for tumour xenograft studies, show hyperglycaemic glucose levels between 10 to 30 mM, which is why tumour progression is enabled, confirming the crucial role of hyperglycaemia in driving tumourigenesis [25]. Furthermore, this is the base reason that tumour cells are standardly grown and passaged in 4.5 g/L (25 mM) glucose-containing medium, and tumour cell studies are performed in media containing 10–30 mM of glucose. There are few reports that cancer cells rapidly adjust to high-glucose-mediated hyper-osmotic (HG-HO) stress by adjusting their shape and volume, which influence the net surface tension and enable cell survival [26,27]. In addition, the rounding of cells is reported to signal cells to enter the cell cycle and divide [28,29,30,31]. More alarmingly, the emergent phenotypes in osmotic stress conditions are associated with increased drug resistance, providing additional survival advantages to tumourous cells [27]. Therefore, given the timely relevance, we explored the contemporary area of mechano-oncology for excavating the mechanisms associated with tumour cell mechanical adaptations to hyperglycaemia-associated hyper-osmotic stress (HG-HO) under high and low oxygen tension. Effects of the osmotic components of glucose can be dissected from its metabolic component and can be investigated by using the glucose osmotic mimics mannitol and l-glucose [32,33]. Hence, the treatment protocols to study HG/HG-HO-mediated molecular changes were performed in the presence of mannitol and l-glucose as independent controls for the media osmolality. Appropriate volumes of mannitol (osmo-mimic for d-glucose, not uptaken, not metabolized) or l-glucose (enantiomer of d-glucose, not metabolized) were added to the physiological or normal glucose baseline of 5.5 mM (NG), to obtain the equivalent osmolality as in the HG condition. Similar changes observed in HG and the HG-HO can be attributed to the function of the osmotic component of HG rather than its signalling component. However, it is essential not to take mannitol/l-glucose directly (bypassing baseline 5.5 mM glucose levels), because this can disturb the glucose-mediated signalling in normal cells and may generate artefacts [33]. The general experimental set-up is explained in Supplementary Figure S1. We escalated the major findings for validation in various human tumour cell lines. We sought convincing evidence of the same in the relevant clinical samples, which is imperative for bench-to-bedside translation. We further investigated whether phytochemical-base perturbation of crucial mechanisms in HG osmoregulation can promote the uptake of chemotherapies and attenuate tumour growth. ## 2. Materials and Methods Please note that additional details on reagents and standard protocols are available in the supporting information file accompanying this article. ## 2.1. Cell Lines, Reagents Human cervical cancer cell lines CaSki (HPV16+ve and HPV18+ve) and C-33A (HPV-ve); metastatic colon cancer cell line HCT116; metastatic pancreatic cancer cell line MIAPaCa-2; metastatic breast cancer cell line MDA-MB-231 and human normal keratinocytes cell line HaCaT were obtained from and authenticated by American Tissue Culture Collection, Manassas, VA, USA. SVG, the immortalized human astrocyte cell line, was a gift from Prof. Pankaj Seth, National Brain Research Centre, Manesar, India and was initially procured from Prof. Eugene O Major, NINDS/NIH, Bathesda, USA. H9C2, rat cardiomyoblasts, were procured from the cell line repository of the National Centre for Cell Science (NCCS), Pune, India. Cancer cells were maintained in DMEM (Gibco-12100-046) with $10\%$ Foetal Bovine Serum (Himedia—RM9955) and 1x antibiotics (Himedia—A002). Normal cells were cultured as described previously [12,13,34]. All common chemicals were purchased from Sigma-Aldrich, St.Louis, MO unless otherwise indicated. The antibodies used are as follows: pH3(Ser10) [9701S], phospho-p$\frac{44}{42}$ MAPK (Thr202/Tyr204) [4370]—Cell Signaling Technologies, Danvers, MA, USA; pH3(Ser10) [ab267372], SUMO1(ab32058), SUMO4(ab126606), SUMO$\frac{2}{3}$ (ab81371)—Abcam, Waltham, MA, USA; SUMO2 (sc26972), HIF1a (sc10790), GAPDH—(sc47724)—Santacruz Biotechnology, Dallas, TX, USA; Cholesterol (abx100311)—Abbexa, Houston, TX, USA; AURKB (A1020), ALDOC (A11618), Phospho p70S6K (T389) [AP1059]—Abclonal, Woburn, MA, USA. ## 2.2. Cell Culture Treatments with Normal, High-Glucose and Hyper-Osmotic Equivalents in Normoxia and Hypoxia Conditions The treatments were conducted by culturing the cells either under normoxia or chemically induced hypoxia by the addition of 150 µM of CoCl2 [35]. In both conditions, the effects of normal as well as high glucose (HG) concentrations and the contribution of its corresponding osmotic component (HG-HO) to the observed effects were studied by treating the cells with either 20 mM of glucose as the hyperglycaemic condition or 20 mM of mannitol as the osmotic control for the hyperglycaemic condition, while maintaining 5.5 mM of glucose as basal level (NG) in all the conditions [32,33]. In the figures, normal glucose level, high glucose level and mannitol osmotic control are designated as Nor Glu, High Glu and Mann Ctl, respectively, except in Figures 1 and 2. Figures 1 and 2 have data on various concentrations of high glucose and two osmotic controls; therefore, normal glucose level, high glucose levels, l-glucose osmotic controls and mannitol osmotic controls are designated as 5.5 mM Glu, $\frac{11.5}{25.5}$ mM High Glu, $\frac{11.5}{25.5}$ mM L-Glu Ctl and $\frac{11.5}{25.5}$ mM Mann Ctl, respectively. In the text, 5.5 mM normal glucose level is abbreviated as NG, high glucose as HG and high-glucose hyper-osmotic stress as HG-HO. ## 2.3. Confluence Analysis Cells were cultured in 6-well plates with an initial seeding density of 15,000 cells/well. After exposing cells to the basic experimental set-up treatments, the cells were fixed with $1.5\%$ PFA for 20 min at RT. The wells were imaged at 4x magnification using the TECAN Multimode reader/imager. The whole well images were thresholded using the Otsu tool in Fiji image analysis software (FIJI is just ImageJ $\frac{2.9.0}{1.53}$t; Java 1.8.0_322; 64-bit) [36],and the number of cells was counted by running the “Analyze particles” option. ## 2.4. Floating Cell Count with Trypan Blue The cells were seeded in 100 mm cell culture dishes and were treated with high glucose/mannitol in conjunction with either normoxia or hypoxia for 96 h. The medium was collected daily, and the floating cells were pelleted down by centrifuging the media at 2000 rpm for 5 min. The floating cell pellet was re-suspended in 30 µL medium, and equal volumes of this cell suspension were mixed with trypan blue solution. Cell count was scored manually using a hemocytometer, and dead and live cell counts per quadrant were registered. ## 2.5. Proximity Ligation Assay (PLA) A Duolink in situ PLA kit from Sigma-Aldrich, St. Louis, MO, USA was used to investigate the interaction between SUMO2 and pH3(Ser10), according to the manufacturer’s instructions. Briefly, the cells were fixed with $1.5\%$ PFA for 20 min at RT after the treatments. Cell permeabilization was conducted with $0.25\%$ saponin in PBS. One drop of blocking solution per 1 cm2 area was added, and the slides were incubated in a pre-heated humidified chamber at 37 °C for 30 min. Respective primary antibodies were diluted to recommended concentrations in the antibody diluent and were added to the cells, and the slides were incubated overnight at 4 °C in a humidified chamber. Post incubation, the slides were washed with 1x wash buffer A, and the plus and minus PLA probes, diluted in 1:5 in antibody diluent, were added to the cells. After incubating the cells for 1 h at 37 °C with the PLA probe solution, the slides were washed with 1x wash buffer A. The slides were then incubated with 40 µL of ligation mix per well in a humidified chamber for one hour at 37 °C. After washing with 1× wash buffer A, the required volume of signal amplification mix with polymerase enzyme was added to the cells. It was incubated for one hundred minutes at 37 °C in a humidified chamber. Final washes were given with wash buffer A and wash buffer B, and the cells were stained for Hoechst for six minutes at RT, followed by mounting in $70\%$ glycerol. ## 2.6. Immunohistochemistry on Tissue Microarrays Uterine cervical tumour tissue array (US Biomax, Derwood, MD, USA, cat no. CR2087): uterine cervical carcinoma tissue microarray containing 104 cases of malignant tumour (87 squamous cell carcinoma, 3 adeno squamous carcinomas and 14 adenocarcinomas) with duplicate cores were used for IHC. The multiple tumour tissue array (cat nos. T6235700-1) was from BioChain Institute Inc., Newark, CA, USA. Antigen retrieval was performed on TMA slides by boiling the tissues in basic heat-induced epitope recovery citrate buffer (0.01 M Sodium Citrate buffer, pH6). The TMA slides were then permeabilized with $0.01\%$ digitonin for 30 min at RT and blocked for 1 h in the blocking mixture, constituting $5\%$ BSA and $2\%$ donkey serum. TMA slides were incubated with primary antibodies overnight at 4 °C. Fluorophore-labelled secondary antibodies were incubated for 1.5 h at RT. Slides were mounted in $70\%$ glycerol in PBS after Hoechst staining for 5 min at RT. The stained sections were visualized and imaged using a confocal microscope and were analysed with Fiji Image analysis software. ## 2.7. FRET Analysis Antibody-mediated FRET assay was performed to confirm the direct interaction between SUMO2 and pH3(Ser10). Briefly, the cells were fixed after treatment with $1.5\%$ PFA for 20 min at RT. The cells were immunostained with pH3(Ser10) (9701S) and SUMO2 (sc-26972) antibodies. Signal amplification was performed for pH3(Ser10) to obtain strong donor fluorescence intensity by performing TSA-Biotin-mediated fluorescence enhancement. Further, immunostaining of pH3(Ser10) and SUMO2 was developed with donkey anti-mouse Cy3 (Jackson Immuno Research, West Grove, PA, USA; 1:200) and donkey anti-goat IgG conjugated with Alexa 647 (Invitrogen, Waltham, MA, USA; 1:100), respectively. The FRET signal intensities on mitotic chromosomes were obtained by outlining the chromosomes’ periphery in Fiji and subsequent use of the ‘Measure’ tool. The nuclear and cytoplasmic FRET channel (acceptor emission after donor excitation) signals were likewise obtained, and nuclear vs cytoplasmic FRET signal intensities were graphically represented. ## 2.8. Three-Dimensional Pellet Cultures and Drug Treatments In 96-wellround-bottom plates, 3 × 104 cells were seeded per well in MEM media [12]. The plates were centrifuged in a swinging bucket rotor at 500× g for 5 min to generate the 3D cell pellet. The HG, HG-HO and oxygen tension treatments were started post 24 h. After establishing each microenvironmental condition for 96 h, individual and in-combination drug treatments were started, which were continued for up to 9 days to capture the response of individual cell lines to the therapeutics. Momordin Ic (MC) was used at a concentration of 25 µM and Gallic Acid (GA) at 100 µM, as these concentrations did not show toxicity to normal cells in the MTT assay. In addition, these dosages have been validated in cell culture studies, where the experiments were escalated to in vivo tumour models [37,38]. The pellet images were captured daily, and the volume analysis was performed using the Reconstruction and Visualization from a Single Projection (ReViSP; https://sourceforge.net/p/revisp, accessed on 21 February 2023) tool, which enables the 3D volume reconstruction by quantitating the voxels (3D pixels) [39,40]. ## 2.9. Image Analysis Confocal images were hyper-stacked to visualize the maximum intensity projections using Fiji software, and the cell outlines were manually drawn. The fluorescence intensity was measured for the individual cells and their respective nuclei. Cytoplasmic fluorescence intensity was obtained by subtracting the integrated fluorescence intensity of the nucleus from the whole cell. Image acquisition parameters for each channel were kept the same across each condition and over independent replicates. Colocalization analysis was performed using the Coloc tool, and the extent of colocalization was expressed as Mander’s coefficient of colocalization (R). ## 2.10. Statistical Analysis Statistical analyses were performed using one-tailed unpaired Bonferroni’s t-test. The following p values represented statistical significance: * p ≤ 0.05, p ≤ 0.01 and * p ≤ 0.001. All comparisons were made with the control 5.5 mM normoxic condition unless otherwise indicated. Data are represented as means ± SD and averaged from at least three independent experiments. ## 3.1. High-Glucose-Induced Hyper-Osmotic Stress Drives Tumour Cell Proliferation under Normoxia In order to delineate the mechanisms by which tumour cells osmo-adapt to high-glucose-associated hyper-osmotic (HG-HO) stress in normoxic and hypoxic microenvironments, the experimental set-up was designed, which is explained in Supplementary Figure S1. Acute hypoxia was generated as described previously [35], and elevated hypoxia marker HIF1a was confirmed in 12 and 48 h (Supplementary Figure S2). Tumour cells exposed to high glucose concentrations (HG, 11.5 and 25.5 mM) and the hyper-osmotic equivalents (HG-HO) showed high proliferation of the bulk tumour population in normoxia (Figure 1A–C and Figure S3A). The 2D colony formation and 3D anchorage-independent spheroid-based growth assays (soft agar assay) that determine the capacities of cells to self-renew also showed that individual tumour cells could efficiently grow in HG-HO condition and thereby adapt to this stress, but only in normoxia (Figure 1D–F, Figure 2A and Figure S3B,C). Overall, these assays suggest that tumour cells can over-proliferate not only in response to better growth conditions provided by HG (5.6–20 mM extra glucose) but also due to HG-HO stress, which has only baseline glucose levels of 5.5 mM. Indeed, proliferation-associated mitotic rounding has been reported to enable osmo-adaptive membrane tension and, thereby, could be a clever mechanoadaptive strategy to survive in adverse conditions that present osmotic challenges [26,27,28,29,30,31]. ## 3.2. Hypoxia Retards High-Glucose Hyper-Osmotic-Stress-Induced Tumour Cell Proliferation by Triggering G2/M Cell Cycle Arrest We observed significantly less survival under acute hypoxia irrespective of the presence of NG, HG or HG-HO conditions. This suggests that acute hypoxia attenuated HG/HG-HO osmo-adaptive mechanisms. Therefore, delineating the identity of hypoxia-induced ‘anti-mechanoadaptive’ molecular mechanisms was particularly interesting at this stage. These were clues to strategies for transforming tumour cells vulnerable to death. Therefore, we delve deeper into the cell fate and molecular analysis of acute hypoxia-induced compromised survival. Under hypoxia, a significant number of floating cell populations were observed that were mostly dead. When this floating cell population was replated under normoxic conditions, a few cells revived and formed colonies, signifying the dangerous capacity to relapse upon the return of conducive conditions (Figure 2B). However, the acute hypoxia-induced cell death and the residual adherent cell growth were not due to senescence (Supplementary Figure S4A). Instead, such phenotypes expressed markers of stemness (Supplementary Figure S4B,C) and were able to reverse growth retardation upon exposure to normoxia (Figure 2C). In addition, the floating cell count, although significant, could not wholly account for the dramatic loss in tumour cell numbers observed in acute hypoxia. Since cell proliferation is regulated by the cell cycle, flow-cytometry-assisted cell cycle analysis was performed, which revealed robust G2/M arrest in acute hypoxia (Figure 3A). Further, a higher mitotic cell count over the total number of cells was registered in hypoxic conditions, but this did not corroborate with the expected rise in proliferation (data shown in Figure 1 and Figure 2), which again suggests that acute hypoxia induces M-phase arrest (Figure 3B). Furthermore, the residual growth in hypoxia showed the formation of micronuclei, an indicator of abnormal cell cycle and mitotic chromosome segregation, which again points to the M-phase defects (Supplementary Figure S4D) [41]. A deeper analysis of the M-phase distribution of tumour cells revealed that cells in hypoxia were stuck in pro-metaphase and late anaphase/telophase stages (Figure 3C–F). Several kinases orchestrate the transition from G2 to the M phase. However, AURKB, predominantly involved in the exit from prophase and transition into metaphase–anaphase, has significant roles in the process of cytokinesis [42]. Most of the molecular players involved in M-phase progression, including AURKB, have been identified as substrates of SUMOylation, especially by the SUMO2 isoform [43,44] (Supplementary Table S1). Therefore, we examined the SUMO2-modified AURKB on mitotic cells, which was identified by immunostaining with pH3(Ser10), a master driver of mitosis [45,46]. We found that in acute hypoxia, the intensities of SUMO2 and AURKB were significantly less on the mitotic chromosomes and during cytokinesis (Figure 4A,B,F and Figure S5). In addition, the pH3(Ser10) intensity was less (Figure 4A,B,D). However, the low levels of AURKB and pH3(Ser10), which reached mitotic chromosomes, were mainly SUMOylated, as indicated by a high colocalization coefficient (R) in acute hypoxia (Figure 4G,H). In the non-cycling cells, hypoxic conditions showed higher cytoplasmic colocalization of SUMO2-pH3(Ser10) (Supplementary Figure S6). ## 3.3. pH3(Ser10), a Master Regulator of Mitosis, Is a Target of SUMO2 and Is Abnormally SUMOylated under Acute Hypoxic Conditions The extensive colocalization of pH3(Ser10) with SUMO2 in the cytoplasm of tumour cells exposed to acute hypoxia prompted us to explore whether pH3(Ser10), a master mitotic driver, is a novel substrate of SUMO2. As SUMOylation is associated with nuclear trafficking and chromosomal loading of various cell-cycle-associated proteins, we wanted to examine whether appropriate SUMOylation of pH3(Ser10) is required for efficient nuclear localization and whether this process is predominantly hindered under acute hypoxia, which may result in observed retarded growth. In situ protein–protein interaction signal detection through proximity ligation (PLA assay) and FRET assay confirmed that a significant pool of pH3(Ser10) is conjugated to SUMO2 in both mitotic and non-mitotic cells (Figure 5, Figure 6 and Figures S7 and S8). The SUMOylated pH3(Ser10) levels were significantly higher in the cytoplasm and lowered on the mitotic chromosomes of cells exposed to acute hypoxia. In addition, cytoplasmic SUMOylated pH3(Ser10) showed aggregated organization in acute hypoxia-induced cells (Supplementary Figures S7 and S8, white arrows). ## 3.4. The Patient Tumour Tissue Array Validates the Extensive Cytoplasmic Sequestration of SUMO2-Modified pH3(Ser10) in Highly Hypoxic Regions In order to validate the correlation between cytoplasmic sequestration of SUMOylated pH3(Ser10) in hypoxia ‘very high’ regions in the context of actual clinical samples, we performed robust immunohistochemistry-based examination of various cancer tissues [47,48]. We confirmed that very high HIF1a-expressing cells have significant cytoplasmic localization of SUMO2-pH3(Ser 10) vs the nuclear enrichment in HIF1a ‘low or medium regions’ (Figure 7 and Figures S9–S15). Since highly hypoxic regions develop in advanced stages of tumour progression, it is noteworthy that significant cytoplasmic colocalization of SUMO2-pH3(Ser10) was identified in HIF1a-positive regions of advanced vs. lower tumour grades (Figure 7 and Figures S11–S13). ## 3.5. Nuclear Trafficking of SUMO2-Conjugated pH3(Ser10) Is Significantly Less in Acute Hypoxia Due to Its Phase Separation (LLPS) and Aggregation in the Cytoplasm We were curious to know what causes cytoplasmic sequestration of SUMOylated pH3(Ser10) in acute hypoxia. SUMO2 can bind to its substrate as a polySUMO chain (chain of SUMOs bound to each other) or can conjugate as individual residues at one or multiple sites. We first tried to seek insights into how SUMO2 binds to pH3(Ser10). For this, we ran docking and molecular simulations experiments and found that SUMO2 can bind efficiently to pH3(Ser 10) in the tail region (Supplementary Figure S16A). This binding was found to be very strong for the SUMO2 isoform vs other isoforms. The tail region of histones is intrinsically disordered and is robustly post-translationally modified to enable various cellular, molecular and biological functions of histones. Co-relatively, two globally constant properties of SUMO2, documented for nuclear localization of its substrates, are the substrate binding preference in the disordered regions of proteins and SUMO-phosphorylation co-modification [44,49]. We identified the intrinsically disordered amino acid sequences in the tail region of pH3(Ser10) to which SUMO2 binds. The crucial lysine in the SUMO binding motif was identified to be at the 9th position in the sequence, near the phosphorylated residue, which is at the 10th position (Supplementary Figure S16B). In addition, the binding was specific to histone H3, which was phosphorylated at the Ser10 amino acid residue vs Ser 28, which is another phosphorylation site within this protein (Supplementary Figure S16C). Intrinsically disordered proteins are highly susceptible to protein aggregation, condensation and liquid–liquid phase separation (LLPS). Since histone H3 variants have significantly high coverage of disordered sequences (Supplementary Figure S16B), we reasoned that any aberrant post-translational modification (PTM) of the histone H3 variant could condense the protein into LLPS, which may significantly impact its nuclear trafficking, cell-cycle-associated functions and its role in cell proliferation [49]. We further speculated that acute hypoxia-induced abnormal site SUMOylation could be the type of PTM that sequesters the pH3(Ser10) in the cytoplasm via LLPS formation [50]. It could thereby explain the M-phase arrested phenotype in hypoxia. To this argument, we examined the tumour cells under NG, HG and HG-HO conditions in normoxia and hypoxia through a high-resolution confocal microscope. We found convincing evidence of SUMO2-pH3(Ser10) membraneless cytoplasmic phase separation (Figure 8). Intriguingly, membranous organelles, identified by membrane labelling with an anti-cholesterol antibody, frequently bordered these LLPS regions, like fences. The significance of this observation is yet to be clear. However, it could be that these organelles are randomly trapped at the periphery of the phase-separated aggregates, or it could be that these organelles actively border the LLPS to prevent its further spread in the cell, as LLPS expansion can invariably choke the cells to death. The latter possibility is likely to help the tumour cells survive acute hypoxia while lying dormant [51,52]. ## 3.6. Abnormal Activity of SENPs, under Acute Hypoxia, Sequesters SUMOylated pH3(Ser10) in the Cytoplasm via LLPS, Thereby Generating Tumour Cell Growth Arrested Phenotype Since SUMO2 conjugation and de-conjugation (de-SUMOylation) are regulated by sentrin proteases, we profiled the transcripts of SENP isoforms. We found that SENP1 and SENP7 were significantly high in acute hypoxia (Figure 9). Subsequent RNA interference studies showed that SENP7 downregulation enhanced proliferation in both hypoxia and normoxia, supported by an observed increase in cell counts in SENP7 downregulation, and the reverse was observed when SENP1 was downregulated (Figure 10A and Figure S17). This suggests that SENP7 does not support survival, while SENP1 does. Interestingly, SENP1 downregulation generated SUMO2-pH3(Ser10) LLPS even in normoxia, and in hypoxia, the LLPS-like phenotype intensified (Figure 10B and Figure S18). SENP1 downregulation can prevent SUMO2 substrate degradation and promote its abnormal accumulation. These observations again suggest that SENP1 is pro-survival and is not involved in LLPS generation in acute hypoxia [53,54,55,56,57,58,59]. It further indicates that SENP7 may be involved in LLPS generation, which is restricted by SENP1, due to which a few dormant stem-like tumour cells resist acute hypoxia challenge. SENP7’s polySUMO chain length shortening activity is structurally specific for SUMO2-conjugated substrates [56]. Therefore, in hypoxia with elevated levels of SENP7, the SUMO2 nuclear signal was less vs other isoforms (although some punctate cytoplasmic signals were observed for other isoforms as well in hypoxic vs normoxic conditions) (Supplementary Figure S19). Further, the SUMO2 substrate, such as AURKB, was not associated with a significant LLPS phenotype in hypoxia, although its nuclear trafficking was inhibited (Supplementary Figure S20). Therefore, LLPS formation is very substrate-specific to certain SUMO targets only [57]. SENP7 is predominantly involved in the deSUMOylation of polySUMO chains, causing shortening in the polySUMO2 chain conjugated to its substrate. The SUMO2 freed from the polySUMO chain can further bind to the same substrate at other conducive sites via the activity of ligases (Figure 11A, diagrammatic representation of SENP7 action). Thus, the nature of SUMOylation and positional location of SUMOylation cumulatively determine the substrate’s physical state, solubility and functionality in a context-dependent manner. Therefore, polySUMO chain deSUMOylating SENPs can lead to the loss of SUMO substrate functions or vice versa. To test abnormal SUMO processing of pH3(Ser10), we needed to perform the immunoprecipitation experiments either from the adherent cell lysates or the cytoplasmic fraction, which were technically challenging for the hypoxic conditions due to the formation of LLPS. However, we extracted the nuclear lysates from tumour cells in HG and HG-HO hypoxic and normoxic conditions, and via co-immunoprecipitation with the SUMO2 antibody, profiled the SUMOylation states of pH3(Ser10) that managed to enter the nucleus (Figure 11B). The input showed relatively less levels of the pH3(Ser10) pool in the hypoxic nuclei. The higher molecular weight bands of SUMOylated pH3(Ser10) were missing in the immunoprecipitated hypoxic samples (Figure 11B, indicated by asterisks). In addition, the SUMOylated pH3(Ser10) levels were less (Figure 11C). This suggests that a pool of SUMOylated pH3(Ser10) that managed to enter the hypoxic nuclei was majorly differentially modified by SUMO2 and, thereby, may not be efficient in association with chromosomes, as documented in Figure 4, Figure 5 and Figure 6. This is correlated with the observation that depletion of SENP7 allows more SUMOylated pH3(Ser10) to enter the nuclei and thereby promotes proliferation (Supplementary Figure S17). ## 3.7. Momordin Ic (MC), an SENP1 Inhibitor, Can Mimic Hypoxia-Induced Abnormal SUMOylation of pH3(Ser10) and Thereby Retards Tumour Cell Proliferation in Normoxic High-Glucose Hyper-Osmotic Conditions Our observations so far suggested that SENP1 downregulation drove SUMOylated pH3(Ser10) into an LLPS-like phenotype in normoxic conditions, thereby inhibiting the cell proliferation functions of pH3(Ser10) and retarding tumour growth (Figure 10A,B). It suggests that we need to reduce the levels of SENP1 to promote the abnormal accumulation of SUMO2-pH3(Ser10), so that it can stall mitosis. Normal cells are also expected to recruit SUMO2 substrates for mitotic events, but the required levels are lower than cancer cells [58,59]. In addition, SENP1 expression is also lower in normal cells, so SENP1 inhibition and resultant aberrant SUMO2 substrate accumulation would be far less in normal cells. We used Momordin Ic (MC), a natural SENP1 inhibitor, to produce the growth-retarding effects associated with abnormal SUMO2 processing of pH3(Ser10) [37]. MC concentrations were titrated for non-toxic effects on normal cells and were determined to be 25 µM (Supplementary Figure S21). Upon Momordin Ic treatment, we noticed an increase in SUMO2-pH3(Ser10) abnormal colocalization. In addition, in several regions of the tumour cell, pH3(Ser10) was observed to form large aggregates (Figure 12A, white arrows). Within 12 h of MC treatment, a significant rise in mitotic cell numbers was noticed in normoxic HO and HG-HO conditions (Supplementary Figure S22). However, by 24 h of MC treatment, there was a significant loss in cell count. It suggests that cells first underwent M-phase arrest, which manifested as higher mitotic cell numbers per field of view, but there was no actual growth. By the next 24 h of treatment (total of 48 h), most cells lifted off the substrate and died by anoikis, as determined by trypan blue hemocytometry-based live/dead assay. The effect of MC on residual cell growth associated with acute hypoxia was far less, probably because SUMOylated pH3(Ser10) was already inhibited from associating with the mitotic chromosomes (Supplementary Figure S23). In addition, there was a pre-existing M-phase arrest. In order to kill the hypoxia-induced residual cell survival, which possesses a capacity for re-growth, a robust strategy was needed. In our previous experience, we reported that the plants from which the bioactive drug molecule is derived also have many other chemotherapeutics and chemosensitizers [34]. Generally, the administration of plant extracts from which the active molecule is derived is preferred in traditional medicine because such plants are enriched in triterpenes and flavonoids in the correct concentrations to harm tumorous growth selectively. We therefore scrutinized the chemical composition of Momordica charantia (enriched in Momordin Ic) and pinned our hope on the Gallic Acid found in it, as the literature survey showed that it could target major cancer survival players such as EGFR, MAPK, autophagy effectors, glycolysis, etc. [ 38,60,61,62,63,64,65]. We titrated the Gallic Acid effective dose that is non-toxic to normal cells (Supplementary Figure S24). We tested its effects on lowering the growth factors such as EGFR and MAPK1, autophagy effectors such as pEIF4EBP1 and p70-S6K (RPS6KB1) and glycolysis regulators ALDOC and GAPDH (Figure 12B). Next, we tested MC and GA’s individual and combined effects in the complex tumouroid assays with both bulk and cancer stem cell populations. This combination therapy was efficient on cervical cancer cells and other tumour cells (Figure 13). The floating cells emerging from this treatment were dead, and even when these pooled floating cells were replated in conducive conditions, no growth was observed for two months. The outcome of this study is summarized in Figure 14. ## 4. Discussion High glucose (HG) is a significant hallmark of tumour progression. However, targeting known glucose signalling, glucose transporters and metabolic enzymes in cancer has been a concern, as it adversely impacts normal physiology. Since high glucose is also an osmotic stressor to which cancer cells intriguingly adapt, we were enthusiastic about teasing out the most prominent “hyper osmolarity adaptative mechanism” with the hope that targeting it may selectively promote the death of tumour cells without affecting the normal cells [8,32,33]. Due to interstitial fluid flow, which bathes the entire tumour tissue, high glucose can even be carried to the hypoxic interiors and exert its hyper-osmotic challenge. Therefore, by molecularly dissecting how tumour cells reshape their gene signatures and adjust to the high-glucose hyper-osmotic stress and oxygen-tension-associated dual biophysical challenge, we were motivated that we may be able to attenuate the mechanoadaptation therapeutically and consequently induce selective vulnerability to death. In a cervical cancer cell model, we report that high glucose and its osmotic mimetics, the mannitol and l-glucose at similar concentrations, promote comparable proliferation effects in normoxia. However, this effect is attenuated in an acutely hypoxic microenvironment, and cells enter dormancy due to mitotic arrest. The mechanism associated with this effect was related to hypoxia-induced excessive accumulation of SUMO2-modified pH3(Ser10) in the cytoplasm, compromising its availability for mitosis progression. Excessive phosphorylation of histone H3 and its variant H3.3 at the Ser 10 position has been associated with aggressive tumours, and it is a known master driver of mitosis [66,67]. Stalling of dynamic mitotic rounding, due to less chromosomal association of SUMOylated pH3(Ser10) in hypoxia, could inhibit tumour cells from balancing their cortical tension to osmo-adaptive values and thereby cause the death of mitotic cells that are under arrest for long time [68,69,70]. SUMO conjugation and substrate stabilization are regulated by the sentrin-specific proteases family of proteins, which has both redundant and unique functions [50]. Acute hypoxia was found to trigger SENP1 and SENP7 isoforms, but SENP1 was identified to be pro-survival. SENP1 higher levels have been associated with tumorigenesis [53,54,55,70]. We found that the over-activity of SENP7 promotes differential processing of SUMOylated pH3(Ser10), which shortens the SUMO2 chain bound to the substrate, impacting pH3(Ser10) solubility. Further, the liberated SUMO2 residues, in effect, likely bind to the alternative sites, rendering pH3(Ser10) accumulation in the cytoplasm with the LLPS-like M-phase phenotype. Induction of mitotic arrest by various anti-cancer drugs has been attractive for drug development because it controls metastasis and circulating tumour cells [68,69,71]. However, we find that such mitotically arrested, adherent and anchorage-independent phenotypes can relapse upon arrival of conducive conditions. Therefore, not only mitotic arrest but attenuation of residual survival cell signalling needs to be targeted simultaneously. However, the first step is to render the tumour dormant for the ‘kill while they sleep’ approach. Based on our observation, we could choose between two routes: the first was to upregulate SENP7 activity in normoxia to induce the LLPS phenotype and promote G2/M arrest. Alternatively, we could deplete SENP1, as our observation suggests that this could lead to massive abnormal accumulation of SUMOylated pH3(Ser10) due to lack of proteasomal degradation. Due to the availability of the natural SENP1 inhibitor Momordin Ic, we chose the second option. Momordin Ic treatment led to excessive accumulation of SUMOylated pH3(Ser10) in the cytoplasm akin to a hypoxic condition and induced dormancy in the normoxic HG-HO conditions. Clinical samples from cervical and other cancer also showed high cytoplasmic accumulation of SUMOylated pH3(Ser10) in the acutely hypoxic zones (identified by high HIF1a expression). Thus, the study suggests that Momordin Ic could effectively hijack hyperglycaemic–hyper-osmotic mechanoadaptation, thereby inducing therapeutic vulnerabilities to cell death. In order to further target the minimal residual growth generated in this treatment, we took the rational approach of delving deeper into the other phytochemicals that are enriched in plants from which MC is extracted and scored Gallic Acid for multiple reasons. Gallic Acid is known for its anti-cancer effects [62,63,64,65]. It can attenuate almost all hallmarks of cancer, such as inflammation, redox imbalances, hyperglycaemia, hyperlipidaemia, extracellular matrix stiffness, immune evasion, etc. Momordin Ic and Gallic Acid are already independently tested in animal models of tumours for safety and efficacy [37,38]. Therefore, plant extracts enriched in Momordin Ic and Gallic Acid, such as Momordica charantia, will be tested in further studies [72,73,74]. Momordica charantia is in Phase II clinical trials in modern medical practice (https://clinicaltrials.gov/ct2/show/results/NCT02397447, accessed on 21 February 2023); https://go.drugbank.com/drugs/DB14265, accessed on 21 February 2023). It is already a certified nutraceutical-based Ayurvedic medicine, a form of a medical practice prevalent in India and Asia. In addition, Gallic Acid is also enriched in other ayurvedic medicines such as Amalakirasayana and Arjunarishta [75,76,77]. ## 5. Conclusions In summary, we find that hyper-osmotic stress associated with high glucose concentrations drives tumour cell proliferation under normoxia, but the effect is attenuated under hypoxia due to G2/M-phase arrest. Mitotic-phase arrested tumour cells, in acute hypoxia, show extensive cytoplasmic aggregates of SUMO2 modified pH3(Ser10) due to liquid–liquid phase separation (LLPS). We validated this phenotype in high HIF1a regions of cancer tissues of different organs of origin. The high levels of SENP7 in acute hypoxia likely dismantle the polySUMO2 chain from pH3(Ser10), which promotes free and accumulated SUMO2 individual residues to bind to pH3(Ser10) at abnormal sites. This aberrant processing impacts pH3(Ser10) cytoplasmic solubility, generates the LLPS phenotype and thereby restricts its critical functions as a mitotic driver, causing tumour dormancy. Momordin Ic (MC), an SENP1 inhibitor, enhances SUMO2-pH3(Ser10) cytoplasmic accumulation in normoxic normal, high-glucose and hyper-osmotic conditions, thereby causing M-phase arrest and death of tumour cells. The combination of Momordin Ic with Gallic Acid (GA), both enriched in the extract of Momordica charantia (under Phase II clinical trials, available for sale under traditional medicine systems), significantly eliminates both normoxic active and hypoxic dormant tumour cells; therefore, this represents a promising lead in anti-cancer drug development. As high-grade patient tissues show the cytoplasmic accumulation of SUMO2-pH3(Ser10) in the HIF1a high areas, we propose SUMO2, pH3(Ser10) and HIF1a co-immunodetection as multiplexed panel biomarkers for prognosis and for predicting the probability of developing metastasis and relapse. ## References 1. Sung H., Ferlay J., Siegel R.L., Laversanne M., Soerjomataram I., Jemal A., Bray F.. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. (2021) 71 209-249. DOI: 10.3322/caac.21660 2. Crucitta S., Cucchiara F., Mathijssen R., Mateo J., Jager A., Joosse A., Passaro A., Attili I., Petrini I., van Schaik R.. Treatment-driven tumour heterogeneity and drug resistance: Lessons from solid tumours. Cancer Treat. Rev. (2022) 104 102340. DOI: 10.1016/j.ctrv.2022.102340 3. Liu Y.P., Zheng C.C., Huang Y.N., He M.L., Xu W.W., Li B.. Molecular mechanisms of chemo- and radiotherapy resistance and the potential implications for cancer treatment. MedComm (2021) 2 315-340. DOI: 10.1002/mco2.55 4. Kumar S., Kushwaha P.P., Gupta S.. Emerging targets in cancer drug resistance. Cancer Drug Resist. (2019) 2 161-177. DOI: 10.20517/cdr.2018.27 5. Rohani N., Hao L., Alexis M.S., Joughin B.A., Krismer K., Moufarrej M.N., Soltis A.R., Lauffenburger D.A., Yaffe M.B., Burge C.B.. Acidification of Tumor at Stromal Boundaries Drives Transcriptome Alterations Associated with Aggressive Phenotypes. Cancer Res. (2019) 79 1952-1966. DOI: 10.1158/0008-5472.CAN-18-1604 6. Wirtz D., Konstantopoulos K., Searson P.C.. The physics of cancer: The role of physical interactions and mechanical forces in metastasis. Nat. Rev. Cancer (2011) 11 512-522. DOI: 10.1038/nrc3080 7. De Felice D., Alaimo A.. Mechanosensitive Piezo Channels in Cancer: Focus on altered Calcium Signaling in Cancer Cells and in Tumor Progression. Cancers (2020) 12. DOI: 10.3390/cancers12071780 8. Brocker C., Thompson D.C., Vasiliou V.. The role of hyperosmotic stress in inflammation and disease. Biomol. Concepts (2012) 3 345-364. DOI: 10.1515/bmc-2012-0001 9. Chang S.F., Chang C.A., Lee D.Y., Lee P.L., Yeh Y.M., Yeh C.R., Cheng C.K., Chien S., Chiu J.J.. Tumor cell cycle arrest induced by shear stress: Roles of integrins and Smad. Proc. Natl. Acad. Sci. USA (2008) 105 3927-3932. DOI: 10.1073/pnas.0712353105 10. Northcott J.M., Dean I.S., Mouw J.K., Weaver V.M.. Feeling Stress: The Mechanics of Cancer Progression and Aggression. Front. Cell Dev. Biol. (2018) 6 17. DOI: 10.3389/fcell.2018.00017 11. Pickup M.W., Mouw J.K., Weaver V.M.. The extracellular matrix modulates the hallmarks of cancer. EMBO Rep. (2014) 15 1243-1253. DOI: 10.15252/embr.201439246 12. John S., Gayathri K.G., Krishna A.P., Mishra R.. Neurotherapeutic implications of sense and respond strategies generated by astrocytes and astrocytic tumours to combat pH mechanical stress. Neuropathol. Appl. Neurobiol. (2022) 48 e12774. DOI: 10.1111/nan.12774 13. John S., Sivakumar K.C., Mishra R.. Extracellular Proton Concentrations Impacts LN229 Glioblastoma Tumor Cell Fate via Differential Modulation of Surface Lipids. Front. Oncol. (2017) 7 20. DOI: 10.3389/fonc.2017.00020 14. Song S., Zhang Y., Ding T., Ji N., Zhao H.. The Dual Role of Macropinocytosis in Cancers: Promoting Growth and Inducing Methuosis to Participate in Anticancer Therapies as Targets. Front. Oncol. (2021) 10 570108. DOI: 10.3389/fonc.2020.570108 15. Weaver V.M.. Cell and tissue mechanics: The new cell biology frontier. Mol. Biol. Cell. (2017) 28 1815-1818. DOI: 10.1091/mbc.e17-05-0320 16. Cheng B., Lin M., Huang G., Li Y., Ji B., Genin G.M., Deshpande V.S., Lu T.J., Xu F.. Cellular mechanosensing of the biophysical microenvironment: A review of mathematical models of biophysical regulation of cell responses. Phys. Life Rev. (2017) 22–23 88-119. DOI: 10.1016/j.plrev.2017.06.016 17. Spill F., Bakal C., Mak M.. Mechanical and Systems Biology of Cancer. Comput. Struct. Biotechnol. J. (2018) 16 237-245. DOI: 10.1016/j.csbj.2018.07.002 18. Jain R.K., Martin J.D., Stylianopoulos T.. The role of mechanical forces in tumor growth and therapy. Annu. Rev. Biomed. Eng. (2014) 16 321-346. DOI: 10.1146/annurev-bioeng-071813-105259 19. Munson J.M., Shieh A.C.. Interstitial fluid flow in cancer: Implications for disease progression and treatment. Cancer Manag. Res. (2014) 6 317-328. DOI: 10.2147/CMAR.S65444 20. Kang H., Wu Q., Sun A., Liu X., Fan Y., Deng X.. Cancer Cell Glycocalyx and Its Significance in Cancer Progression. Int. J. Mol. Sci. (2018) 19. DOI: 10.3390/ijms19092484 21. Goldman J.W., Mendenhall M.A., Rettinger S.R.. Hyperglycemia Associated With Targeted Oncologic Treatment: Mechanisms and Management. Oncologist (2016) 21 1326-1336. DOI: 10.1634/theoncologist.2015-0519 22. Duan W., Shen X., Lei J., Xu Q., Yu Y., Li R., Wu E., Ma Q.. Hyperglycemia, a neglected factor during cancer progression. Biomed. Res. Int. (2014) 2014 461917. DOI: 10.1155/2014/461917 23. Derr R.L., Ye X., Islas M.U., Desideri S., Saudek C.D., Grossman S.A.. Association between hyperglycemia and survival in patients with newly diagnosed glioblastoma. J. Clin. Oncol. (2009) 27 1082-1086. DOI: 10.1200/JCO.2008.19.1098 24. Xue F., Lin L.L., Dehdashti F., Miller T.R., Siegel B.A., Grigsby P.W.. F-18 fluorodeoxyglucose uptake in primary cervical cancer as an indicator of prognosis after radiation therapy. Gynecol. Oncol. (2006) 101 147-151. DOI: 10.1016/j.ygyno.2005.10.005 25. Novikova L., Smirnova I.V., Rawal S., Dotson A.L., Benedict S.H., Stehno-Bittel L.. Variations in rodent models of type 1 diabetes: Islet morphology. J. Diabetes Res. (2013) 2013 965832. DOI: 10.1155/2013/965832 26. Pietuch A., Brückner B.R., Janshoff A.. Membrane tension homeostasis of epithelial cells through surface area regulation in response to osmotic stress. Biochim. Biophys. Acta (2013) 1833 712-722. DOI: 10.1016/j.bbamcr.2012.11.006 27. Cohen B.E.. Membrane Thickness as a Key Factor Contributing to the Activation of Osmosensors and Essential Ras Signaling Pathways. Front. Cell Dev. Biol. (2018) 6 76. DOI: 10.3389/fcell.2018.00076 28. Taubenberger A.V., Baum B., Matthews H.K.. The Mechanics of Mitotic Cell Rounding. Front. Cell Dev. Biol. (2020) 8 687. DOI: 10.3389/fcell.2020.00687 29. Stewart M.P., Helenius J., Toyoda Y., Ramanathan S.P., Muller D.J., Hyman A.A.. Hydrostatic pressure and the actomyosin cortex drive mitotic cell rounding. Nature (2011) 469 226-230. DOI: 10.1038/nature09642 30. Fischer-Friedrich E., Hyman A.A., Jülicher F., Müller D.J., Helenius J.. Quantification of surface tension and internal pressure generated by single mitotic cells. Sci. Rep. (2014) 4 6213. DOI: 10.1038/srep06213 31. Toyoda Y., Cattin C.J., Stewart M.P., Poser I., Theis M., Kurzchalia T.V., Buchholz F., Hyman A.A., Müller D.J.. Genome-scale single-cell mechanical phenotyping reveals disease-related genes involved in mitotic rounding. Nat. Commun. (2017) 8 1266. DOI: 10.1038/s41467-017-01147-6 32. Madonna R., Giovannelli G., Confalone P., Renna F.V., Geng Y.J., De Caterina R.. High glucose-induced hyperosmolarity contributes to COX-2 expression and angiogenesis: Implications for diabetic retinopathy. Cardiovasc. Diabetol. (2016) 15 18. DOI: 10.1186/s12933-016-0342-4 33. Russo V.C., Higgins S., Werther G.A., Cameron F.J.. Effects of fluctuating glucose levels on neuronal cells in vitro. Neurochem. Res. (2012) 37 1768-1782. DOI: 10.1007/s11064-012-0789-y 34. John S., Bacoside A.. Induces Tumor Cell Death in Human Glioblastoma Cell Lines through Catastrophic Macropinocytosis. Front. Mol. Neurosci. (2017) 10. DOI: 10.3389/fnmol.2017.00171 35. Ao Q., Su W., Guo S., Cai L., Huang L.. SENP1 desensitizes hypoxic ovarian cancer cells to cisplatin by up-regulating HIF-1α. Sci. Rep. (2015) 5 16396. DOI: 10.1038/srep16396 36. Schindelin J., Arganda-Carreras I., Frise E., Kaynig V., Longair M., Pietzsch T., Preibisch S., Rueden C., Saalfeld S., Schmid B.. Fiji: An open-source platform for biological-image analysis. Nat. Methods (2012) 9 676-682. DOI: 10.1038/nmeth.2019 37. Wu J., Lei H., Zhang J., Chen X., Tang C., Wang W., Xu H., Xiao W., Gu W., Wu Y.. Momordin Ic, a new natural SENP1 inhibitor, inhibits prostate cancer cell proliferation. Oncotarget (2016) 7 58995-59005. DOI: 10.18632/oncotarget.10636 38. Wang D., Bao B.. Gallic Acid Impedes Non-Small Cell Lung Cancer Progression via Suppression of EGFR-Dependent CARM1-PELP1 Complex. Drug Des. Dev. Ther. (2020) 14 1583-1592. DOI: 10.2147/DDDT.S228123 39. Piccinini F., Tesei A., Arienti C., Bevilacqua A.. Cancer multicellular spheroids: Volume assessment from a single 2D projection. Comput. Methods Programs Biomed. (2015) 118 95-106. DOI: 10.1016/j.cmpb.2014.12.003 40. Piccinini F., Tesei A., Bevilacqua A.. Single-image based methods used for non-invasive volume estimation of cancer spheroids: A practical assessing approach based on entry-level equipment. Comput. Methods Programs Biomed. (2016) 135 51-60. DOI: 10.1016/j.cmpb.2016.07.024 41. Luzhna L., Kathiria P., Kovalchuk O.. Micronuclei in genotoxicity assessment: From genetics to epigenetics and beyond. Front. Genet. (2013) 4 131. DOI: 10.3389/fgene.2013.00131 42. Eifler K., Vertegaal A.C.O.. SUMOylation-Mediated Regulation of Cell Cycle Progression and Cancer. Trends Biochem. Sci. (2015) 40 779-793. DOI: 10.1016/j.tibs.2015.09.006 43. Hendriks I.A., D’Souza R.C., Chang J.G., Mann M., Vertegaal A.C.. System-wide identification of wild-type SUMO-2 conjugation sites. Nat. Commun. (2015) 6 7289. DOI: 10.1038/ncomms8289 44. Hendriks I.A., Lyon D., Young C., Jensen L.J., Vertegaal A.C., Nielsen M.L.. Site-specific mapping of the human SUMO proteome reveals co-modification with phosphorylation. Nat. Struct. Mol. Biol. (2017) 24 325-336. DOI: 10.1038/nsmb.3366 45. Prigent C., Dimitrov S.. Phosphorylation of serine 10 in histone H3, what for?. J. Cell Sci. (2003) 116 (Pt 18) 3677-3685. DOI: 10.1242/jcs.00735 46. Li D.W., Yang Q., Chen J.T., Zhou H., Liu R.M., Huang X.T.. Dynamic distribution of Ser-10 phosphorylated histone H3 in cytoplasm of MCF-7 and CHO cells during mitosis. Cell Res. (2005) 15 120-126. DOI: 10.1038/sj.cr.7290276 47. Kim J.I., Choi K.U., Lee I.S., Choi Y.J., Kim W.T., Shin D.H., Kim K., Lee J.H., Kim J.Y., Sol M.Y.. Expression of hypoxic markers and their prognostic significance in soft tissue sarcoma. Oncol. Lett. (2015) 9 1699-1706. DOI: 10.3892/ol.2015.2914 48. Burri P., Djonov V., Aebersold D.M., Lindel K., Studer U., Altermatt H.J., Mazzucchelli L., Greiner R.H., Gruber G.. Significant correlation of hypoxia-inducible factor-1alpha with treatment outcome in cervical cancer treated with radical radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. (2003) 56 494-501. DOI: 10.1016/S0360-3016(02)04579-0 49. Peng Z., Mizianty M.J., Xue B., Kurgan L., Uversky V.N.. More than just tails: Intrinsic disorder in histone proteins. Mol. Biosyst. (2012) 8 1886-1901. DOI: 10.1039/c2mb25102g 50. Lara-Ureña N., Jafari V., García-Domínguez M.. Cancer-Associated Dysregulation of Sumo Regulators: Proteases and Ligases. Int. J. Mol. Sci. (2022) 23. DOI: 10.3390/ijms23148012 51. Hoppe-Seyler K., Bossler F., Lohrey C., Bulkescher J., Rösl F., Jansen L., Mayer A., Vaupel P., Dürst M., Hoppe-Seyler F.. Induction of dormancy in hypoxic human papillomavirus-positive cancer cells. Proc. Natl. Acad. Sci. USA (2017) 114 E990-E998. DOI: 10.1073/pnas.1615758114 52. Butturini E., Carcereri de Prati A., Boriero D., Mariotto S.. Tumor Dormancy and Interplay with Hypoxic Tumor Microenvironment. Int. J. Mol. Sci. (2019) 20. DOI: 10.3390/ijms20174305 53. Lee M.H., Sung K., Beebe D., Huang W., Shapiro D., Miyamoto S., Abel E.J.. The SUMO protease SENP1 promotes aggressive behaviors of high HIF2α expressing renal cell carcinoma cells. Oncogenesis (2022) 11 65. DOI: 10.1038/s41389-022-00440-4 54. Zhu S., Hu J., Cui Y., Liang S., Gao X., Zhang J., Jia W.. Knockdown of SENP1 inhibits HIF-1α SUMOylation and suppresses oncogenic CCNE1 in Wilms tumor. Mol. Ther. Oncolytics (2021) 23 355-366. DOI: 10.1016/j.omto.2021.07.007 55. Zhang Q.S., Zhang M., Huang X.J., Liu X.J., Li W.P.. Downregulation of SENP1 inhibits cell proliferation, migration and promotes apoptosis in human glioma cells. Oncol. Lett. (2016) 12 217-221. DOI: 10.3892/ol.2016.4558 56. Li Y., De Bolòs A., Amador V., Reverter D.. Structural Basis for the SUMO2 Isoform Specificity of SENP7. J. Mol. Biol. (2022) 434 167875. DOI: 10.1016/j.jmb.2022.167875 57. Martin E.W., Holehouse A.S.. Intrinsically disordered protein regions and phase separation: Sequence determinants of assembly or lack thereof. Emerg. Top. Life Sci. (2020) 4 307-329. DOI: 10.1042/ETLS20190164 58. Gâtel P., Brockly F., Reynes C., Pastore M., Hicheri Y., Cartron G., Piechaczyk M., Bossis G.. Ubiquitin and SUMO conjugation as biomarkers of acute myeloid leukemias response to chemotherapies. Life Sci. Alliance (2020) 3 e201900577. DOI: 10.26508/lsa.201900577 59. Chang H.M., Yeh E.T.H.. SUMO: From Bench to Bedside. Physiol. Rev. (2020) 100 1599-1619. DOI: 10.1152/physrev.00025.2019 60. Jia S., Shen M., Zhang F., Xie J.. Recent Advances in Momordica charantia: Functional Components and Biological Activities. Int. J. Mol. Sci. (2017) 18. DOI: 10.3390/ijms18122555 61. Bortolotti M., Mercatelli D., Polito L.. Front. Pharmacol. (2019) 10 486. DOI: 10.3389/fphar.2019.00486 62. Bai J., Zhang Y., Tang C., Hou Y., Ai X., Chen X., Zhang Y., Wang X., Meng X.. Gallic acid: Pharmacological activities and molecular mechanisms involved in inflammation-related diseases. Biomed. Pharmacother. (2021) 133 110985. DOI: 10.1016/j.biopha.2020.110985 63. Tuli H.S., Mistry H., Kaur G., Aggarwal D., Garg V.K., Mittal S., Yerer M.B., Sak K., Khan M.A.. Gallic Acid: A Dietary Polyphenol that Exhibits Anti-neoplastic Activities by Modulating Multiple Oncogenic Targets. Anticancer Agents Med. Chem. (2022) 22 499-514. DOI: 10.2174/1871520621666211119085834 64. Kahkeshani N., Farzaei F., Fotouhi M., Alavi S.S., Bahramsoltani R., Naseri R., Momtaz S., Abbasabadi Z., Rahimi R., Farzaei M.H.. Pharmacological effects of gallic acid in health and diseases: A mechanistic review. Iran J. Basic Med. Sci. (2019) 22 225-237. DOI: 10.22038/ijbms.2019.32806.7897 65. Aborehab N.M., Osama N.. Effect of Gallic acid in potentiating chemotherapeutic effect of Paclitaxel in HeLa cervical cancer cells. Cancer Cell Int. (2019) 19 154. DOI: 10.1186/s12935-019-0868-0 66. Cui X., Harada S., Shen D., Siegal G.P., Wei S.. The Utility of Phosphohistone H3 in Breast Cancer Grading. Appl. Immunohistochem. Mol. Morphol. (2015) 23 689-695. DOI: 10.1097/PAI.0000000000000137 67. Hans F., Dimitrov S.. Histone H3 phosphorylation and cell division. Oncogene (2001) 20 3021-3027. DOI: 10.1038/sj.onc.1204326 68. Burgess A., Rasouli M., Rogers S.. Stressing mitosis to death. Front. Oncol. (2014) 4 140. DOI: 10.3389/fonc.2014.00140 69. Xia W., Spector S., Hardy L., Zhao S., Saluk A., Alemane L., Spector N.L.. Tumor selective G2/M cell cycle arrest and apoptosis of epithelial and hematological malignancies by BBL22, a benzazepine. Proc. Natl. Acad. Sci. USA (2000) 97 7494-7499. DOI: 10.1073/pnas.97.13.7494 70. Cui C.P., Wong C.C., Kai A.K., Ho D.W., Lau E.Y., Tsui Y.M., Chan L.K., Cheung T.T., Chok K.S., Chan A.C.Y.. SENP1 promotes hypoxia-induced cancer stemness by HIF-1α deSUMOylation and SENP1/HIF-1α positive feedback loop. Gut (2017) 66 2149-2159. DOI: 10.1136/gutjnl-2016-313264 71. Visconti R., Della Monica R., Grieco D.. Cell cycle checkpoint in cancer: A therapeutically targetable double-edged sword. J. Exp. Clin. Cancer Res. (2016) 35 153. DOI: 10.1186/s13046-016-0433-9 72. Liu Z., Gong J., Huang W., Lu F., Dong H.. The Effect of. Evid. Based Complement. Alternat. Med. (2021) 2021 3796265. DOI: 10.1155/2021/3796265 73. Peter E.L., Deyno S., Mtewa A., Kasali F.M., Nagendrappa P.B., Sesaazi D., Tolo C.U., Ogwang P.E.. Safety and efficacy of Momordica charantia Linnaeus in pre-diabetes and type 2 diabetes mellitus patients: A systematic review and meta-analysis protocol. Syst. Rev. (2018) 7 192. DOI: 10.1186/s13643-018-0847-x 74. Rauniyar G.P., Sinha R., Chapagain K., Maskey R., Pandey D.R.. Effects of Momordica Charantia (Karela/bitterguord) in Type 2 Diabetic Patients Taking Allopathic Drugs: A pilot study. Kathmandu Univ. Med. J (2021) 19 143-147. DOI: 10.3126/kumj.v19i2.49655 75. Kumar V., Aneesh K.A., Kshemada K., Ajith K.G.S., Binil R.S.S., Deora N., Sanjay G., Jaleel A., Muraleedharan T.S., Anandan E.M.. Amalaki rasayana, a traditional Indian drug enhances cardiac mitochondrial and contractile functions and improves cardiac function in rats with hypertrophy. Sci. Rep. (2017) 7 8588. DOI: 10.1038/s41598-017-09225-x 76. Shengule S.A., Mishra S., Joshi K., Apte K., Patil D., Kale P., Shah T., Deshpande M., Puranik A.. Anti-hyperglycemic and anti-hyperlipidaemic effect of Arjunarishta in high-fat fed animals. J. Ayurveda Integr. Med. (2018) 9 45-52. DOI: 10.1016/j.jaim.2017.07.004 77. Tiwari P., Patel R.K.. Quantification of Gallic acid and Ellagic acid in Arjunarishta by validated HPTLC densitometry. Int. J. Pharm. Sci. Res. (2012) 3 2215. DOI: 10.13040/IJPSR.0975-8232.3(7).2215-23
--- title: Sociodemographic Variables and Body Mass Index Associated with the Risk of Eating Disorders in Spanish University Students authors: - María-Camino Escolar-Llamazares - María-Ángeles Martínez-Martín - María-Begoña Medina-Gómez - María-Yolanda González-Alonso - Elvira Mercado-Val - Fernando Lara-Ortega journal: European Journal of Investigation in Health, Psychology and Education year: 2023 pmcid: PMC10047306 doi: 10.3390/ejihpe13030046 license: CC BY 4.0 --- # Sociodemographic Variables and Body Mass Index Associated with the Risk of Eating Disorders in Spanish University Students ## Abstract Background: The passage through university is a complex experience that can heighten personal susceptibility to eating disorders. The objective of this research is to determine how gender, age, course, educational faculty, and body mass index (BMI) can influence the risk of eating disorders among university students. Method: A transversal and descriptive study is conducted with a sample of 516 Spanish students ($57.2\%$ female, $42.8\%$ male; Mage = 21.7, SDage = 4.1) following 26 university degrees. The Inventory Eating Disorder-Reference criterion (EDI-3-RF) was administered to the students. Contingency tables were used between categorical variables with the chi-squared statistic, at a significance level of $p \leq 0.05.$ The Student t-test was used for two independent samples and a one-way ANOVA test with the post hoc Bonferroni test for more than two groups. Pearson’s correlation and a simple linear regression analysis were used to analyze the relationship between the variables in its quantitative version. Results: It was found that the female students enrolled in the second year presented a greater obsession with thinness and body dissatisfaction ($$p \leq 0.029$$; $d = 0.338$); the male students practiced more physical exercise to control their weight ($$p \leq 0.003$$); and that students under the age of twenty ($p \leq 0.010$; $d = 0.584$) and students from both the Health ($$p \leq 0.0.13$$) and Law ($$p \leq 0.021$$) educational faculties showed greater bulimic behavior ($d = 0.070$). More females are underweight ($z = 2.8$), and more men are overweight ($z = 2.4$). Normal-weight students scored significantly higher in thinness obsession ($$p \leq 0.033$$). Overweight students scored significantly higher on thinness obsession ($p \leq 0.001$) and body dissatisfaction ($p \leq 0.001$). Obese students scored significantly higher on body dissatisfaction ($$p \leq 0.04$$). Conclusions: The data obtained in this study, reinforce the hypothesis that the female gender, at an age within the limits of early adolescence, in the first year of the degree courses, with specific university qualifications, and a high BMI constituted factors that could provoke an eating disorder. Consequently, it is necessary to implement preventive measures adapted to the circumstances of each university student. ## 1. Introduction The passage through university for many young people represents a complex experience, in which they will encounter stressors that heighten their susceptibility to succumbing to mental health disorders of various kinds [1,2,3]. The scientific evidence suggests that both the particular characteristics of the life cycle and the demands of university life (heavier academic workload, adaptation to changing social solidarity and networks, greater autonomy, increased responsibility, vocational and academic failure, moving home, living alone) increase the probability of emergent emotional difficulties that, if not properly treated, will metamorphose into clinical disorders [1,4,5]. Eating disorders (ED) are hardly unknown to university students who are, in fact, a high-risk population, due in part to all the changes associated with the start of university life [6,7,8]. ED are serious mental disorders, related to body dissatisfaction, excessive preoccupation with food, weight, body image and what it represents. Likewise, they produce important behavioral alterations, as a consequence of the attempt to control the body and weight [9,10,11]. EDs represent a public health problem, given their prevalence, severity, prolonged clinical course, tendency to chronification, need for pluri- and interdisciplinary treatment, and recurrent hospitalization [5,12,13]. The scientific literature has established that over $60\%$ of university students hold erroneous perceptions of their physical image, overestimating their body mass index (BMI) [14,15,16,17]. This body dissatisfaction accompanied by slimming diets constitutes one of the main risk factors of EDs [15,18,19,20,21,22]. Body dissatisfaction is characterized by the presence of value judgments about the body that do not match the actual characteristics [10,23,24]. Similarly, body dissatisfaction represents dissatisfaction with overall body shape and with the size of specific body parts (e.g., stomach, hips, thighs, and buttocks) that are of extraordinary concern to people with eating disorders [5,25,26]. Aspects such as feeling bloated after eating a normal amount of food are common features among people who feel dissatisfied with their own bodies [11,27,28]. Given that body dissatisfaction is endemic among young women in Western cultures it cannot be concluded that this construct alone causes EDs. However, it is a risk factor responsible for the initiation and maintenance of extreme weight control behaviors that lead to the development of EDs in those who are vulnerable [5,11,28]. On the other hand, weight assessment is important in EDs because it provides information about the context in which the symptoms occur. Moreover, it is necessary to understand subjective psychological distress related to figure and appearance [11,29,30]. Studies conducted on body dissatisfaction consistently show that it is positively correlated with body weight and body mass index (BMI). Similarly, BMI correlates positively with obsession with thinness and problematic eating patterns (binge eating) in response to negative emotional states. Finally, BMI is particularly important since a very low value of this index is a very serious sign of an eating disorder or physical illness [5,11,29]. In relation to gender, epidemiological studies have consistently made clear that an ED is more common among women than among men [15,18,22,31]. In this sense, Martínez-González, L. et al. [ 32] found a risk prevalence of $19\%$ in a university population, which was higher among women. However, the impact of one ED or another has also been shown to affect the male population [15,31,33]. Although the masculine and feminine risk profiles for an ED differ, risk factors also arise due to peer pressures surrounding an ideal of beauty. In the case of men, an ideal musculature can never be an ideal of slimness. However, it also exposes men to ED risk behaviors, such as: the use of steroids, high consumption of laxatives, desire for greater muscular mass, dietary complements, diet and excessive sports activity [7,34,35,36]. Authors such as Sepúlveda et al. [ 37] and Chin et al. [ 10] signaled some significative gender-related differences in risk behavior. For example, vomiting was present in $9.6\%$ of men as opposed to $16\%$ of women and the use of laxatives to control body weight stood at $10.6\%$ among men and $14.5\%$ among women [38]. Other studies, such as the one by Lameiras et al. [ 39] with university students from the Autonomous Community of Galicia (Spain), affirmed that women showed higher levels of concern, due to weight and body image, and they resorted to diets to reach the ideal weight. Franco et al. [ 15] found that the three risk behaviors reported to a greater extent among women were compensatory behaviors, the use of slimming products, and binge eating, and among the men, compensatory behaviors, binge eating, and exercise to burn off calories. In relation to age, González-Carrascosa et al. [ 7], Daly and Costigan [3] and Schilder et al. [ 40], among others, pointed out that the start of these disorders is habitually at prepubescent ages and early adolescence, with a higher percentage prevalence among young adults, who form the majority of the university population. An epidemiological study carried out by the Asociación contra la Anorexia y la Bulimia (ACAB) [Association against Anorexia and Bulimia] and Andersen [41] with universities (18–25 years old) from the Catalan Autonomous Community (Spain), confirmed that $11.48\%$ presented a high risk of suffering from an ED, while $6.38\%$ could have been suffering from it at that time. A significant proportion of university students therefore appear to be at-risk of developing an ED in the future. As much is also confirmed in the results of work completed at the Autonomous University of Madrid by Sepúlveda et al. [ 37] in which $14.9\%$ of men and $20.8\%$ of women presented a high risk of suffering an ED. Although social pressures tend to be more intense at the adolescent stage, the cult of the body is maintained throughout all stages of life [42]. EDs are therefore increasingly present at early ages and are at the same time maintained at more advanced ages [7,18,43]. In studying the influence of the year in which students are enrolled, the scientific literature points to the importance of the first years of university. The first year is an especially critical period for the onset of pathological or disordered eating patterns. The increase in independence and responsibilities, as well as questions over personal identity all contributed in part to the pathological eating patterns [44,45]. Increased weight, by more than two-thirds, occurred among first-year students, during their first semester [46,47,48]. Body mass tended to be added, on average, by slightly over $1\%$ [44,49]. Other studies have reported weight gain and increased body mass among students throughout the different university courses [44]. With regard to the selection of university qualification or degree, it could be influenced by pre-existing eating disorders. Personality traits, motivations, and lifestyles correlate with the choice of a future profession [50,51]. In other words, there is usually a higher impact on risk factors among young university students from degrees within the area of health, such as nutrition, dietetics, physical education, nursing and medicine, where physical appearance and concern for health are very important [22,52]. Bo et al. [ 50] pointed out that the students enrolled on health sciences courses such nutrition showed a high prevalence of EDs (specifically, one fifth), and those with pre-existing pathological eating patterns were especially inclined to enroll on health sciences courses. A European survey reported that $12.8\%$ of students of nutrition presented an earlier or a current ED, such as anorexia nervosa, bulimia nervosa, and binge-eating disorder [53]. Along these same lines, Rocks et al. [ 54] and Toral et al. [ 33] found that the students of nutrition showed a double prevalence of the psychological characteristics and behavior often associated with EDs in comparison with students from other degrees. Other studies, such as the one by Peña Salgado et al. [ 21] showed that student susceptibility to eating disorders was higher among the students enrolled on business administration, followed by law. In turn, Cancela and Ayán [55] found that physical inactivity and eating disorders had a significant prevalence among the students enrolled on primary school teaching and nursing. In relation to students of primary school teaching, some works showed that they have distorted attitudes and knowledge of the etiology of obesity, balanced nutrition and dietary regimes, and that the women especially presented risk factors such as inappropriate weight control techniques (use of laxatives and vomiting) [34,56]. Despite the abundance of the literature, some authors [27,34,57] point out the importance of to carry out more research to better understand the risk of suffering DEs in the university population. As noted out by [30] eating disorders are more prevalent among university students compared to the general population. In this regard,, the objective of this investigation will be centered on determining whether, in Spanish university students, variables such as gender, age, course, educational faculty, and BMI are associated with a higher risk of suffering from eating disorders (ED). ## 2. Materials and Methods A transversal, descriptive survey of a probabilistic sample population was designed [58]. Different data sets were then compiled for the study: the sample characteristics, the instrument, the procedure, and the analysis of the results. ## 2.1. Participants The investigation took place at the University of Burgos (Spain). The sample was formed of 561 students taken from a population of 6,277 students. Stratified sampling by degree and faculty resulted in the selection of the participants from among the 26 degrees taught at the nine faculties of the university. The sample was calculated using the formula for finite stratified samples, at a confidence level of $95.6\%$ and with an error margin of +/−$4\%$ [59]. Convenience sampling was used to group the individual respondents to the survey. In all, $42.8\%$ of the sample were men ($$n = 240$$) and $57.2\%$ were women ($$n = 276$$), with an average age of 21.7 (SD = 4.1). Regarding their origin, $75\%$ of the students were from Burgos and the province, $12\%$ from the other Spanish provinces and $13\%$ were from other countries (China, Mexico, and France). In relation to the course in which the students were enrolled, $37.1\%$ were enrolled in the fourth year of studies, $28.5\%$ in the second year, $21.7\%$ in the third year, and $12.7\%$ in the first year. Frequencies and percentages of students by educational faculty are shown in Table 1. The 561 male and female students had average heights of 1.78 meters (SD = 0.06) and 1.64 meters (SD = 0.06), respectively. Their average actual weights were 74.51 kilos (SD = 9.98) and 58.61 kilos (SD = 8.81), respectively. The ideal weight of the male students was 73.29 kilos (SD = 7.80) (1.22 kilos less than their average actual weight), and the ideal weight of the female students was 55.11 kilos (SD = 6.25). The body mass indexes (BMI) of the men and the women were 23.28 (SD = 2.79) and 21.77 (SD = 2.94), respectively. In relation to the data on the BMI classification, 22 ($3.9\%$) students were underweight, 451 ($80.45\%$) were normal weight, 86 ($15.3\%$) were overweight (>22 and ≤25) and 2 ($0.4\%$) were obese. ## 2.2. Instrument Two instruments were designed: one, ad hoc, to collect sociodemographic data (gender, age, residence, educational faculty, degree, and course enrolment), and another standardized instrument, specifically, the Spanish version of the Eating Disorder Inventory-Reference criterion (EDI-3-RF) developed by Garner [11]. The EDI-3-RF consists of a brief self-administered questionnaire on the risk of developing an ED, based on concerns over food and feeding, body weight, stature, and the presence of extreme behaviors to control weight [11]. The inventory is formed of three scales of risk: 1. Obsession with thinness (OT): a scale with 7 items that measure obsession with thinness, worry over food, and an intense fear of gaining weight. This scale is a good predictor of the appearance of binge eating and the development of an ED. The range of direct scores is from 0 to 20, where 12 is the critical value (situation of real risk of an ED) [34]. 2. Bulimia (B): an 8-item scale used to evaluate problematic eating patterns (binges) as a response to negative emotional states, constituting a risk factor. The range of direct scores runs from 0 to 32, and the critical values are between 5 and 8, in accordance with the body mass index (BMI) of each individual. 3. Body Dissatisfaction (BD): includes 10 items that evaluate dissatisfaction with the general shape of the body and with the size of specific parts that that cause extraordinary concern among people with EDs (e.g., stomach, hips, thighs, and buttocks). The direct scores of the scale from 0 to 40 are on a qualitative range: 0–6 low, 7–27 medium, 28–40 high body dissatisfaction. In addition, the instrument includes socio-demographic questions, individual weight records, and five behavioral questions that examine the presence of extreme behaviors to control weight. In particular, they are: [1] Presence of binge eating: range of direct scores between 0–5 where, the critical score is between 2 and 5; [2] Induced vomiting or purges: range of direct scores between 0–5, where the critical score is between 1 and 5; [3] Use of laxatives: range of direct scores between 0–5, where the critical score is between 1 and 5; [4] Physical exercise as a means of losing or controlling weight: range of direct scores between 0–5, where the critical score is 5; [5] Weight loss of nine kilos or more during the last six months: a dichotomous Yes/No question where the critical score is Yes [11]. The responses, grouped into two categories to facilitate the analysis, were as follows: No Risk (when the response to each one of the five questions was outside the criteria of pathology) and Risk (when the responses to each of these five questions indicated a risk of suffering an ED). The psychometric data yielded values between 0.82 and 0.96 for internal consistency (Cronbach’s alpha) with the Spanish adaptation in Spanish clinical samples and between 0.64 and 0.92 in non-clinical samples [60]. Equally, García et al. [ 34] analyzed the internal consistency with Cronbach’s alpha, which yielded a value of 0.91, a reliability level that was considered excellent [61]. The value of internal consistency in the sample from the present study was 0.886 (Cronbach’s alpha). ## 2.3. Procedure Before the study commenced, the authorization of the Bioethics Committee of the University of Burgos and informed consent forms from all participants were obtained. The questionnaire was administered by researchers with previous training, the application of which was possible thanks to the disinterested collaboration of teachers from the different faculties at which the 26 degree courses were taught. The class selection criteria and therefore the selection of the students to be administered the questionnaires was random. One of the researchers visited the classrooms, gave information on data confidentiality, and requested the informed consent of the participants. Questionnaire administration time fluctuated between 15 and 30 min. ## 2.4. Data Analysis The data were processed with the statistical program SPSS (version 28). A descriptive analysis of the variables under study was performed using frequency tables and percentages. Likewise, contingency tables were used to observe the relations between categoric variables with the Chi-squared statistic, at a significance level of $p \leq 0.05.$ The Student t-test was used for two independent samples (with Levene’s test for equality of variances) and a one-way ANOVA test with the post hoc Bonferroni test for more than two groups. The effect size was calculated for each of the significant differences (d of Cohen). Gender, age, course, faculty and BMI constituted the independent variables. Two categories (man-woman) in relation to the variable gender were used for data collection. Three categories were used for the age variable: 20 years or less, 21-to-25 years, and 26 years or over. Four categories were used for the year of the course: 1. First year; 2. Second year; 3. Third year; 4. Fourth year. Nine categories were considered for the variable faculty (Table 1). Four categories were used for BMI: Underweight (≤18.5), Normal weight (>18 and ≤22), Overweight (>22 and ≤25) and Obesity (>25). The results of the EDI-3-RF on the three scales of risk (Obsession with Thinness -DT-, Bulimia -B- and Body Dissatisfaction -BD-) and the five behavioral questions (binge eating, vomiting, laxatives, physical exercise, loss of weight), constituted the dependent variables. Pearson correlation and simple linear regression analysis were used to analyze the relationship between the variables age and BMI in its quantitative version. ## 3. Results The results were drawn from the impact of the sociodemographic, academic and BMI as independent variables (IV) on the dependent variables (DV) previously described for the study. ## 3.1. Differences as a Function of Gender With regard to the values of the central tendency among women and the dispersion of the scores on the three scales of risk, the mean average both for Obsession with Thinness (OT) (Mwomen = 7.25 and SDwomen = 5.63; Mmen = 4.72 and SDmen = 4.40) and for Body Dissatisfaction (BD) (Mwomen = 11.61 and SDwomen = 8.11; Mmen = 7.83 and DTmen = 6.61) were, in both cases, higher and significantly different among women ($p \leq 0.000$). An effect size was obtained (Cohen d index) considered as moderate (0.500 and 0.510, respectively) as it was located between 0.50 ≤ d ≤ 0.79. So, in the present study there is a rather small risk of random chance-based differences between both groups [62]. On the Bulimia (B) scale, the two measures (men and women) were similar (Mwomen = 4.35 and SDwomen = 4.25; Mmen= 4.32 and SDmen = 4.16, $$p \leq 0.938$$). Contingency Tables were applied to examine the qualitative range of the Body Dissatisfaction (BD) scale and to analyze its relationship with gender (Table 2). A chi-square test revealed that there is a statistically significant relationship between gender and the qualitative range of the BD, χ2 [2] = 18.46, $p \leq 0.001.$ The corrected residue value reveals that significantly more men than women presented low Body Dissatisfaction ($z = 4.2$). Likewise, more women than men presented moderate Body Dissatisfaction ($z = 3.8$). In relation to the presence of risk factors -binge eating, vomiting, use of laxatives, physical exercise, and weight loss-, different chi-square tests revealed that there is only a statistically significant relationship between gender and the presence of risk for excess physical exercise, χ2 [1] = 8.79, $$p \leq 0.003.$$ The corrected residual value determined that significantly more men than women are at risk for excess physical exercise as a way to control their weight ($z = 3$) (Table 3). ## 3.2. Differences as a Function of Age Table 4 describes the mean scores and standard deviation obtained by the students according to their age on the OT, B and BD scales. An ANOVA test revealed that there was a significant difference in the scores obtained by the students only in the Bulimia scale, F[2,558] = 4.52, $$p \leq 0.011.$$ No significant differences were observed on the OT and BD scales as a function of age. The analysis of multiple comparisons after the Bonferroni statistical test was looking for significative relations between scale B and the groups at extreme ends of the age scale; the results for the students under 20 years old and over 26 years old were in favor of the first group ($p \leq 0.010$), in other words, in favor of the younger students. The effect size can be considered as moderate ($d = 0.584$). No significant differences were observed for the presence of binge-eating, vomiting, use of laxatives, physical exercise, and weight loss as a function of age. Nevertheless, the age range with higher percentages for all types of behavior was between 20 and 25 years old, except for the behavior of binge-eating where the age range extended from under 20 years old to people over 26 years old. ## 3.3. Differences as a Function of the Year of the Course Table 5 shows the mean scores and standard deviation obtained by the students on the OT, B and DT scales, according to the year of the course in which they are enrolled. An ANOVA test established that there were significant differences in the scores obtained by the students only in the Body Dissatisfaction (BD) scale, F[3,557] = 3.461, $$p \leq 0.016.$$ The multiple comparison analysis following the application of the Bonferroni statistic pointed to significative relations between the results of the BD scale for the 2nd and the 3rd year students, in favor of the 2nd year students ($$p \leq 0.029$$), with a small effect size ($d = 0.338$). These students therefore presented significatively higher levels of body dissatisfaction. No significant differences were observed, as a function of the year, on the other two scales, DT and B, nor for the five behavioral symptoms (binge eating, vomiting, laxatives, physical exercise, and weight loss). ## 3.4. Differences as a Function of the Educational Faculty Table 6 shows the mean scores and standard deviations obtained by the students on the OT, B, and BD scales according to the faculty in which they are enrolled. An ANOVA test established that there were significant differences in the scores obtained by the students only in the B scale, F[8,552] = 2.903, $$p \leq 0.004.$$ A Bonferroni post-hoc test revealed that the differences are statistically significant between students from the Faculty of Science and students from the Faculties of Law ($$p \leq 0.021$$) and Health ($$p \leq 0.013$$), in favor of the latter two. With a small effect size ($d = 0.070$). No significant differences were observed, in accordance with the faculty for the other two scales, DT and BD. Significant differences were only found between the five behavioral symptoms for the presence of binge eating ($p \leq 0.011$), in relation to the Health faculty with $40.4\%$ of students, followed by the Law faculty with $38.8\%$ of students (a small effect size, $d = 0.075$). No other significant differences were observed between the other groups. ## 3.5. Differences as a Function of Body Mass Indexes (BMI) Contingency Tables were applied to examine the BMI classification and to analyze its relation with gender (Table 7). A chi-square test reveals that there is a statistically significant relationship between gender and body mass index, χ2 [3] = 15.480, $$p \leq 0.001.$$ The corrected residual value reveals that significantly more females than men are underweight ($z = 2.8$). Similarly, more men than females are overweight ($z = 2.4$). Table 8 shows the mean scores and standard deviations obtained by the students on the OT, B, and BD scales according to the BMI classification. An ANOVA test established that there were significant differences in the scores obtained by the students in the OT scale, F[3,557] = 6.487, $p \leq 0.001$, and in the BD scale F[3,557] = 12.740, $p \leq 0.001.$ A Bonferroni post-hoc test revealed that in the OT scale, normal weight students scored significantly higher in thinness obsession than underweight students ($$p \leq 0.033$$). Similarly, overweight students scored significantly higher on thinness obsession than underweight students ($p \leq 0.001$) and normal weight students ($$p \leq 0.017$$). The effect size was small ($d = 0.034$). On the BD scale, it was found that overweight students scored significantly higher on body dissatisfaction than underweight students ($p \leq 0.001$) and normal weight students ($p \leq 0.001$). On the other hand, obese students scored significantly higher on body dissatisfaction than underweight students ($$p \leq 0.04$$) and normal weight students ($$p \leq 0.029$$). The effect size was small ($d = 0.064$). In relation to the presence of risk factors -binge eating, vomiting, use of laxatives, physical exercise, and weight loss-, different chi-square tests revealed that there is only a statistically significant relationship between BMI classification and the presence of risk for loss of weight, χ2 [3] = 8.388, $$p \leq 0.039$$ (Table 9). The corrected residual value determined that significantly more overweight students than expected have lost 9 kilograms or more during last 6 months ($z = 2.8$). Table 10 presents the percentages of students as a function of age in three categories and BMI. In relation to the percentages of BMI in four ranges, it can be noted that the highest percentages based on their BMI are in Normal weight and Overweight for students from 21 to 25 years old ($54.5\%$ and $51.2\%$, respectively). In Underweight and Obesity equal percentage ($50\%$ and $50\%$) for students 20 years old or less, and 21 to 25 years old. A chi-square test revealed that there was no statistically significant relationship between age in ranges and BMI classification, χ2 [6] = 3.281, $$p \leq 0.773.$$ When analyzing these two variables quantitatively (Mage = 21.72, SDage = 4.11; MBMI = 22.42, SDBMI = 2.97), a Pearson correlation revealed a positive relationship between the two variables that was statistically significant ($$p \leq 0.004$$) although with a low strength ($r = 0.120$). This may suggest that the older the age, the higher the BMI. In addition, a simple linear regression analysis indicated that age explains $1.3\%$ of the variance of BMI. That is, in $1.3\%$ of the cases, age predicts BMI [F[1,559] = 8.234, $$p \leq 0.004$$; R2 = 0.015; R2 corrected = 0.013; Age in years (β = 0.120, $$p \leq 0.004$$)]. ## 4. Discussion The purpose of this investigation was to estimate whether, in Spanish university students, variables such as gender, age, course, educational faculty, and BMI are associated with a higher risk of suffering from eating disorders (ED), with the aim of improving the understanding of the risk of suffering from eating disorders in the university population The results as a function of gender showed that, although women scored higher than men on the three scales of risk, the significative differences were found for Obsession with Thinness (OT) and Body Dissatisfaction (BD). These results and those of other works [5,7,10,18,21,62] are coincident, thereby supporting the theory that the female population is subjected to greater social pressure. Equally remarkable is that no significative differences were found between men and women for bulimic behavior (scale B). As is the case in the study by Kowalkowsk and Poínos [29], who found that emotional eating and uncontrolled eating are positively correlated in both sexes. This finding may be related with the results obtained by García et al. [ 34], who reported significantly higher scores for bulimic behavior among men. The authors proposed no explanation for that result, which they qualified as unexpected, although they pointed to a significant increase in BD and Eating Disorders (ED) among men from Western countries [41,42]. Moreover, significant differences were only found for practicing physical exercise among the five behavioral symptoms. In particular, men presented risks of excessive physical exercise as a form of controlling their weight. This notable point that Bo et al. [ 50] and González-Carrascosa et al. [ 7] also underlined is that sport is among the compensatory behaviors used by men suffering from some sort of ET, especially anorexia nervosa [63]. In the same line, Chin et al. [ 10] points out in a study among university students in the United Arab Emirates, that women who wanted to lose weight preferred dieting to physical exercise. While men preferred exercise to diet. With regard to the differences as a function of age, the students under 20 years old presented a more acute problematic (bulimic) behavior than the older students. These students did not present a higher BMI than the rest of the age ranges. Specifically, $78.9\%$ of the students 20 years old or younger were at a BMI of normal weight. Therefore, these results can be attributed to the critical age for the development of ED [3,5] and not to having a higher BMI than the rest. These results were similar to those of Sáenz et al. [ 6] who observed a greater risk of ED among university students under 19 years old. Although there were no significative differences in the presence of binge eating, vomiting, use of laxatives, physical exercise, and weight loss, the highest percentages for age were between 20 and 25 years, except for the behavior of binge eating where it was extended to 26 years old or over. Toro conducted a study in 2000 on people in need of specialized assistance, estimating that $4.5\%$ of the population between 12 and 25 years fell into that category, alerting the health authorities to the susceptibility of people within that age range. García et al. [ 34] also found high risks among over 25-year-old students, suggesting the need to implement preventive intervention strategies that covered broader age groups than the standard ones. Equally, Cooper and Goodyer [64], in an assessment of concern over weight and body image among girls of different ages, concluded that despite the concerns over body form and appearance that arise at the start of adolescence, the behaviors relating to eating disorders occurred much later [20]. As a function of the year in which the student was enrolled, it was observed that the students in the second year presented significantly higher body dissatisfaction than those in the third year. Students are normally between 18 and 20 years old, in the second year, and students younger than 20 years old felt greater dissatisfaction with their bodies, which may possibly be related to that growth phase of life in late adolescence [7,18,21]. On the other hand, body dissatisfaction was accompanied by negative emotions that might be responsible for a bulimic type of ED among the students under 20 years old observed in this work and that adds to the risk of suffering an ED. These results are close to those of Gropper et al. [ 44], who pointed out that the first years of university are a critical period for the development of an ED [65]. Increased independence and responsibilities, as well as concern over their own identity all contributed in turn to the development of eating pathologies [1]. Differences were found as a function of the study faculty between the students enrolled at the Health Sciences Faculty (Occupational Therapy and Nursing Degrees) and at the Law Faculty (Law; Political Science and Public Management Degrees; double Degree in Law and Administration and Business Management). In particular, they showed a greater tendency to think of uncontrolled attacks of binge eating and a higher tendency to indulge in them (scale B). From among the five behavioral questions, the students from these two faculties only showed a greater presence of binge eating. These results reinforce the results found in the literature [18,33], in so far as a higher risk of ED was attributed to the students from Health Sciences, as a consequence of greater concern for health and physical appearance. It would be necessary to investigate whether the students in this study presented pathological EDs prior to their entry into university, which might moreover be a reason for their choice of this area of study [50,52]. The results referring to the students from the Law Faculty were congruent with those obtained by other authors such as Peña et al. [ 21] who found a higher percentage of students at risk of ED following the Business Administration and Law degrees. Perhaps the students of those degrees considered that they were under greater pressure than on other degree courses, due to esthetic ideals of beauty that have traditionally been attributed to students on degrees such as Law and Administration and Business Management. Nevertheless, investigation will have to continue to analyze the underlying causes of these results. Regarding the differences according to the BMI value, we found that women were underweight and men were overweight. This finding is in accordance with the scientific literature. For example, Radwan et al. [ 66] found that women desired a thinner or smaller body, as opposed to men who desire a more muscular body and a larger body size [10]. Similarly, authors such as Momeni et al. [ 5] and Alipour et al. [ 67] note that thinness was the most desirable body image for female college students. On the other hand, we found that overweight students scored significantly higher in obsession with thinness (DT) and body dissatisfaction (BD) than their normal and underweight peers. Similarly, obese students have higher body dissatisfaction (BD) than their normal and underweight peers. These results in overweight and obese students are to be expected and are in line with the scientific literature [5,10,11]. These students suffer constant internal and external pressures to lose weight for health or social reasons. Numerous studies have found a positive correlation between body dissatisfaction and obsession with thinness with weight and BMI [8,10,11,29,30]. It is rare to find overweight subjects with low body dissatisfaction scores, although acceptance of body size without the need to lose weight is one of the primary goals of non-weight loss treatments for obesity [11,29,30]. With respect to students with normal weight showed greater obsession with thinness (DT) than those with low weight, data that seem particularly relevant. As reported by other researchers [5,11,68,69], if a person’s weight is within the normal range, high scores in thinness obsession indicate the possible presence of disturbing symptomatology. It should be noted that the same level of psychological distress (intense desire to be thinner) may have different meanings depending on weight [11]. For example, many patients with bulimia nervosa have a normal weight, but this value may represent a very significant loss of their previous weight [11]. However, as Momeni et al. [ 5] point out this negative evaluation of body image and weight is quite common among women in modern societies, even among those with a normal BMI. ## 5. Conclusions In conclusion, the female university students from the University of Burgos presented as significantly underweight, and with higher levels of obsession with thinness and greater body dissatisfaction than the men in the same sample. In turn, this body dissatisfaction was significantly more present among students enrolled in the second year of the degree. The men, in addition, presented as significantly overweight, practiced significantly more physical exercise as a form of controlling their weight and the younger students under 20 years old and those from the Faculties of Health Sciences and Law presented significantly higher bulimic and binge-eating behaviors. In general, overweight students in the study score significantly high in obsession with thinness and body dissatisfaction, and have also lost 9 kilos or more during the last 9 months. Similarly, students with obesity have high body dissatisfaction. On the other hand, students with normal weight score significantly high in obsession with thinness, which may indicate an alarm factor. The data obtained in this study, as well in other previous ones [10,23,24,70,71,72] reinforce the hypothesis that the female gender, at an age within the limits of early adolescence, the first year of the degree courses, certain university qualifications and a high BMI constituted factors that can influence the appearance or the continuance of risk-related patterns of eating, which could provoke one among various sorts of ED. In consequence, these results determine the need to implement measures that are specifically adapted to university students that stimulate heathy eating habits, improving the perception of their body image and reducing obsessive concerns over thinness. As García et al. [ 34] and Du et al. [ 30] pointed out, although maximum risk is reached during adolescence, the levels observed among university students are sufficiently important to propose these sorts of interventions. In addition, these symptoms are relatively stable during the university period [68,73]. In this sense, and with the aim of preventing ED among students at the University of Burgos (Spain), a group prevention program has been implemented this academic year (2022–2023). Fortunately, recent systematic revisions and meta-analyses have demonstrated theoretical and methodological advances in the field of the prevention of EDs. The calculations of the latest meta-analyses suggest that at present over half ($51\%$) of preventive interventions reduce the risk factors and somewhat over a quarter ($29\%$) reduce the prevalence and the incidence of present and future eating pathologies [3,27,57,74,75]. Preventive eating disorder interventions may also be effective in preventing the onset of subclinical and full syndrome eating disorders [27]. Consequently, as Harrer et al. [ 27] and Du et al. [ 30] points out future research should investigate how such interventions can be effectively implemented in university settings, and how to motivate students to participate. Therefore, universities should consider implementing screenings for maladaptive eating behaviors for students and providing interventions as needed [30]. Finally, the limitations of this research are principally related with the EDI-3-RF inventory, as a self-administered questionnaire in which some of the responses could have been both false positive and false negative. Other limitations of the study are related to the lack of follow-up to assess whether the results are maintained over time. On the other hand, this is a cross-sectional study, so a single measurement of the study variables was collected. Consequently, it is not possible to infer causal relationships between the variables. Furthermore, since the data were collected from a sample of students from a medium-sized public university, the results cannot be generalized to the rest of the Spanish population. With regard to future investigations, important areas include the development of ED-related awareness-raising measures, especially within the context of Burgos University, and the evaluation of their benefits. Future research projects will evaluate the benefits of applying the ED prevention program, implemented in the context of the University of Burgos. Likewise, as future research, we will analyze the risk of suffering ED in university students in relation to other aspects of mental health, such as anxiety and depression, aspects on which we are currently working. ## References 1. Martínez Martín M.A., Bilbao León M.C., Martínez Martín M.. Los Trastornos de La Conducta Alimentaria En El Contexto Universitario. Todo sobre los Trastornos de la Conducta Alimentaria (2015) 535-564 2. Barrera-Herrera A., Neira-Cofré M., Raipán-Gómez P., Riquelme-Lobos P., Escobar B.. Apoyo Social Percibido y Factores Sociodemográficos En Relación Con Los Síntomas de Ansiedad, Depresión y Estrés En Universitarios Chilenos. Rev. Psicopatología y Psicol. Clínica (2019) 24 105. DOI: 10.5944/rppc.23676 3. Daly M., Costigan E.. Trends in Eating Disorder Risk among U.S. College Students, 2013–2021. Psychiatry Res. (2022) 317 10-13. DOI: 10.1016/j.psychres.2022.114882 4. Foran A.M., Muldoon O.T., O’Donnell A.T.. Identity Processes and Eating Disorder Symptoms during University Adjustment: A Cross-Sectional Study. J. Eat. Disord. (2021) 9 44. DOI: 10.1186/s40337-021-00399-4 5. Momeni M., Ghorbani A., Arjeini Z.. Disordered Eating Attitudes among Iranian University Students of Medical Sciences: The Role of Body Image Perception. Nutr. Health (2020) 26 127-133. DOI: 10.1177/0260106020912657 6. Sáenz Duran S., González Martínez F., Díaz Cárdenas S.. Hábitos y Trastornos Alimenticios Asociados a Factores Socio-Demográficos, Físicos y Conductuales En Universitarios de Cartagena, Colombia. Rev. Clínica Med. Fam. (2011) 4 193-204. DOI: 10.4321/S1699-695X2011000300003 7. González-Carrascosa R., García-Segovia P., Martínez-Monzo J.. Valoraciónn de La Imagen Corporal y de Los Comportamientos Alimentarios En Universitarios. Rev. Psicopatol. Psicol. Clin. (2013) 18 45-59 8. Nasrallah C., Kimmel L., Khaled S.M.. Associations between Weight Loss Difficulty, Disordered Eating Behaviors and Poor Weight Loss Outcomes in Arab Female University Students. Eat. Behav. (2020) 36 101363. DOI: 10.1016/j.eatbeh.2020.101363 9. 9. Association American Psychiatry Manual Diagnóstico y Estadístico de Los Trastornos Mentales (DSM-5)Editorial Médica PanamericanaMadrid, Spain2014. Manual Diagnóstico y Estadístico de Los Trastornos Mentales (DSM-5) (2014) 10. Chin Y.S., Appukutty M., Kagawa M., Gan W.Y., Wong J.E., Poh B.K., Shariff Z.M., Taib M.N.M.. Comparison of Factors Associated with Disordered Eating between Male and Female Malaysian University Students. Nutrients (2020) 12. DOI: 10.3390/nu12020318 11. Garner D.M.. EDI-3 RF. Inventario de Trastornos de La Conducta Alimentaria-3 (2010) 12. González-Alonso M.Y., Escolar-Llamazares M.C., Martín M.Á.M., Gómez M.B.M., Val E.M.. Comorbidity of Eating Disorders with Anxiety and Depression in University Students: Systematic Review. Rev. Argentina Clin. Psicol. (2019) 28 375-384. DOI: 10.24205/03276716.2019.1149 13. Escolar-LLamazares M.C., Martínez Martín M.A., González Alonso M.Y., Medina Gómez M.B., Mercado Val E., Lara Ortega F.. Risk Factors of Eating Disorders in University Students. Rev. Mex. Trastor. Aliment. (2017) 8 105-112. DOI: 10.1016/j.rmta.2017.05.003 14. Keshk M.M., Fahim H.I., Hassan A.M., Boulos D.N.. Body Image Perception and Self-Esteem among University Students in Cairo. Egypt. J. Community Med. (2019) 37 82-96. DOI: 10.21608/ejcm.2019.28135 15. Franco K., Jesús Díaz F., López-Espinoza A., Escoto M.C., Camacho E.J.. Variables Predictoras de Riesgo de Trastorno Del Comportamiento Alimentario En Mujeres. Ter. Psicológica (2013) 31 219-225. DOI: 10.4067/S0718-48082013000200008 16. Sandoz E.K., Boullion G.Q., Mallik D., Hebert E.R.. Relative Associations of Body Image Avoidance Constructs with Eating Disorder Pathology in a Large College Student Sample. Body Image (2020) 34 242-248. DOI: 10.1016/j.bodyim.2020.07.002 17. Leiva V.A., Tapia C.N., Bergeret Á.Q., Pérez C.U.. Dissatisfaction with Body Image and Nutritional Status in University Students. Rev. Psicoter. (2020) 31 279-294 18. Aguiar dos Reis J., Rodrigues Silva Júnior C.R., De Pinho L.. Factors Associated with the Risk of Eating Disorders among Academics in the Area of Health. Rev. Gaúcha Enferm. (2014) 35 73-78. DOI: 10.1590/1983-1447.2014.02.42441 19. Chan Y.L., Samy A.L., Tong W.T., Islam M.A., Low W.Y.. Eating Disorder Among Malaysian University Students and Its Associated Factors. Asia Pacific J. Public Health (2020) 32 334-339. DOI: 10.1177/1010539520947879 20. Galarsi M.F., Correche M.S., Ledezma C.. Conductas y Actitudes Hacia La Alimentación En Una Muestra de Mujeres Jóvenes. Fundam. Humanid. (2010) 1 193-205 21. Peña Salgado N.D.P., López de Arco S.P., Liévano Fiesco M.C.. Conductas Alimentarias Asociadas a TCA En Estudiantes Universitarios Que Asisten a Un Centro de Formación Deportiva. Rev. Iberoam. Psicol. Cienc. Tecnol. (2014) 7 29-38 22. Barnhart W.R., Braden A.L., Price E.. Emotion Regulation Difficulties Interact with Negative, Not Positive, Emotional Eating to Strengthen Relationships with Disordered Eating: An Exploratory Study. Appetite (2021) 158 105038. DOI: 10.1016/j.appet.2020.105038 23. Berengüí R., Castejón M.Á., Torregrosa M.S.. Body Dissatisfaction, Risk Behaviors and Eating Disorders in University Students. Rev. Mex. Trastor. Aliment. (2016) 7 1-8. DOI: 10.1016/j.rmta.2016.02.004 24. Medina-Gómez M.B., Martínez-Martín M.Á., Escolar-Llamazares M.C., González-Alonso Y., Mercado-Val E.. Anxiety and Body Dissatisfaction in University Students. Acta Colomb. Psicol. (2019) 22 13-21. DOI: 10.14718/ACP.2019.22.1.2 25. Bratić B., Triglianos V., Kurbalija V., Pautasso C., Ivanović M.. Role of Interactive Presentation Platform ASQ in Delivering Web Design Course. Smart Learn. Environ. (2020) 7 44. DOI: 10.1186/s40561-020-00123-w 26. Hamamoto Y., Suzuki S., Sugiura M.. Two Components of Body-Image Disturbance Are Differentially Associated with Distinct Eating Disorder Characteristics in Healthy Young Women. PLoS ONE (2022) 17. DOI: 10.1371/journal.pone.0262513 27. Harrer M., Adam S.H., Messner E.M., Baumeister H., Cuijpers P., Bruffaerts R., Auerbach R.P., Kessler R.C., Jacobi C., Taylor C.B.. Prevention of Eating Disorders at Universities: A Systematic Review and Meta-Analysis. Int. J. Eat. Disord. (2020) 53 813-833. DOI: 10.1002/eat.23224 28. Castellano S., Rizzotto A., Neri S., Currenti W., Guerrera C.S., Pirrone C., Coco M., Di Corrado D.. The Relationship between Body Dissatisfaction and Eating Disorder Symptoms in Young Women Aspiring Fashion Models: The Mediating Role of Stress. Eur. J. Investig. Health Psychol. Educ. (2021) 11 607-615. DOI: 10.3390/ejihpe11020043 29. Kowalkowska J., Poínhos R.. Eating Behaviour among University Students: Relationships with Age, Socioeconomic Status, Physical Activity, Body Mass Index, Waist-to-Height Ratio and Social Desirability. Nutrients (2021) 13. DOI: 10.3390/nu13103622 30. Du C., Adjepong M., Chong M., Zan H., Cho M.J., Fenton J.I., Hsiao P.Y., Keaver L., Lee H., Ludy M.. Gender Differences in the Relationships between Perceived Stress, Eating Behaviors, Sleep, Dietary Risk, and Body Mass Index. Nutrients (2022) 14. DOI: 10.3390/nu14051045 31. Faus Boronat G., Matínez Martin M.Á.. Trastornos de La Conducta Alimentaria En Varones. Todo sobre los Trastornos de la Conducta Alimentaria. Una Visión Multidisciplinar Desde la Experiencia y la Evidencia Científica (2015) 191-2014 32. Martínez-González L., Fernandez Villa T., Molina De La Torre A.J., Ayan Pérez C., Bueno Cavanillas A., Capelo Álvarez R., Mateo Campos R., Martín Sández V.. Prevalencia de Trastornos de La Conducta Alimentaria En Universitarios Españoles y Factores Asociados: Proyecto UniHcos. Nutr. Hosp. (2014) 30 927-934. DOI: 10.3305/nh.2014.30.4.7689 33. Toral N., Gubert M.B., Spaniol A.M., Monteiro R.A.. Eating Disorders and Body Image Satisfaction among Brazilian Undergraduate Nutrition Students and Dietitians. Arch. Latinoam. Nutr. (2016) 66 129-134. PMID: 29737669 34. García Alba J., Solbes Canales I., Expósito Casas E., Navarro Asencio E.. Imagen Corporal y Riesgo de Trastornos de La Conducta Alimentaria En Población Universitaria Española: Diferencias de Género. Rev. Orientación Educ. (2012) 26 31-46 35. Kanayama G., Hudson J.I., Pope Jr H.G.. Anabolic-Androgenic Steroid Use and Body Image in Men: A Growing Concern for Clinicians. Psychother. Psychosom. (2020) 89 65-73. DOI: 10.1159/000505978 36. Ramos-Jiménez A., Hernández-Torres R.P., Urquidez-Romero R., Wall-Medrano A., Villalobos-Molina R.. Body Image Satisfaction as a Physical Activity Indicator in University Students. Am. J. Health Behav. (2017) 41 599-607. DOI: 10.5993/AJHB.41.5.9 37. Sepúlveda A.R., Gandarillas A., Carrobles J.A.. Prevalencia de Trastornos Del Comportamiento En La Población Universitaria. Rev. Psiquiatr. Com (2004) 8 2 38. Gómez Sánchez R., Gómez Sánhez R., Díaz Cuenca A.M., Fortea Gorbe M.I., Salorio del Moral P.. Prevalencia de Los Trastornos de Alimentaciόn En Una Muestra Universitaria. Ansiedad Como Factor de Modulación. Index Enfermería (2010) 19 124-128. DOI: 10.4321/S1132-12962010000200011 39. Lameiras M., Calado M., Rodríguez Y., Fernández M.. Hábitos Alimentarios e Imagen Corporal En Estudiantes Universitarios Sin Trastornos Alimentarios. Rev. Int. Psicol. Clínica Salud. Int. J. Clin. Health Psychol. (2003) 3 23-33 40. Schilder C.M.T., Sternheim L.C., Aarts E., van Elburg A.A., Danner U.N.. Relationships between Educational Achievement, Intelligence, and Perfectionism in Adolescents with Eating Disorders. Int. J. Eat. Disord. (2021) 54 794-801. DOI: 10.1002/eat.23482 41. Andersen A.. Eating Disorders and Obesity (2005) 42. Schaefer L.M., Rodgers R.F., Thompson J.K., Griffiths S.. A Test of the Tripartite Influence Model of Disordered Eating among Men. Body Image (2021) 36 172-179. DOI: 10.1016/j.bodyim.2020.11.009 43. Bibiloni M.M., Coll J.L., Pich J., Pons A., Tur J.A.. Body Image Satisfaction and Weight Concerns among a Mediterranean Adult Population. BMC Public Health (2017) 17. DOI: 10.1186/s12889-016-3919-7 44. Gropper S.S., Arsiwalla D.D., Lord D.C., Huggins K.W., Simmons K.P., Ulrich P.V.. Associations among Eating Regulation and Body Mass Index, Weight, and Body Fat in College Students: The Moderating Role of Gender. Eat. Behav. (2014) 15 321-327. DOI: 10.1016/j.eatbeh.2014.04.002 45. Kriaucioniene V., Raskiliene A., Petrauskas D., Petkeviciene J.. Trends in Eating Habits and Body Weight Status, Perception Patterns and Management Practices among First-Year Students of Kaunas (Lithuania) Universities, 2000–2017. Nutrients (2021) 13. DOI: 10.3390/nu13051599 46. Beaudry K.M., Ludwa I.A., Thomas A.M., Ward W.E., Falk B., Josse A.R.. First-Year University Is Associated with Greater Body Weight, Body Composition and Adverse Dietary Changes in Males than Females. PLoS ONE (2019) 14. DOI: 10.1371/journal.pone.0218554 47. Mueller M.P., Blondin S.A., Korn A.R., Bakun P.J., Tucker K.L., Economos C.D.. Behavioral Correlates of Empirically-Derived Dietary Patterns among University Students. Nutrients (2018) 10. DOI: 10.3390/nu10060716 48. Novais A.I.D., de Oliveira I.B.L., Martins P.C., Santos V.M., de Oliveira Penaforte F.R.. Fatores Associados Ao Ganho de Peso Após o Ingresso Na Universidade. El J. Food Cult. Am. (2020) 2 60-79. DOI: 10.35953/raca.v1i1.17 49. Sousa T.F.D., Barbosa A.R.. Prevalence of Body Weight Excess in Undergraduate Students: Analysis of Repeated Surveys. Rev. Bras. Epidemiol. (2017) 20 586-597. DOI: 10.1590/1980-5497201700040003 50. Bo S., Zoccali R., Ponzo V., Soldati L., De Carli L., Benso A., Fea E., Rainoldi A., Durazzo M., Fassino S.. University Courses, Eating Problems and Muscle Dysmorphia: Are There Any Associations?. J. Transl. Med. (2014) 12 221. DOI: 10.1186/s12967-014-0221-2 51. Silva Vereau J.E.. Factores Asociados a La Elección de Una Especialidad Quirúrgica en Estudiantes de Sesenta y Tres Escuelas de Medicina de Latinoamérica (2016) 52. Lupo R., Zaminga M., Carriero M.C., Santoro P., Artioli G., Calabrò A., Ilari F., Benedetto A., Caslini M., Clerici M.. “Eating Disorders and Related Stigma”: Analysis among a Population of Italian Nursing Students. Acta Biomed. (2020) 91 1-12. DOI: 10.23750/abm.v91i12-S.10797 53. Kinzl J.F., Hauer K., Traweger C., Kiefer I.. Orthorexia Nervosa in Dieticians. Psychother. Psychosom. (2006) 75 395-396. DOI: 10.1159/000095447 54. Rocks T., Pelly F., Slater G., Martin L.A.. Eating Attitudes and Behaviours of Students Enrolled in Undergraduate Nutrition and Dietetics Degrees. Nutr. Diet. (2017) 74 381-387. DOI: 10.1111/1747-0080.12298 55. Cancela Carral J.M., Ayán Pérez C.. Prevalencia y Relación Entre El Nivel de Actividad Física y Las Actitudes Alimenticias Anómalas En Estudiantes Universitarias Españolas de Ciencias de La Salud y La Educación. Rev. Esp. Salud Publica (2011) 85 499-505. DOI: 10.1590/S1135-57272011000500009 56. Yager Z., Gray T., Curry C., McLean S.A.. Body Dissatisfaction, Excessive Exercise, and Weight Change Strategies Used by First-Year Undergraduate Students: Comparing Health and Physical Education and Other Education Students. J. Eat. Disord. (2017) 5 10. DOI: 10.1186/s40337-016-0133-z 57. Jiménez-Limas K., Miranda-Barrera V.A., Muñoz-Díaz K.F., Novales-Huidobro S.R., Chico-Barba G.. Body Dissatisfaction, Distorted Body Image and Disordered Eating Behaviors in University Students: An Analysis from 2017–2022. Int. J. Environ. Res. Public Health (2022) 19. DOI: 10.3390/ijerph191811482 58. Montero I., León O.G.. A Guide for Naming Research Studies in Psychology. Int. J. Clin. Health Psichol. (2007) 7 847-862 59. Suárez Falcón J.C., Recio Saboya P., San Luis Costas M.C., Pozo Cabanillas M.P.. Introducción Al Análisis de Datos. Aplicaciones En Psicología y Ciencias de La Salud (2019) 60. George D., Mallery P.. SPSS/PC+ Step by Step: A Simple Guide and Reference (1995) 61. Cohen J.. Statistical Power Analysis for the Behavioral Sciences (2013) 62. Míguez Bernárdez M., De la Montaña Miguélez J., González Carnero J., González Rodríguez M.. Concordancia Entre La Autopercepción de La Imagen Corporal y El Estado Nutricional En Universitarios de Orense. Nutr. Hosp. (2011) 26 472-479. DOI: 10.3305/nh.2011.26.3.4604 63. Al-Jumayan A.A., Al-Eid N.A., AlShamlan N.A., AlOmar R.S.. Prevalence and Associated Factors of Eating Disorders in Patrons of Sport Centers in Saudi Arabia. J. Fam. Community Med. (2021) 28 94-102. DOI: 10.4103/jfcm.jfcm_113_21 64. Cooper P.J., Goodyer I.. Prevalence and Significance of Weight and Shape Concerns in Girls Aged 11–16 Years. Br. J. Psychiatry (1997) 171 542-544. DOI: 10.1192/bjp.171.6.542 65. McAndrew A.J., Menna R., Oldershaw M.. Disordered Eating and Barriers to Help-Seeking: A Brief Report. Eat. Disord. (2020) 30 239-248. DOI: 10.1080/10640266.2020.1771166 66. Radwan H., Hasan H.A., Ismat H., Hakim H., Khalid H., Al-Fityani L., Mohammed R., Ayman A.. Body Mass Index Perception, Body Image Dissatisfaction and Their Relations with Weight-Related Behaviors among University Students. Int. J. Environ. Res. Public Health (2019) 16. DOI: 10.3390/ijerph16091541 67. Alipour B., Abbasalizad Farhangi M., Dehghan P., Alipour M.. Body Image Perception and Its Association with Body Mass Index and Nutrient Intakes among Female College Students Aged 18–35 Years from Tabriz, Iran. Eat. Weight Disord. (2015) 20 465-471. DOI: 10.1007/s40519-015-0184-1 68. Dakanalis A., Pla-Sanjuanelo J., Caslini M., Volpato C., Riva G., Clerici M., Carrà G.. Predicting Onset and Maintenance of Men’s Eating Disorders. Int. J. Clin. Health Psychol. (2016) 16 247-255. DOI: 10.1016/j.ijchp.2016.05.002 69. Nikniaz Z., Mahdavi R., Amiri S., Ostadrahimi A., Nikniaz L.. Factors Associated with Body Image Dissatisfaction and Distortion among Iranian Women. Eat. Behav. (2016) 22 5-9. DOI: 10.1016/j.eatbeh.2016.03.018 70. Invitada E.. Editorial Invitada. World (2007) 1 1-2 71. Fernández Aranda F., Turon Gil V.. Trastornos de La Alimentación. Guía Básica de Tratamiento En Anorexia y Bulimia (1998) 72. Castelao-Naval O., Blanco-Fernández A., Meseguer-Barros C.M., Thuissard-Vasallo I.J., Cerdá B., Larrosa M.. Estilo de Vida y Riesgo de Trastorno Alimentario Atípico En Estudiantes Universitarios: Realidad versus Percepción. Enfermería Clínica (2019) 29 280-290. DOI: 10.1016/j.enfcli.2018.03.002 73. Cain A.S., Epler A.J., Steinley D., Sher K.J.. Concerns Related to Eating, Weight, and Shape: Typologies and Transitions in Men during the College Years. Int. J. Eat. Disord. (2012) 45 768-775. DOI: 10.1002/eat.20945 74. Ruiz Lázaro P.M., Martínez Martín M.A.. Prevención de Los Trastornos de La Conducta Alimentaria y Los Diversos Ámbitos de Actuación. Todo sobre los Trastornos de la Conducta Alimentaria (2015) 431-476 75. Barnett P., Arundell L.L., Saunders R., Matthews H., Pilling S.. The Efficacy of Psychological Interventions for the Prevention and Treatment of Mental Health Disorders in University Students: A Systematic Review and Meta-Analysis. J. Affect. Disord. (2021) 280 381-406. DOI: 10.1016/j.jad.2020.10.060
--- title: A Feedback Loop between TGF-β1 and ATG5 Mediated by miR-122-5p Regulates Fibrosis and EMT in Human Trabecular Meshwork Cells authors: - Munmun Chakraborthy - Aparna Rao journal: Current Issues in Molecular Biology year: 2023 pmcid: PMC10047315 doi: 10.3390/cimb45030154 license: CC BY 4.0 --- # A Feedback Loop between TGF-β1 and ATG5 Mediated by miR-122-5p Regulates Fibrosis and EMT in Human Trabecular Meshwork Cells ## Abstract Autophagy is a cell’s evolutionary conserved process for degrading and recycling cellular proteins and removing damaged organelles. There has been an increasing interest in identifying the basic cellular mechanism of autophagy and its implications in health and illness during the last decade. Many proteinopathies such as Alzheimer’s and Huntington’s disease are reported to be associated with impaired autophagy. The functional significance of autophagy in exfoliation syndrome/exfoliation glaucoma (XFS/XFG), remains unknown though it is presumed to be impaired autophagy to be responsible for the aggregopathy characteristic of this disease. In the current study we have shown that autophagy or ATG5 is enhanced in response to TGF-β1 in human trabecular meshwork (HTM) cells and TGF-β1 induced autophagy is necessary for increased expression of profibrotic proteins and epithelial to mesenchymal (EMT) through Smad3 that lead to aggregopathy. Inhibition of ATG5 by siRNA mediated knockdown reduced profibrotic and EMT markers and increased protein aggregates in the presence of TGF-β1 stimulation. The miR-122-5p, which was increased upon TGF exposure, was also reduced upon ATG5 inhibition. We thus conclude that TGF-β1 induces autophagy in primary HTM cells and a positive feedback loop exists between TGF-β1 and ATG5 that regulated TGF downstream effects mainly mediated by Smad3 signaling with miR-122-5p also playing a role. ## 1. Introduction Pseudoexfoliation syndrome (XFS) is a protein aggregation disorder and the most common cause of exfoliation glaucoma (XFG). XFS is characterized by white flaky deposits seen over ocular surfaces. Transforming growth factor-beta1 (TGF-β1) which has been shown to be increased in XFS and XFG [1,2], not only plays a specialized role in the body, it also regulates other pathways, which helps to explain why TGF-signalling dysregulation is linked to a variety of disorders. TGF also regulates extracellular matrix homeostasis, which is known to be dysregulated in glaucoma, in XFS/XFG [1,2]. Abnormal extracellular matrix (ECM) production and reduced clearance of ECM aggregates are known to be caused by TGF in several fibroproliferative diseases, including XFG [1,3,4]. In XFG, the accumulation of protein complex aggregates in the trabecular meshwork causes a blockage of the outflow channels leading to raised intraocular pressure, glaucoma, and subsequent blindness. The trabecular meshwork, therefore, constitutes the tissue of end-organ damage in glaucoma and XFG, which therefore is the primary site of molecular events preceding cell death or dysfunction. In several studies [3,4,5], the role of dysregulated autophagy and TGF-β signaling pathways in the pathogenesis of kidney fibrosis, diabetes, and its complications, such as cardiomyopathy, retinopathy, and nephropathy, has been reported. In our previous study, we have also shown that 10 ng/mL TGF-β1 causes an increase in ECM protein production, causes apoptosis, induces epithelial to mesenchymal transition (EMT), and protein aggregation in HTM cells in vitro [6]. The reduction of cellular degradative processes, particularly the autophagy–lysosomal pathway, is linked to protein aggregation. We recently discovered that in all stages of XFG, lower unfolded protein response (UPR) clearance is related to elevated TGF levels, implying that the autophagy pathway and TGF-β autophagy crosstalk may be involved in aggregate clearance [1,7]. Autophagy is an intracellular trafficking system that transports cytosolic elements to the lysosome for destruction, which is necessary for misfolded protein clearance, ubiquitin–proteasomal degradation, and cell repair [7,8,9,10,11,12]. Autophagy is a protective mechanism that can be triggered by a variety of intracellular and extracellular stimuli, including a lack of amino acids or growth hormones, hypoxia, a low cellular energy status, endoplasmic reticulum stress or oxidative stress, organelle injury, and pathogen infection [5,7,13,14,15,16,17,18,19]. While autophagy is a driving force in the regulation of cell viability and function, it is also regarded as the second kind of programmed cell death, with autophagosome accumulation differing from apoptosis [20,21]. Micro ribonucleic acids (MicroRNAs), which are noncoding RNAs, have also been shown to regulate autophagy either directly or through a variety of routes, and so have an important role in the cause and treatment of a variety of disorders. Given the role of autophagic clearance of protein aggregates, autophagy-related genes (ATG genes) may be involved in XFG pathophysiology. Furthermore, the potential for autophagy and TGF to interact has attracted a lot of attention. The general consensus is that autophagic pathways are beneficial for cell functions since they facilitate the elimination of aggregates that are too big for proteasome-mediated clearance. In fact, there is a clear link reported between autophagy induction and the existence of protein aggregates in various neurodegenerative disorders, including Alzheimer’s disease [22], Huntington’s disease [23], and amyotrophic lateral sclerosis [24]. In light of this evidence, it would be interesting to look into the role of autophagy and TGF signaling pathways, as well as their crosstalk, in the progression of XFS/XFG, in order to spur the development of targeted therapeutic strategies. We have reported that a feedback loop exists between TGF-β1 and ATG5 which regulates the induction of profibrotic proteins and EMT. Further research into autophagy in XFS and XFG could lead to new knowledge of the disease’s pathogenesis as well as new therapeutic options in the future. ## 2.1. siRNA Transfection The siRNA transfection was carried out as previously described [25]. Briefly, 60 pmol/well of siRNAs against autophagy-essential ATG5 (siATG5) (6345S, Cell signalling technology, Beverly, MA, USA)and nontargeting siRNA (negative control, siNC) (6568S, Cell signalling technology, Beverly, MA, USA) in serum-free medium was mixed with Lipofectamine RNAiMAX reagent (13778, Invitrogen, Beverly, MA, USA) and incubated for 60 min according to the instructions provided by the manufacturers. Passage 4 HTM cells were then suspended in the siRNA–lipofectamine mixture and plated on 6 well plates(reverse transfection approach for successful siRNA distribution). Cells were then cultivated and treated accordingly and used for further analysis. Transient knockdown of the ATG5 gene was chosen because ATG5 participates in both for autophagy induction and conjugation cascades leading to microtubule-associated protein light chain (LC3) lipidation and autophagosome formation [12], thus silencing autophagy. ## 2.2. Stimulation of HTM Cells with TGF-β1 We isolated primary HTM cells from healthy cadaver eyes procured from the institutional eye bank (age of donors 28 ± 1.2 years, all cadaver corneas procured within 4 h of death). The use of human tissue in this study adhered to the tenets of the Declaration of Helsinki. Briefly, the trabecular meshwork (TM) stripped from the procured corneoscleral rims were thoroughly washed with Dulbecco’s-phosphate buffer saline (D-PBS, A12856-01, Gibco, Carlsbad, CA, USA) followed by dissection of the TM tissue into tiny pieces with microscissors under a dissection microscope (Olympus SZX7, Tokyo, Japan). The cut TM pieces were digested with collagenase treatment (1 mg/mL at 37 °C for 2 h) which was followed by centrifugation (1200 rpm, 5 min) and trypsinization ($0.25\%$ trypsin, 5 min). The TM cells were pelleted down by centrifugation (1200 rpm, 5 min) and plated on a 35 mm dish with an endothelial cell basal medium, EBM-2 media (cc-3156, Lonza, Basel, Switzerland) at 37 °C and $5\%$ CO2. Primary HTM cells (passage 1 to 5) infected with and without siRNA (siATG5 and siNC) were cultured to $80\%$ confluency in Dulbecco’s modified eagle medium (DMEM) + $10\%$ fetal bovine serum (FBS). Cells were then serum-starved DMEM ($0\%$ FBS) for 24 h before being treated with TGF-β1 (10 ng/mL) for 0–72 h in preparation for the experiment. ## 2.3. RNA Isolation and Quantitative Polymerase Chain Reaction (qPCR) QIAzol lysis reagent was used to extract total RNA from control and treated cells (passage 3) (QIAGEN, Hilden, Germany). One µg of RNA was reverse-transcribed using the Reverse Transcriptase Core Kit (RT-RTCK-03, EUROGENTEC, Liege, Belgium), and the cDNA was subjected to quantitative PCR using the PowerUp SYBR Green qPCR Master Mix(A25741, Applied Biosystems, Foster City, CA, USA) with the following reaction conditions: Stage 1: 50 °C for 2 min, Stage 2 (40 cycles): 94 °C for 2 min, 95 °C for 3 s and 60 °C for 30 s (for Primers: ATG5 Forward-TGGGATTGCTCAGGCAACGAA, Reverse-TTCCCCATCTTCAGGATCAA and LC3-II Forward-GAGAAGCTTCCTGTTCTGG, Reverse-GTGTCCGTTCACCAACAGGAAG). The data were examined using the threshold (Ct) method after the Ct values were standardized to GAPDH, which served as an internal reference. We also used the Applied Biosystems TaqMan Advanced miRNA Assay kit (A25576, Applied Biosystems, Foster City, CA, USA) which detects and quantifies mature miRNA from as little as 1 pg of total RNA. After RNA preparation, cDNA was synthesized using the TaqMan Advanced miRNA cDNA Synthesis Kit. The cDNA is then preamplified using universal primers and a master mix to uniformly increase the amount of cDNA for each target, maintaining the relative differential levels after which the expression levels of miR-122-5p, miR-124-3p, and miR-424-5p) are quantified. The Ct values were standardized to U6 (RNU6-1), which served as an internal reference. ## 2.4. Annexin V/propidium Iodide (PI) Staining Assay Apoptosis was examined in TGF-β1 and siRNA-treated and untreated cultures (passage 3) using the annexin V-fluorescein isothiocyanate (FITC)/prodium iodide (PI) dual staining method with the BD FACS Canto II Flow Cytometer (BD Biosciences, San Jose, CA, USA) using the Annexin V-FITC Apoptosis Detection kit (MintenyiBiotec, Bergisch, Gladbach, Germany). The percentage of the total (early + late) apoptotic cells were plotted. ## 2.5. Protein Whole Cell Lysate and Immunoblotting Immunoblotting was performed using a method previously described [6]. Briefly, cells (passage 4) were treated accordingly, and protein lysates were extracted in RIPA buffer containing protease inhibitors, 2.0 mM N-ethylmaleimide, 2.0 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride, 0.05 M Tris-HCl (pH 8.0), 0.15 M sodium chloride (NaCl), 5.0 mM ethylenediaminetetraacetic acid (EDTA), and $1\%$ NP-40. The Bradford assay was used to quantify the protein, and an equal amount of protein was placed onto each lane of 10–$12\%$ polyacrylamide SDS-PAGE gels and transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore). Membranes were blocked with $5\%$ skimmed milk and incubated with antibodies. The primary antibodies used in the study were mouse anti-SMA (ab7817, 1:1000, Abcam, Cambridge, MA, USA), antivimentin (D21H3, 1:1000, Cell Signalling Technology, Beverly, MA, USA), rabbit antifibronectin (AF5335, 1:1000, Affinity Biosciences, Brisbane, QLD, Australia), mouse anti-fibrillin (AF0429, 1:1000, Affinity Biosciences, Brisbane, QLD, Australia), rabbit anti-ATG-5 (DF6010, 1:800, Affinity Biosciences, Brisbane, QLD, Australia), anti-LC3I/II (AF5402, 1:1000, Affinity Biosciences, Brisbane, QLD, Australia), and SMAD3 (MA5-14939, Abcam, Cambridge, MA, USA). The loading control used was rabbit anti-GAPDH (1:10,000; Abcam, Cambridge, MA, USA). The primary HTM cells were treated with Smad3 inhibitor, SIS3 (CAS 1009104-85-1), with and without TGF-β1 treatment and the protein levels of ATG-5, LC3-I and LC3-II were validated. ## 2.6. Electron Micrograph and Immunofluorescence Staining Cells (passage 5) were treated with TGF-β1 for 72h, pelleted down, and fixed in $2\%$ glutaraldehyde in 0.1 mol/L phosphate buffer for 15 h, then postfixed in $2\%$ buffered osmium tetroxide and embedded in epoxy resin (Epon) in a usual method. Ultrathin sections stained with uranyl acetate–lead citrate mixture were then seen under an electron microscope (JEOL, JEM-2100 PLUS, Akishima, Tokyo, Japan). For immunostaining, control and siATG5 transfected cells were sown in coverslips covered with $0.1\%$ gelatin before TGF-β1 treatment. The cell culture medium was separated and the cells were rinsed in phosphate buffer saline (PBS) before being fixed for 15 min at room temperature (RT) with $4\%$ paraformaldehyde. The cells were then permeabilized with a $0.2\%$ Triton X-100 solution for 10 min before being incubated with a blocking solution ($5\%$ FBS in 1X PBS) for 20 min at RT with gentle rocking to block nonspecific sites. Following this, the cells were incubated overnight at 4 °C with primary antibodies, Oligomer 11 (AHB0052, 1:1500, Invitrogen, Beverly, MA, USA) and amyloid fibril (PA5-77843, 1:1500, Invitrogen, Beverly, MA, USA), followed by 1 h with secondary antibody (antirabbit Alexa 488, 1:2000, Invitrogen, Beverly, MA, USA) mixed with 2 g/mL Hoechst (four nucleus staining) at RT. The coverslips were mounted and examined under a fluorescence microscope (Olympus, BX53, Tokyo, Japan). ## 2.7. Statistical Analysis Each experiment was carried out three times. The mean data were reported as mean +/− standard deviation. Graphpad prism (Version 7, San Diego, CA, USA) was used for statistical analysis. An unpaired Student t-test was performed to determine statistical differences between the two groups, and a one-way ANOVA with post hoc analysis was employed for >2 groups, with $p \leq 0.05$ set as statistical significance. ## 3.1. TGF-β1 Induced ATG5 Activation, Fibrosis, and Epithelial to Mesenchymal Transition in HTM TGF-β1 is one of a number of factors involved in fibrogenesis and EMT. Our group has previously studied and reported TGF-β1-induced fibrosis and EMT in HTM cells [6]. Our first objective was to see if TGF-β1 also induces autophagy in HTM cells. We observed that 10 ng/mL TGF-β1 caused considerable increases in autophagy markers, ATG5, and LC3-II at 72 h as compared to time-matched controls. A significant increase in protein levels of ATG5 and LC-I and II was also found (Figure 1). ## 3.2. TGF-β1-Induced Fibrosis and Epithelial to Mesenchymal Transition Is Regulated by Autophagy Induction We investigated the effects of TGF-β1 activation of HTM in the presence of autophagy inhibition to see if the link between autophagy and TGF-β1-induced fibrosis was more than correlational. We employed siRNA-mediated transient ATG5 inhibition (siATG5) to curtail autophagy (autophagy inhibition was confirmed by reduced ATG5 protein levels) and analyzed the existence of autophagy flux to get a more quantitative assessment of autophagy induction. Inhibition of the autophagy gene ATG5 dramatically reduced the α-SMA and vimentin increase induced by TGF-β1. We also investigated the levels of fibronectin and fibrillin to see if this inhibitory impact was also relevant to the profibrotic proteins. Silencing ATG5 also reduced the TGF-induced expression of fibronectin and fibrillin, as demonstrated in Figure 2. The accumulation of LC3-II caused by TGF-β1 was increased in the siATG5 transfected cells. These results in summary support our hypothesis that TGF-β1 regulates and induces EMT and fibrosis in HTM cells via crosstalk with the autophagy pathway. ## 3.3. TGF-Autophagy Crosstalk Regulates Apoptosis in HTM Cells Autophagy and apoptosis have been previously demonstrated to either regulate each other or occur independently. Our data showed that 10 ng/mL TGF-β1 activates both autophagy and apoptosis and autophagy inhibition increased apoptosis suggesting that apoptosis is regulated by ATG5 in HTM cells and inhibition of autophagy causes apoptosis in HTM cells (Figure 3). ## 3.4. TGF Induced Autophagy Involves a Positive Feedback Loop of ATG5 on the TGF Pathway As shown in Figure 4, Smad3 inhibitor-treated cells had lower levels of autophagy proteins ATG5 and LC3-II. This also paralleled with reduced Smad3 protein levels in siATG5 transfected cells confirming that ATG5 regulates TGF-β1 induced EMT and fibrosis in HTM cells via a positive feedback loop of ATG5 involving TGF-β1 and SMAD signaling and downstream effects. ## 3.5. Crosstalk between ATG5 and TGF May Be Mediated by miR 122-5p In our previous study [1], we found three novel miRNAs (miR-122-5p, miR-124-3p, and miR-424-5p) that were involved in pathways, namely TGF-β1, fibrosis/ECM, and proteoglycan metabolism. Since ATG5 was found to regulate TGF-β1 induced EMT and fibrosis through the Smad3 pathway, we evaluated the expression of these three miRNAs in TGF-β1 treated and siATG5-inhibited HTM cells to see if the TGF and autophagy crosstalk involves miRNAs. We observed an increase in the expression of miR-122-5p and a decrease in miR-124-3p at 72 h in TGF-β1 stimulated cells (Figure 5A). This paralleled with a decrease in the expression of miR-122-5p at 72 h in siATG5 cells stimulated with TGF-β1 (Figure 5B) indicating the possibility that ATG5 could be modulating miR-122-5p which in turn modulates Smad3 signaling in HTM cells. This, however, requires further validation. ## 3.6. TGF-Induced ATG5 Activation Is Essential to Prevent TGF-Induced Aggregate Formation in HTM Cells Electron micrograph results confirmed the presence of protein aggregates occurring in 72h TGF-β1 treated cells (Figure 6). Immunofluorescence images of siATG5 HTM cells treated with TGF-β1 at 10 ng/mL for 72 h showed aggregate deposition, suggesting the role of autophagy in protein aggregation. Amyloid fibrills were restricted near the nucleus and oligomer 11 was spread throughout the cell and ECM (Figure 7). ATG-5 inhibition however resulted in increased protein aggregation implying that autophagy activation is crucial for aggregate clearance. The results also suggest that TGF possibly mitigates aggregate formation independently and clearance requires autophagy activation. ## 4. Discussion In this investigation, we found that TGF-β1 increases autophagy and the siRNA-mediated restriction of ATG5 led to a positive feedback loop repressing TGF-induced fibrosis and EMT. This feedback loop between ATG5 and TGF-β1 is possibly mediated via miR-122-5p and Smad3 pathway, which may be a therapeutic target to reduce EMT, fibrosis, and prevent or reduce aggregate formation in XFS. Autophagy is a biological process that provides a system for protein breakdown, which is necessary for tissue regeneration, cell survival, and homeostasis. Several studies have reported the crosstalk between autophagy and TGF-β1 in various diseases. TGF-induced fibrosis and EMT have been well studied in the pathophysiology of XFS and XFG [2,6]. Literature has recently focused on the interaction between autophagy, TGF, and the fibrotic response in various degenerative diseases [25,26,27,28,29,30,31]. Autophagy can either positively or negatively control fibrosis, depending on the cell type, tissue, or pathological circumstances. We found in this study that suppressing autophagy partially dramatically reduces TGF-β1-mediated Smad signaling and downstream effects on profibrotic factors in HTM cells. Furthermore, in siATG5-transfected cultures, the vimentin and SMA levels in response to exogenous TGF-β1 therapy were significantly reduced. Fibrosis has been linked to both autophagic upregulation and downregulation in many organs, highlighting the diversity of autophagy’s functional involvement in tissue repair. Consistent with the findings of our study, another study has reported that TGF-β1 promotes both COL1A2 synthesis and autophagy induction in human atrial myofibroblasts. Knocking out ATG5 in mouse embryonic fibroblasts also has been reported to result in a decrease in TGF-β1-induced fibrosis, when compared to wild-type cells, emphasizing the importance of autophagy in TGF-β1 induced fibrosis [32]. Interestingly, autophagy inhibition in human and murine macrophages resulted in a greater reduction in TGFB1 production [33]. Likewise, another group showed that the reduction in TGF-induced fibrosis was linked to genetic and pharmacological suppression of autophagy, which was generated by a BAMBI-mediated lower activation of Smad$\frac{2}{3}$ signaling in autophagy-deficient cells [34]. A feedback loop between ATG5 and TGF, as shown in this study, is a protective mechanism to control TGF-mediated side effects like EMT, cell death, and fibrosis, that may be detrimental in vivo in the HTM. Our earlier study showed that continued expression of TGF causes dysregulation of the downstream effects with enhanced cell death and preferential upregulation of specific profibrotic markers and miRNAs. While TGF is directly known to regulate ATG5 and autophagosome functioning, [3,25,26], this study suggests a dysfunctional feedback loop and aberrant crosstalk of TGF with ATG5, which may explain some of the pathogenic features seen in PXF. Eyes with XFS have flaky protein aggregate deposits that have been associated with increased TGF-β1 expression in XFS/XFG [1,2,7]. An in vitro model that involved costimulating retinal pigment epithelial (RPE) cell monolayers with TNF-α and TGF-β2 resulted in the formation of cellular aggregates [35]. Mutant TGF-β1was shown to enhance Aβ peptide aggregation in corneal dystrophy, suggesting the potential direct role of TGF in mediating aggregate formation [36]. In our previous study, we reported protein aggregates in HTM cells upon continued TGF-β1 exposure. We also showed a decreased UPR response in patients with pseudoexfoliation glaucoma, suggesting that reduced autophagy-induced UPR regulation was key in the formation of aggregates in the trabecular meshwork in advanced PXF disease. Therapeutic strategies targeting the formation or dissolution of aggregates in XFS/XFG may involve targets that mediate these feedback loops between TGF, ATG5, and the UPR pathway. A growing body of evidence suggesting the function of hsmiR-122-5p in downregulating the TGF–Smad pathway has been established. Further, miR-122 plays a role in skeletal muscle myogenesis by regulating the TGF–Smad pathway [37]. MicroRNA and miR-122-5p were found to be downregulated in skeletal muscle fibrosis as a result of TGF [38]. Findings of one study revealed that the miR-122–PKM2 autophagy axis protects hepatocytes against arsenite stress via the PI3K–Akt–mTOR pathway, suggesting that miR-122 could be a possibility for arseniasis treatment [39]. This study shows miR-122-5p upregulation upon TGF response that was downregulated in siATG5 knockdown cells, indicating the possibility that ATG5 feedback loop regulates TGF/Smad signaling via miR-122-5p. Our findings highlight the importance of autophagy and its link to the TGF pathway in the processes of protein aggregates in HTM cells. They also provide a new paradigm for understanding the causes of fibrosis and HTM cell dysfunction that are key to understanding the pathophysiology of glaucoma in XFS. Overall, the findings strongly suggest that autophagy and TGF play key balancing roles in HTM cell function and regulation of fibrotic response. Finally, these findings point to autophagy and aggregated proteins as potential new therapeutic targets for reversing or preventing HTM cell death in XFG. ## References 1. Rao A., Chakraborty M., Roy A., Sahay P., Pradhan A., Raj N.. Differential miRNA Expression: Signature for Glaucoma in Pseudoexfoliation. Clin. Ophthalmol. (2020) 14 3025-3038. DOI: 10.2147/OPTH.S254504 2. Sahay P., Reddy S., Prusty B.K., Modak R., Rao A.. TGFβ1, MMPs and cytokines profiles in ocular surface: Possible tear biomarkers for pseudoexfoliation. PLoS ONE (2021) 16. DOI: 10.1371/journal.pone.0249759 3. Heydarpour F., Sajadimajd S., Mirzarazi E., Haratipour P., Joshi T., Farzaei M.H., Khan H., Echeverría J.. Involvement of TGF-β and Autophagy Pathways in Pathogenesis of Diabetes: A Comprehensive Review on Biological and Pharmacological Insights. Front. Pharmacol. (2020) 11 498758. DOI: 10.3389/fphar.2020.498758 4. Ding Y., Choi M.E.. Regulation of autophagy by TGF-β: Emerging role in kidney fibrosis. Semin. Nephrol. (2014) 34 62-71. DOI: 10.1016/j.semnephrol.2013.11.009 5. Quan W., Lim Y.M., Lee M.S.. Role of autophagy in diabetes and endoplasmic reticulum stress of pancreatic β-cells. Exp. Mol. Med. (2012) 44 81-88. DOI: 10.3858/emm.2012.44.2.030 6. Chakraborty M., Sahay P., Rao A.. Primary Human Trabecular Meshwork Model for Pseudoexfoliation. Cells (2021) 10. DOI: 10.3390/cells10123448 7. Chakraborty M., Rao A.. Alternate Causes for Pathogenesis of Exfoliation Glaucoma, a Multifactorial Elastotic Disorder: A Literature Review. Curr. Issues Mol. Biol. (2022) 44 1191-1202. DOI: 10.3390/cimb44030078 8. Sharma S., Chataway T., Klebe S., Griggs K., Martin S., Chegeni N., Dave A., Zhou T., Ronci M., Voelcker N.H.. Novel protein constituents of pathological ocular exfoliation syndrome deposits identified with mass spectrometry. Mol. Vis. (2018) 24 801-917. PMID: 30713420 9. Wolosin J.M., Ritch R., Bernstein A.M.. Is Autophagy Dysfunction a Key to Exfoliation Glaucoma?. J. Glaucoma (2018) 27 197-201. DOI: 10.1097/IJG.0000000000000606 10. Rao A., Sahay P., Chakraborty M., Prusty B.K., Srinivasan S., Jhingan G.D., Mishra P., Modak R., Suar M.. Switch to Autophagy the Key Mechanism for Trabecular Meshwork Death in Severe Glaucoma. Clin. Ophthalmol. (2021) 15 3027-3039. DOI: 10.2147/OPTH.S292218 11. De Juan-Marcos L., Escudero-Domínguez F.A., Hernández-Galilea E., Cruz-González F., Follana-Neira I., González-Sarmiento R.. Investigation of Association between Autophagy-Related Gene Polymorphisms and Exfoliation Syndrome and Exfoliation Glaucoma in a Spanish Population. Semin. Ophthalmol. (2018) 33 361-366. DOI: 10.1080/08820538.2016.1247177 12. Want A., Gillespie S.R., Wang Z., Gordon R., Iomini C., Ritch R., Wolosin J.M., Bernstein A.M.. Autophagy and Mitochondrial Dysfunction in Tenon Fibroblasts from Exfoliation Glaucoma Patients. PLoS ONE (2016) 11. DOI: 10.1371/journal.pone.0157404 13. Ecker N., Mor A., Journo D., Abeliovich H.. Induction of autophagic flux by amino acid deprivation is distinct from nitrogen starvation-induced macroautophagy. Autophagy (2010) 6 879-890. DOI: 10.4161/auto.6.7.12753 14. Liang J., Shao S.H., Xu Z.X., Hennessy B., Ding Z., Larrea M., Kondo S., Dumont D.J., Gutterman J.U., Walker C.L.. The energy sensing LKB1-AMPK pathway regulates p27(kip1) phosphorylation mediating the decision to enter autophagy or apoptosis. Nat. Cell Biol. (2007) 9 218-224. DOI: 10.1038/ncb1537 15. Meijer A.J., Codogno P.. Autophagy: Regulation by energy sensing. Curr. Biol. (2011) 21 R227-R229. DOI: 10.1016/j.cub.2011.02.007 16. Nijholt D.A., de Graaf T.R., van Haastert E.S., Oliveira A.O., Berkers C.R., Zwart R., Ovaa H., Baas F., Hoozemans J.J., Scheper W.. Endoplasmic reticulum stress activates autophagy but not the proteasome in neuronal cells: Implications for Alzheimer’s disease. Cell Death Differ. (2011) 18 1071-1081. DOI: 10.1038/cdd.2010.176 17. Majmundar A.J., Wong W.J., Simon M.C.. Hypoxia-inducible factors and the response to hypoxic stress. Mol. Cell (2010) 40 294-309. DOI: 10.1016/j.molcel.2010.09.022 18. Scherz-Shouval R., Shvets E., Fass E., Shorer H., Gil L., Elazar Z.. Reactive oxygen species are essential for autophagy and specifically regulate the activity of Atg4. EMBO J. (2007) 26 1749-1760. DOI: 10.1038/sj.emboj.7601623 19. Levine B., Kroemer G.. Autophagy in the pathogenesis of disease. Cell (2008) 132 27-42. DOI: 10.1016/j.cell.2007.12.018 20. Das G., Shravage B.V., Baehrecke E.H.. Regulation and function of autophagy during cell survival and cell death. Cold Spring Harb. Perspect. Biol. (2012) 4 a008813. DOI: 10.1101/cshperspect.a008813 21. Chen N., Karantza V.. Autophagy as a therapeutic target in cancer. Cancer Biol. Ther. (2011) 11 157-168. DOI: 10.4161/cbt.11.2.14622 22. Uddin M.S., Stachowiak A., Mamun A.A., Tzvetkov N.T., Takeda S., Atanasov A.G., Bergantin L.B., Abdel-Daim M.M., Stankiewicz A.M.. Autophagy and Alzheimer’s Disease: From Molecular Mechanisms to Therapeutic Implications. Front. Aging Neurosci. (2018) 10 4. DOI: 10.3389/fnagi.2018.00004 23. Ravikumar B., Duden R., Rubinsztein D.C.. Aggregate-prone proteins with polyglutamine and polyalanine expansions are degraded by autophagy. Hum. Mol. Genet. (2002) 11 1107-1117. DOI: 10.1093/hmg/11.9.1107 24. Fortun J., Dunn W.A., Joy S., Li J., Notterpek L.. Emerging role for autophagy in the removal of aggresomes in Schwann cells. J. Neurosci. (2003) 23 10672-10680. DOI: 10.1523/JNEUROSCI.23-33-10672.2003 25. Xu Y., Yang S., Huang J., Ruan S., Zheng Z., Lin J.. Tgf-beta1 induces autophagy and promotes apoptosis in renal tubular epithelial cells. Int. J. Mol. Med. (2012) 29 781-790. PMID: 22322529 26. Bolt A.M., Klimecki W.T.. Autophagy in toxicology: Self-consumption in times of stress and plenty. J. Appl. Toxicol. (2012) 32 465-479. DOI: 10.1002/jat.1787 27. Poon A.H., Chouiali F., Tse S.M., Litonjua A.A., Hussain S.N., Baglole C.J., Eidelman D.H., Olivenstein R., Martin J.G., Weiss S.T.. Genetic and histologic evidence for autophagy in asthma pathogenesis. J. Allergy Clin. Immunol. (2012) 129 569-571. DOI: 10.1016/j.jaci.2011.09.035 28. Hernandez-Gea V., Ghiassi-Nejad Z., Rozenfeld R., Gordon R., Fiel M.I., Yue Z., Czaja M.J., Friedman S.L.. Autophagy releases lipid that promotes fibrogenesis by activated hepatic stellate cells in mice and in human tissues. Gastroenterology (2012) 142 938-946. DOI: 10.1053/j.gastro.2011.12.044 29. Lijnen P.J., Petrov V.V., Fagard R.H.. Induction of cardiac fibrosis by transforming growth factor-beta(1). Mol. Genet. Metab. (2000) 71 418-435. DOI: 10.1006/mgme.2000.3032 30. Zhang J., Jiang N., Ping J., Xu L.. TGF-β1-induced autophagy activates hepatic stellate cells via the ERK and JNK signaling pathways. Int. J. Mol. Med. (2021) 47 256-266. DOI: 10.3892/ijmm.2020.4778 31. Tabor-Godwin J.M., Tsueng G., Sayen M.R., Gottlieb R.A., Feuer R.. The role of autophagy during coxsackievirus infection of neural progenitor and stem cells. Autophagy (2012) 8 938-953. DOI: 10.4161/auto.19781 32. Ghavami S., Cunnington R.H., Gupta S., Yeganeh B., Filomeno K.L., Freed D.H., Chen S., Klonisch T., Halayko A.J., Ambrose E.. Autophagy is a regulator of TGF-[beta]1-induced fibrogenesis in primary human atrial myofibroblasts. Cell Death Dis. (2015) 6 e1696. DOI: 10.1038/cddis.2015.36 33. Porter K.M., Jeyabalan N., Liton P.B.. MTOR-independent induction of autophagy in trabecular meshwork cells subjected to biaxial stretch. Biochim. Biophys. Acta (2014) 1843 1054-1062. DOI: 10.1016/j.bbamcr.2014.02.010 34. Nettesheim A., Shim M.S., Hirt J., Liton P.B.. Transcriptome analysis reveals autophagy as regulator of TGFβ/Smad-induced fibrogenesis in trabecular meshwork cells. Sci. Rep. (2019) 9 16092. DOI: 10.1038/s41598-019-52627-2 35. Matoba R., Morizane Y., Shiode Y., Hirano M., Doi S., Toshima S., Araki R., Hosogi M., Yonezawa T., Shiraga F.. Suppressive effect of AMP-activated protein kinase on the epithelial-mesenchymal transition in retinal pigment epithelial cells. PloS ONE (2017) 12. DOI: 10.1371/journal.pone.0181481 36. Yam G.H., Wang K., Jhanji V., Choy K.W., Baum L., Pang C.P.. In vitro amyloid aggregate forming ability of TGFBI mutants that cause corneal dystrophies. Investig. Ophthalmol. Vis. Sci. (2012) 53 5890-5898. DOI: 10.1167/iovs.11-9068 37. Ding Z., Lin J., Sun Y., Cong S., Liu S., Zhang Y., Chen J.. miR-122-5p negatively regulates the transforming growth factor-β/Smad signaling pathway in skeletal muscle myogenesis. Cell Biochem. Funct. (2020) 38 231-238. DOI: 10.1002/cbf.3460 38. Sun Y., Wang H., Li Y., Liu S., Chen J., Ying H.. miR-24 and miR-122 Negatively Regulate the Transforming Growth Factor-β/Smad Signaling Pathway in Skeletal Muscle Fibrosis. Mol. Ther. Nucleic Acids (2018) 11 528-537. DOI: 10.1016/j.omtn.2018.04.005 39. Wang Y., Zhao H., Guo M., Fei D., Zhang L., Xing M.. Targeting the miR-122/PKM2 autophagy axis relieves arsenic stress. J. Hazard. Mater. (2020) 383 121217. DOI: 10.1016/j.jhazmat.2019.121217
--- title: 'Complementary Feeding Methods: Associations with Feeding and Emotional Responsiveness' authors: - Carla Fernandes - Fátima Martins - Ana F. Santos - Marília Fernandes - Manuela Veríssimo journal: Children year: 2023 pmcid: PMC10047322 doi: 10.3390/children10030464 license: CC BY 4.0 --- # Complementary Feeding Methods: Associations with Feeding and Emotional Responsiveness ## Abstract Learning to eat complementary foods is a crucial milestone for infants, having implications across development. The most used method for introducing complementary foods is Traditional Spoon-Feeding (TSF). However, the alternative method Baby-Led Weaning (BLW) is increasingly becoming used as it has been associated with positive outcomes. Research analyzing associations between complementary feeding methods and responsive parenting is practically non-existent. Therefore, the objective of this study was to analyze differences in emotional and feeding responsiveness between caregivers who previously implemented traditional vs. non-traditional feeding approaches. Caregivers (mostly mothers) of 179 children between 3 and 5 years were asked about the complementary feeding method that they had followed previously ($70.4\%$ reported using the TSF, $16.8\%$ said they used the BLW and $12.8\%$ used both methods simultaneously). In addition, they reported on their feeding practices using the Comprehensive Feeding Practices Questionnaire and on their responses to children’s distress using the Coping with Children’s Negative Emotions Scale. The results showed that parents who reported using a non-traditional (BLW or both) complementary feeding method reported less pressure to eat and minimization of reactions to children’s negative emotions, compared to parents who used a traditional method (although these reported using more problem-focused reactions). The findings suggest that complementary feeding methods and responsive parenting may be linked, leaving the question of which one sets the stage for the other. ## 1. Introduction It is during early childhood that food habits start to develop, making this a crucial moment to promote healthy food choices. The change from breast milk or formula to complementary foods is a crucial developmental milestone and is related with eating behaviors, food preferences and body weight across development [1,2,3,4,5]. ## 1.1. Complementary Feeding Methods The WHO’s most recommended complementary feeding method is Traditional Spoon-Feeding (TSF) [6,7]. In this method, infants are spoon-fed by caregivers, and the first solids offered are pureed foods, with gradual exposure to more varied textures and flavors over time, until family foods are introduced [8,9]. An alternative method named Baby-Led Weaning (BLW) has become increasingly widely used. According to Rapley [10], this method allows the infant to lead the weaning process, choosing what, when and how fast to eat. Within this approach, family foods are offered to the infant, in a texture and form that are adapted to the child’s developmental stage, for example, in the form of pieces (finger foods) that he/she can grasp with their hands and eat independently [11,12]. Thus, infants can experience and participate in family meals [10,13,14]. Occasionally, spoon-feeding or serving pureed foods may occur for up to $10\%$ of the total feeding time [1]. However, not all authors agree with this definition of BLW, suggesting that this method implies the infant self-feeds all or most of the time [15,16]. Although previous studies are limited in number and most of them are correlational, BLW has been associated with positive outcomes such as lower body mass index [17], preference for healthy foods [5], increased satiety responsiveness [1], a longer period of exclusive breastfeeding [8,15,18,19], children’s more frequent participation in family meals and enjoyment at mealtimes [8,15,19,20], lower maternal anxiety [2,17,19] and infant irritability [18,20]. In sum, in the TSF method, the caregivers’ control appears to be high, as they guide with a spoon the amount, speed, type and consistency of food given to the infant [19,21]. By contrast, in the BLW method, caregivers present a variety of solid foods, and the infant takes an active role, being allowed to self-feed and choose the food and quantities to eat [10]. Evidence suggests that there are differences between parents and children who follow each of these methods. Still, research has not yet been able to keep up with the growing interest in BLW, with many questions still open or to be clarified [2]. At this level, one question of interest is related to parents’ responsiveness to the child [22]. ## 1.2. Feeding and Emotional Responsiveness and Complementary Feeding Responsive parenting is characterized by the caregiver’s ability to adequately identify the different signals the child transmits and, in turn, give a developmentally appropriate response, in an emotionally supportive way, not intrusive or controlling [23]. Similarly, in the context of responsive feeding, the caregiver can recognize and respond appropriately to the child’s internal appetite cues, whereas, in non-responsive feeding, caregivers use excessive controlling and coercive feeding practices, failing to respond to these cues [23,24,25]. As a result, non-responsive feeding practices can undermine the child’s self-regulation of energy intake, increasing the risk of overweight or obesity development [23,26,27,28,29]. On the other hand, caregivers’ responsive feeding practices promote children’s ability to regulate their internal hunger and satiety cues [23]. Similarly, as previously referred to, evidence shows that BLW is also associated with greater satiety responsiveness, which could result from the opportunity infants have to regulate the amount of food they eat as they are the ones who feed themselves in this complementary method [1,9]. So, like responsive feeding, the BLW method increases children’s attention to hunger and satiety cues. In this sense, it is possible that BLW could be somehow related to responsive feeding practices or consist of a form of responsive feeding. However, to our knowledge, only one study has examined this link. In this study, Brown and Lee [30] found that mothers who used BLW reported less nonresponsive practices, namely, pressure to eat and restriction of food, compared with mothers who followed a more conventional method. Furthermore, when caregivers doubt children’s ability to learn to self-feed and consume enough food or feel stressed or pressured for some reason, they could end up dominating the feeding situation, turning to non-responsive feeding practices [23] and not endorsing the BLW method. Thus, caregivers may resort to a complementary food method in which their control is higher, such as the TSF method. In addition, emotions are strongly present in mealtimes [31]. In particular, the introduction of new flavors and consistencies can lead the child to demonstrate positive and negative emotions. Moreover, children seem to have a biological predisposition to be reluctant or refuse new foods [32]. Thus, this refusal could lead to tensions between caregivers and children, and mealtime may become stressful and prompt negative emotions. The way caregivers choose to deal with children’s emotions during feeding may impact both children’s emotion regulation and regulation of energy intake [31]. In fact, recent research shows that emotional responsiveness and feeding responsiveness are intertwined, with the use of unsupportive emotional responses (e.g., distress, punitive and minimization responses) being a risk factor for the use of non-responsive feeding practices (e.g., pressure to eat, restriction, food as a reward, and emotion regulation) [31,33,34,35,36]. Therefore, as emotional unresponsiveness could lead caregivers to exert excessive control in feeding, which, in turn, places children at risk for excessive weight gain [27], it is important to assess whether emotional responsiveness could have a role in complementary feeding. Evidence at this level is nonexistent or at least difficult to locate. ## 1.3. Current Study A small group of previous studies have highlighted the importance of emotional responsiveness to feeding responsiveness and, therefore, to children’s self-regulation of energy intake [31,33,34,35,36]. Additionally, there is also evidence associating BLW to children’s self-regulation of energy intake [1,37]. However, there is a dearth of literature about the relationship between emotional and feeding responsiveness and complementary feeding methods. Therefore, the main aim of this study is to analyze the differences in emotional and feeding responsiveness between caregivers who previously implemented the BLW or TSF method with their children. ## 2.1. Participants Participants were the caregivers (mostly mothers, $98.3\%$) of 179 children (104 boys, 75 girls), aged between 3 and 5 years old ($M = 51.5$ months; SD = 11.3) from Lisbon, Portugal. Most parents were either married or cohabitating ($83.7\%$). Mother’s age ranged between 23 and 48 years ($M = 36.7$; SD = 5.1), $42.5\%$ had a master’s degree ($30.2\%$ had high school diploma, $2.8\%$ had a bachelor’s, $20.7\%$ had an advanced professional degree and $1.7\%$ a PhD) and most of them ($78.8\%$) worked full-time ($12.3\%$ work part-time, and $8.4\%$ were unemployed). Father’s age ranged between 22 and 58 years ($M = 38.8$; SD = 5.9), $53.1\%$ had a high school diploma ($0.6\%$ bachelor, $27.4\%$ master, $8.4\%$ advanced professional degree and $4.5\%$ and a PhD), most fathers ($93.9\%$) work full-time ($1.7\%$ work part-time, and $2.2\%$ were unemployed). Children usually spend an average of 7 h at school (SD = 1.5), $53.7\%$ were firstborn and $67\%$ had siblings. Children’s weight varied between 10 kg and 32 kg ($M = 17.7$; SD = 3.7), and their height varied between 87 cm and 130 cm ($M = 106.1$; SD = 8.3). The beginning of the children’s food introduction ranged between 3 months and 8 months ($M = 5.3$; SD = 1.2). ## 2.2.1. Complementary Feeding Methods Caregivers were asked to retrospectively self-identify themselves as following a TSF, BLW or a mixed complementary feeding method. A definition of TSF and BLW was provided to caregivers. For parents that felt that they did not fit in either of these complementary feeding methods, an additional option was provided. In this, they were asked to describe the method they used. The mixed feeding method emerged after the analysis of caregivers’ responses corresponding to those who indicated to use a combination of both TSF and BLW methods. In order to double check this information, they were also asked to estimate the frequency of its use and the proportion of food that they provided as purées or spoon-fed to their children when they were babies. These notions are present among caregivers who perceive themselves as following a particular complementary feeding method, for example, BLW (e.g., [1,2,22]). ## 2.2.2. Parental Feeding Practices Caregivers’ feeding practices were assessed using the Comprehensive Feeding Practices Questionnaire (CFPQ) [38], a questionnaire composed of 49 items that are answered by caregivers using a 5-point rating scale, indicating their degree of agreement (1 = disagree, to 5 = agree; items 1–13) or their frequency of use a specific feeding approach (1 = never, to 5 = always; items 14–49). Items can be aggregated into 12 subscales reflecting distinct caregivers’ feeding practices. Six of these subscales reflect more positive or healthy feeding styles, namely: encourage balance and variety (α = 0.71), i.e., promoting of healthy and varied food consumption; environment (α = 0.63), i.e., providing healthy foods; involvement (α = 0.72), i.e., encouraging child’s involvement in food preparation and in meal planning; modelling (α = 0.63), i.e., being an active and enthusiastic model of healthy eating for the child; monitoring (α = 0.82), i.e., keeping track of child’s intake of unhealthy foods; and teaching about nutrition (α = 0.42), i.e., encouraging the child’s intake of healthy foods through didactic techniques. The other six subscales reflect unhealthy, emotion-related or pressuring feeding styles, including emotion regulation (α = 0.82), i.e., using food to regulate the child’s emotions; child control (α = 0.58), i.e., allowing the child to control their feeding interactions and own eating behaviors; food as reward (α = 0.51), i.e., using of food as a reward for child’s behavior; pressure (α = 0.70), i.e., encouraging the child to eat more food at meals, ignoring the child’s satiety/hunger cues; restriction for weight (α = 0.79), i.e., controlling the child’s intake to maintain or decrease the child’s weight; and restriction for health (α = 0.58), i.e., controlling the child’s intake to refers to parental control of child’s intake in order to limit unhealthy foods. Following Bost and colleagues [33], and excluding subscales with α < 0.60, two composites were generated: the pressuring feeding styles (average of emotion regulation, and pressure α = 0.74) and the healthy feeding styles (average of modeling, involvement, encourage balance and variety, and environment; α = 0.71). ## 2.2.3. Caregivers’ Responses to Children’s Negative Emotions Caregivers’ responsiveness to children’s negative emotions was assessed using the Coping with Children’s Negative Emotions Scale (CCNES) [39]. It includes 12 hypothetical scenarios, each with 6 possible and qualitatively different parental reactions to the child when he/she is upset or expressing negative emotions (e.g., “If my child becomes angry because he/she is sick or hurt and can’t go to his/her friend’s birthday party, I would…”). For each one, parents should indicate how likely they are to react in those specific ways using a 7-point rating scale (1 = very unlikely; 7 = very likely). These distinct parental reactions to control or regulate the child’s negative emotional expression correspond to 6 subscales. Three of them reflect negative reactions: punitive reactions (α = 0.78), involving caregivers’ use of punishment (verbal/physical); distress reactions (α = 0.53), reflecting caregivers’ discomfort; minimization reactions (α = 0.79), reflecting caregivers’ devaluation of the child’s problem or emotions. The other three reflect positive reactions: expressive encouragement (α = 0.87), reflecting caregivers’ acceptance and promotion of the child’s negative emotional expressions; emotion-focused reactions (α = 0.87), reflecting strategies used to help the child feel better; and problem-focused reactions (α = 0.80), reflecting strategies used to help the child to solve the problem that caused distress. Following Bost and colleagues [33], and excluding subscales with α < 0.60, two composites were generated: the negative emotion regulation (average of punitive and minimization reactions subscales; α = 0.86), and the positive emotion regulation (average of expressive encouragement, emotion-focused reactions, and problem focused reactions subscales; α = 0.91). ## 2.3. Procedures Data collection was carried out in schools in Lisbon ($26\%$) and online using Qualtrics during the 2021–2022 school year. We used convenience sampling. First, participants were presented with informed consent, informing the objectives of the study and anonymity. Additionally, before completing the questionnaires, they were asked to provide information regarding demographic data and the complementary feeding method previously adopted when the children were babies. This study was approved by the Ethics Committee (I/$\frac{038}{06}$/2020). ## 2.4. Analytic Plan Before our main analyses, descriptive statistics were explored. Normality and homoscedasticity of the variances were tested. ANOVAs were used to test for significant differences between caregivers’ who previously adopted distinct complementary food methods regarding demographics and variables in the study (i.e., caregivers’ feeding practices and emotional regulation strategies). In the variables in which the assumption of homoscedasticity was not verified, the statistical analysis was performed using the Welch correction of ANOVA. Hierarchical regression analysis was also performed. All statistics were run using the Statistical Package for the Social Sciences (SPSS, Version 28.0, Armonk, NY, USA). ## 3. Results Most of the participants ($70.4\%$) reported using the TSF method to introduce complementary food, $16.8\%$ said they used the BLW method and $12.8\%$ used both methods simultaneously (mixed method). For the following analyses, two groups were created: a traditional (which included caregivers who used the TSF method) and a mixed group (which included caregivers who used BLW exclusively and those who used it simultaneously with the TSF). There was a significant difference in mothers’ age (F[175,1] = 23.39, $p \leq 0.001$; $M = 37.84$, SD = 5.03 for traditional and $M = 34.02$, SD = 4.14 for mixed) and in fathers’ age (F[179,1] = 9.10, $p \leq 0.01$; $M = 39.65$, SD = 5.68 for traditional and $M = 36.73$, SD = 6.03 for mixed), with older parents using the traditional method more often. Children in the traditional group spend more time at school compared to the mixed group (Fwelch = 62.26, $p \leq 0.05$; $M = 7.57$, SD = 1.10 for traditional and $M = 6.90$; SD = 2.15 for mixed). A significant difference was also found regarding firstborns (χ2[1,179] = 6.09, $p \leq 0.01$, φ = −0.18), with more firstborns than expected in the mixed group. No other differences were found. ## 3.1. Parental Feeding Practices Depending on the Complementary Feeding Method As represented in Table 1, our results revealed significant differences in relation to pressure practices, depending on the method of complementary food introduction (F[1,178] = 5.00; $p \leq 0.05$). Specifically, parents who reported using a non-traditional complementary feeding method revealed using less pressure to eat ($M = 2.43$; SD = 0.98) compared to parents who used a TSF method ($M = 2.80$; SD = 0.97). A hierarchical regression analysis was performed with demographics on the first block and adding the feeding method on the second block. The first block was not significant. Adding the feeding method improved the model (ΔR = 0.02; ΔF = 3.91, $p \leq 0.05$), and this was the only one with a significant coefficient (β = −0.16, $p \leq 0.05$). ## 3.2. Parental Responses to Children’s Negative Emotions Depending on the Complementary Feeding Method Parents’ distress reactions were not included in the following analysis due to the poor alpha (α = 0.53). The results showed significant differences in minimization reactions depending on the complementary food introduction group ($F = 8.49$; $p \leq 0.01$). Specifically, parents who used the non-traditional method reported using fewer minimization reactions to children’s negative emotions ($M = 2.49$; SD = 0.87), compared to parents who used a non-traditional method ($M = 2.95$; SD = 0.86). Significant differences in emotion-focused reactions were also revealed ($F = 4.75$; $p \leq 0.05$). Specifically, parents who reported using a traditional complementary feeding method were found to have more problem-focused reactions ($M = 6.12$; SD = 0.77), compared with parents who used a non-traditional complementary feeding method ($M = 5.76$; SD = 1.08) (see Table 2). ## 4. Discussion Our study aimed to explore if the prior implementation of different methods of complementary food introduction by parents was related to differences in their emotional and feeding responsiveness. Specifically, we intended to explore how different they may be in terms of the emotional regulation strategies and feeding practices they report using with their children during the preschool years. The results show that caregivers who followed different complementary feeding methods tend to be different at several levels, for instance, in terms of their demographic characteristics (e.g., age). This is in line with previous evidence showing that parents who follow these approaches may be different [2,15,16,22]. Our findings also suggest that these differences may continue to be observed later, during preschool years, expanding previous knowledge by showing some differences in the way caregivers respond to children in feeding contexts and when children become upset. Regarding feeding responsiveness, the results indicate that, compared with the caregivers who followed a traditional method, those who followed a BLW/mixed approach are less likely to pressure their preschool children to eat more foods or specific foods, ignoring their signals of hunger and satiety. This is in line with the results of Brown and Lee [30] and is consistent with a philosophy in which children should not be forced to eat, respecting their needs in terms of feeding time and what and how they want to eat. This reflects a more responsive eating style, which can be evident since parents decide how to introduce complementary foods to their children [1,9,23]. If evidence relating different complementary feeding methods with responsive feeding is scarce, it is even more difficult to locate data considering emotional responsiveness at this level. Concerning emotional responsiveness to children’s negative emotions, our findings reveal that, compared with the caregivers who followed a traditional method, caregivers who followed a non-traditional approach are less likely to minimize or devaluate the child’s negative emotional expression and state. On the other hand, caregivers who followed a traditional method are more likely to use a problem-focused strategy to deal with preschoolers’ negative emotions, by helping the child to solve the problem that caused him/her distress. These findings suggest that both caregivers that followed a more standard vs. a less standard weaning method may later be responsive to their preschoolers’ negative emotions, albeit they may do it in different ways. In this sense, it is possible that caregivers’ who followed a baby-led related approach can be responsive to their children’s negative emotions in a more emotionally supportive manner [23], whereas caregivers who followed a traditional method can be responsive to their children’s negative emotions in a more practical supportive manner. Although these results are interesting and promising, they should be viewed with caution due to the nature and limitations of this study. This is a retrospective, comparative cross-sectional, exploratory study with a small sample size, which includes a limited number of participants who indicated they had followed the BLW method. Additionally, as in most studies, participants self-identify themselves as previously followed a specific complementary feeding method and were asked to estimate the frequency of its use and the proportion of food that they provided as purées or spoon-fed to their children when they were babies, with all the ambiguity this may imply [2,22]. Future studies should seek to overcome these limitations and will also benefit from a multi-informant (e.g., mothers and fathers), multi-method (e.g., self-reported and observational measures) and multi-context approach (e.g., family and schools). Specifically, longitudinal research is needed to disentangle associations between prior complementary feeding methods and later responsive parenting during preschool years. It is plausible to speculate that parents who followed a BLW/mixed method are more likely to use responsive practices, but it is also possible that responsive parenting was the primary reason why they had implemented this complementary feeding method previously and why they continue to respond to their children during preschool years using more responsive practices. Thus, prospective data will help to understand if a particular complementary feeding method could encourage/discourage some responsive/nonresponsive feeding and emotion regulation strategies or whether these parenting practices make more/less sense and became more/less attractive to parents depending on their degree of responsiveness [22,23]. Additionally, it is possible that no single method, even BLW that appears to be more responsive in its nature (e.g., [22]), is feasible or suitable for all children at all times, and being a responsive caregiver is also being able to choose what fits better the child’s characteristics and needs. A child’s temperament, feeding or weight problems, and experience in choking are some infant characteristics highlighted as playing a role in parents’ decision about a feeding method. Additionally, caregivers’ own characteristics (e.g., maternal anxiety), knowledge and understanding about the advantages and disadvantages of a particular feeding method, as well as fears and concerns (frequently related to the possibility of chocking, nutrient deficits, allergic reactions and lower energy intakes) may condition these choices [2,14,17,22,23,40,41]. Considering all these factors, caregivers can even opt for a mixed approach to the detriment of an exclusively traditional or BLW approach, taking advantage of the potential that each of these methods can offer to a given child in certain circumstances. Most important is that caregivers’ choices reflect responsive parenting [23]. Regardless of the limitations of the present study, our findings contribute to the advancement of knowledge about the associations between complementary feeding methods and responsive parenting. The evidence suggests that emotional responsiveness and feeding responsiveness are linked facets of responsive parenting [31,33,34,35,36], and we know that responsive feeding is a major factor enhancing healthy eating behaviors and reducing the risk for weight problems from early in life [22,42]. Thus, data generated at this level are useful for the design of preventive practices to reduce childhood obesity risk and contribute positively to helping caregivers, through psychoeducation. ## References 1. Brown A., Lee M.A.. Early influences on child satiety-responsiveness: The role of weaning style. Pediatr. Obes. (2015) 10 57-66. DOI: 10.1111/j.2047-6310.2013.00207.x 2. D’Auria E., Bergamini M., Staiano A., Banderali G., Pendezza E., Penagini F., Zuccotti G.V., Peroni D.G.. Baby-led weaning: What a systematic review of the literature adds on. Ital. J. Pediatr. (2018) 44 49. DOI: 10.1186/s13052-018-0487-8 3. Rose C.M., Birch L.L., Savage J.S.. Dietary patterns in infancy are associated with child diet and weight outcomes at 6 years. Int. J. Obes. (2017) 41 783-788. DOI: 10.1038/ijo.2017.27 4. Taylor R.W., Williams S.M., Fangupo L.J., Wheeler B.J., Taylor B.J., Daniels L., Fleming E.A., McArthur J., Morison B., Erickson L.W.. Effect of a baby-led approach to complementary feeding on infant growth and overweight: A randomized clinical trial. JAMA Pediatr. (2017) 171 838-846. DOI: 10.1001/jamapediatrics.2017.1284 5. Townsend E., Pitchford N.J.. Baby knows best? The impact of weaning style on food preferences and body mass index in early childhood in a case-controlled sample. BMJ Open (2012) 2 e000298. DOI: 10.1136/bmjopen-2011-000298 6. 6. World Health Organization Infant and Young Child Feeding: Model Chapter for Textbooks for Medical Students and Allied Health ProfessionalsWorld Health OrganizationGeneva, Switzerland2009Available online: https://apps.who.int/iris/handle/10665/44117(accessed on 1 December 2022). Infant and Young Child Feeding: Model Chapter for Textbooks for Medical Students and Allied Health Professionals (2009) 7. 7. World Health Organization UNICEF Global Strategy for Infant and Young Child FeedingWorld Health OrganizationGeneva, Switzerland2003Available online: https://www.who.int/publications/i/item/9241562218(accessed on 1 December 2022). Global Strategy for Infant and Young Child Feeding (2003) 8. Morison B.J., Taylor R.W., Haszard J.J., Schramm C.J., Erickson L.W., Fangupo L.J., Fleming E.A., Luciano A., Heath A.M.. How different are baby-led weaning and conventional complementary feeding? A cross-sectional study of infants aged 6–8 months. BMJ Open (2016) 6 e010665. DOI: 10.1136/bmjopen-2015-010665 9. Utami A.F., Wanda D., Hayati H., Fowler C.. “Becoming an independent feeder”: Infant’s transition in solid food introduction through baby-led weaning. BMC Proc. (2020) 14. DOI: 10.1186/s12919-020-00198-w 10. Rapley G.. Baby-led weaning: Transitioning to solid foods at the baby’s own pace. Community Pract. (2011) 84 20-23 11. Bocquet A., Brancato S., Turck D., Chalumeau M., Darmaun D., De Luca A., Feillet F., Frelut M.-L., Guimber D., Lapillonne A.. Committee on Nutrition of the French Society of Pediatrics. “Baby-led weaning”—Progress in infant feeding or risky trend?. Arch. Pediatr. (2022) 29 516-525. DOI: 10.1016/j.arcped.2022.08.012 12. Rapley G., Murkett T.. Baby-Led Weaning: Helping Your Baby to Love Good Food (2008) 13. Addessi E., Galloway A.T., Wingrove T., Brochu H., Pierantozzi A., Bellagamba F., Farrow C.V.. Baby-led weaning in Italy and potential implications for infant development. Appetite (2021) 164 105286. DOI: 10.1016/j.appet.2021.105286 14. Brown A., Jones S.W., Rowan H.. Baby-led weaning: The evidence to date. Curr. Nutr. Rep. (2017) 6 148-156. DOI: 10.1007/s13668-017-0201-2 15. Cameron S.L., Taylor R.W., Heath A.L.M.. Parent-led or baby-led? Associations between complementary feeding practices and health-related behaviours in a survey of New Zealand families. BMJ Open (2013) 3 e003946. DOI: 10.1136/bmjopen-2013-003946 16. Pérez-Ríos M., Santiago-Pérez M.I., Butler H., Mourino N., Malvar A., Hervada X.. Baby-led weaning: Prevalence and associated factors in Spain. Eur. J. Pediatr. (2020) 179 849-853. DOI: 10.1007/s00431-020-03579-7 17. Utami A.F., Wanda D.. Is the baby-led weaning approach an effective choice for introducing first foods? A literature review. Enferm. Clin. (2019) 29 87-95. DOI: 10.1016/j.enfcli.2019.04.014 18. Fu X., Conlon C.A., Haszard J.J., Beck K.L., von Hurst P.R., Taylor R.W., Heath A.M.. Food fussiness and early feeding characteristics of infants following baby-led weaning and traditional spoon-feeding in New Zealand: An internet survey. Appetite (2018) 130 110-116. DOI: 10.1016/j.appet.2018.07.033 19. Brown A., Lee M.A.. Descriptive study investigating the use and nature of baby-led weaning in a UK sample of mothers. Matern. Child Nutr. (2011) 7 34-47. DOI: 10.1111/j.1740-8709.2010.00243.x 20. Komninou S., Halford J.C.G., Harrold J.A.. Differences in parental feeding styles and practices and toddler eating behaviour across complementary feeding methods: Managing expectations through consideration of effect size. Appetite (2019) 137 198-206. DOI: 10.1016/j.appet.2019.03.001 21. Poniedziałek B., Paszkowiak N., Rzymski P.. Baby-Led-Weaning (BLW) from maternal perspective: Polish experience. J. Med. Sci. (2018) 87 75-81. DOI: 10.20883/jms.2018.269 22. Cameron S.L., Heath A.L., Taylor R.W.. How feasible is Baby-led Weaning as an approach to infant feeding? A review of the evidence. Nutrients (2012) 4 1575-1609. DOI: 10.3390/nu4111575 23. Black M.M., Aboud F.E.. Responsive feeding is embedded in a theoretical framework of responsive parenting. J. Nutr. (2011) 141 490-494. DOI: 10.3945/jn.110.129973 24. DiSantis K.I., Hodges E.A., Johnson S.L., Fisher J.O.. The role of responsive feeding in overweight during infancy and toddlerhood: A systematic review. Int. J. Obes. (2011) 35 480-492. DOI: 10.1038/ijo.2011.3 25. Hodges E.A., Johnson S.L., Hughes S.O., Hopkinson J.M., Butte N.F., Fisher J.O.. Development of the responsiveness to child feeding cues scale. Appetite (2013) 65 210-219. DOI: 10.1016/j.appet.2013.02.010 26. Daniels L.A.. Feeding practices and parenting: A pathway to child health and family happiness. Ann. Nutr. Metab. (2019) 74 29-42. DOI: 10.1159/000499145 27. Faith M.S., Scanlon K.S., Birch L.L., Francis L.A., Sherry B.. Parent-child feeding strategies and their relationships to child eating and weight status. Obes. Res. (2004) 12 1711-1722. DOI: 10.1038/oby.2004.212 28. Hurley K.M., Cross M.B., Hughes S.O.. A systematic review of responsive feeding and child obesity in high-income countries. J. Nutr. (2011) 141 495-501. DOI: 10.3945/jn.110.130047 29. Larsen J.K., Hermans R.C., Sleddens E.F., Engels R.C., Fisher J.O., Kremers S.S.. How parental dietary behavior and food parenting practices affect children’s dietary behavior. Interacting sources of influence?. Appetite (2015) 89 246-257. DOI: 10.1016/j.appet.2015.02.012 30. Brown A., Lee M.A.. Maternal control of child feeding during the weaning period: Differences between mothers following a baby-led or standard weaning approach. Matern. Child Health J. (2011) 15 1265-1271. DOI: 10.1007/s10995-010-0678-4 31. Frankel L.A., Hughes S.O., O’Connor T.M., Power T.G., Fisher J.O., Hazen N.L.. Parental influences on children’s self-regulation of energy intake: Insights from developmental literature on emotion regulation. J. Obes. (2012) 2012 327259. DOI: 10.1155/2012/327259 32. Pliner P., Hobden K.. Development of a scale to measure the trait of food neophobia in humans. Appetite (1992) 19 105-120. DOI: 10.1016/0195-6663(92)90014-W 33. Bost K.K., Wiley A.R., Fiese B., Hammons A., McBride B.. Associations between adult attachment style, emotion regulation, and preschool children’s food consumption. J. Dev. Behav. Pediatr. (2014) 35 50-61. DOI: 10.1097/01.DBP.0000439103.29889.18 34. Fernandes C., Santos A.F., Fernandes M., Santos A.J., Bost K., Veríssimo M.. Caregivers’ perceived emotional and feeding responsiveness toward preschool children: Associations and paths of influence. Nutrients (2021) 13. DOI: 10.3390/nu13041334 35. Saltzman J.A., Pineros-Leano M., Liechty J.M., Bost K.K., Fiese B.H.. Eating, feeding, and feeling: Emotional responsiveness mediates longitudinal associations between maternal binge eating, feeding practices, and child weight. Int. J. Behav. Nutr. Phys. Act. (2016) 13 89. DOI: 10.1186/s12966-016-0415-5 36. Saltzman J.A., Liechty J.M., Bost K.K., Fiese B.H.. Parent binge eating and restrictive feeding practices: Indirect effects of parent’s responses to child’s negative emotion. Eat. Behav. (2016) 21 150-154. DOI: 10.1016/j.eatbeh.2016.02.001 37. Arden M., Abbott R.L.. Experiences of baby-led weaning: Trust, control and renegotiation. Matern. Child Nutr. (2014) 11 829-844. DOI: 10.1111/mcn.12106 38. Musher-Eizenman D., Holub S.. Comprehensive Feeding Practices Questionnaire: Validation of a new measure of parental feeding practices. J. Pediatr. Psychol. (2007) 32 960-972. DOI: 10.1093/jpepsy/jsm037 39. Fabes R.A., Eisenberg N., Bernzweig J.. Coping with Children’s Negative Emotions Scale (CCNES): Description and Scoring (1990) 40. Arias-Ramos N., Andina-Díaz E., Granado-Soto M., Álvarez Rodríguez R., Liébana-Presa C.. Baby-led weaning: Health professionals ‘knowledge and attitudes and parents’ experiences from Spain. A mixed methods approach. Health Soc. Care Community (2022) 30 1352-1363. DOI: 10.1111/hsc.13543 41. Białek-Dratwa A., Soczewka M., Grajek M., Szczepańska E., Kowalski O.. Use of the baby-led weaning (BLW) method in complementary feeding of the infant—A cross-sectional study of mothers using and not using the BLW method. Nutrients (2022) 14. DOI: 10.3390/nu14122372 42. Schwartz C., Scholtens P.A., Lalanne A., Weenen H., Nicklaus S.. Development of healthy eating habits early in life. Review of recent evidence and selected guidelines. Appetite (2011) 57 796-807. DOI: 10.1016/j.appet.2011.05.316
--- title: Effects of Socioeconomic Environment on Physical Activity Levels and Sleep Quality in Basque Schoolchildren authors: - Arkaitz Larrinaga-Undabarrena - Xabier Río - Iker Sáez - Aitor Martinez Aguirre-Betolaza - Neritzel Albisua - Gorka Martínez de Lahidalga Aguirre - José Ramón Sánchez Isla - Mikel Urbano - Myriam Guerra-Balic - Juan Ramón Fernández - Aitor Coca journal: Children year: 2023 pmcid: PMC10047327 doi: 10.3390/children10030551 license: CC BY 4.0 --- # Effects of Socioeconomic Environment on Physical Activity Levels and Sleep Quality in Basque Schoolchildren ## Abstract The socioeconomic and built environment of an area are interrelated with health data and have a direct influence on children’s development. There are facilitators and barriers for schools to promote physical activity depending on the socioeconomic status of the school. The aim of this study was to analyse the relationship between physical activity and sleep and the socioeconomic level of children in the Basque Country. The sample consisted of 1139 schoolchildren between the ages of six and seventeen (566 boys and 573 girls) from 75 schools (43 public and 32 private). Differences between groups were compared using the Mann–Whitney U test (two samples), Kruskal–Wallis one-factor ANOVA (k samples), and Spearman’s Rho correlation. There are sex differences in light (200.8 ± 62.5 vs. 215.9 ± 54.7) and moderate (69.0 ± 34.3 vs. 79.9 ± 32.1) physical activity in favour of the female group of higher socioeconomic status compared to male group of higher socioeconomic status. In the case of vigorous physical activity, the female group performed less than the male group across all socioeconomic statuses, which was statistically significant in the groups of high socioeconomic status (11.6 ± 9.3 vs. 6.9 ± 5.7) in group 2 and medium socioeconomic status (11.1 ± 9.3 vs. 7.7 ± 6.1) in group 3. There is an inverse relationship between sedentary behaviour and BMI, total bed time, total sleep time, and night-time awakenings. There is also an inverse relationship between all levels of physical activity performed with respect to BMI and total sleep efficiency. These data point towards notable inequalities in physical activity and daily sleep in Basque schoolchildren, which in turn may be marginalised in our current school system due to the effects of the socioeconomic environment. ## 1. Introduction The health of individuals and populations is determined by a set of factors that go far beyond those of a biomedical nature, i.e., the genetic load and biological characteristics of individuals and their interaction with their environment [1]. On the contrary, it is living and working conditions, and other more structural factors such as the characteristics of the social, economic, and political context, which we call social determinants of health, that have a more direct influence on the healths of individuals and populations in our societies [2,3]. The fact that these social determinants of health are unequally distributed in the population generates social inequalities in health, i.e., systematic differences in health between people of different social levels, sex, ethnicity or place of residence, among other factors, meaning that the most disadvantaged groups systematically present a worse state of health. Therefore, equity in health is conditioned by the so-called structural determinants and intermediate determinants [1]. The former include aspects related to the socioeconomic and political context, which refer to the characteristics of the social structure of a society. These contextual factors exert a strong influence on patterns of social stratification, which determine the social position that people occupy in society according to their socioeconomic status, sex, level of education, place of birth, and other dimensions [3,4]. This unequal social position in turn generates inequalities in the distribution of intermediate determinants, which include living and working conditions, psychosocial factors such as the extent and quality of social networks, stress and perceived control over one’s life, and health-related behaviour such as alcohol consumption, smoking, diet, and physical activity (PA) [5]. The practice of PA and exercise has multiple health benefits: it is associated with reduced all-cause mortality, improved health-related quality of life (vitality, general health, and mental health) [6], and a reduced risk of adult-onset diabetes, obesity, osteoporosis, some cancers, and cardiovascular disease [7,8,9,10]. Accessibility and shorter distance to environments associated with physical exercise are known to increase the frequency of exercise [11,12,13,14,15]. According to ecological models, the built environment exerts a crucial influence on PA behaviour [16]. This is supported by several systematic reviews showing that people living in walkable, safer, and greener neighbourhoods tend to have higher levels of PA [17,18,19,20,21,22]. One’s socio-economic position, at the individual or area level, and the built environment are interrelated, and the path for mediation and moderation should be considered when related to health outcomes [23]. The influence of the proximity of health-related facilities on the practice of physical exercise can occur in two ways; the first has to do with the influence of seeing people in the vicinity performing physical exercise, which translates into its perception as a positive social norm [14]. The second is that one of the reasons most frequently given for abandoning physical exercise is related to the distance to the environment in which it is practiced, so the proximity of these infrastructures can eliminate the physical and psychological barriers and increase the frequency of physical exercise [14]. The importance of sleep for health at all stages of life has been widely demonstrated, but it is of vital importance in childhood and adolescence [24,25]. A deficiency in the quality of sleep can affect school performance and one’s appropriate PA levels, result in health disorders, etc. [ 26]. Different studies have considered the hypothesis of bidirectionality between PA and sleep, more specifically, that appropriate levels of PA are associated with better sleep efficiency and thus, deficiencies in sleep lead to lower PA levels [27]. Different studies [28,29,30] report the difficulties of children and adolescents and how it is necessary to deepen the study of these variables due to the importance of childhood and adolescence as stages of development and adherence to behaviours that will lead to future healthy adults. The socioeconomic variable has a direct influence on children’s development [31,32,33,34]. Socioeconomic inequalities may cause altered sleep and PA patterns [35], as there is evidence linking family socioeconomic status with sleep quality [36] and PA performance [35,37]. A low socioeconomic status affects children’s health and can have lifelong consequences, in both early and later life [38,39]. In addition, physical activity is not equally distributed across social classes, with the most disadvantaged performing less exercise during both adolescence [40] and adulthood [41]. Current policies are clear about the importance of sport and PA: sport is healthy and there is still much to be achieve in the area of healthy lifestyles [42]; however, in low-, middle-, and high-income countries, PA levels are still insufficient [31,43]. Moreover, few low- and middle-income countries have PA policies, while policies have been well developed in many high-income countries, although often with very limited implementation [44]. In addition, for many families today, PA involves a financial cost and family time that not everyone has the resources to afford [45]. In this situation, it is easier to choose other solutions to introduce PA, such as free school sports and informal PA in public spaces such as parks [46,47]. However, these children are more likely to engage in sedentary activities, such as playing video games instead of PA, which can lead to poor physical fitness in adolescence [48]. Contemporary global socio-cultural and physical environments are generally not conducive to high levels of regular PA among children and adolescents [31]. Instead, these have produced abnormal activity habits and social norms, limited access to the means of fulfilling their basic biological needs, and denied children the human right to physically active games [49]. In this regard, children of lower socioeconomic status residing in different European countries are less likely to participate in sports clubs, more likely to spend more than two hours a day in front of a screen [47], and less likely to spend more than two hours in an open environment on weekends compared to those of higher socioeconomic status [50]. The environmental changes that have reduced PA among children and adolescents have also led to biological changes, including reduced motor competence, reduced physical fitness, and high body fatness, even among those who are not overweight or obese as defined by their body mass index. In turn, these biological changes have further reduced the PA by producing feedback loops that amplify adverse environmental impacts on PA [51,52]. In the United States, there is a direct relationship between family wealth and the ability to participate in organised sports [29]; therefore, the children of families of higher socioeconomic status are more likely to meet the recommended levels of PA and sports participation [39]. Families of a higher socioeconomic status usually have more financial resources for their children to engage in extracurricular activities and may know more about the importance of the impact of PA on health. Therefore, it is easier to encourage these parents to actively participate in sports clubs [45,47]. In summary, children of high socioeconomic status have higher levels of PA [33,39,48]. In addition, children’s access to green spaces in urban areas is closely related to their physical and psychological well-being [53]. The size of the spaces available for play in the school environment directly influences the PA to be practiced. Schools of higher socioeconomic status will have more learning materials and equipment at their disposal, such as balls, ropes, or other materials for children to play with. At the same time, this influences the PA to be practiced at recess. Sports fields, green areas, trees, games, concrete, and shaded areas make it easier to perform PA, and in less discriminatory centres, the children obtained higher caloric expenditure in the games [54]. In order to have the capacity to improve the PA levels of children, centres must give importance to the provision of resources for PA, contributing to the creation of a culture of PA [55]. In relation to this, there are facilitators and barriers for schools to promote PA and raise the levels of PA among children, with some predominant facilitators or barriers depending on the socioeconomic level of the school [55,56]. In primary schools of high socioeconomic level, the lack of barriers related to the curriculum, teacher proficiency, and the intrinsic factors of individual pupils mean that schoolchildren have higher levels of PA [56]. At the compulsory secondary education stage, on the other hand, schools of low socioeconomic level have more barriers than high socioeconomic schools, such as those related to school policy, environment, and individual students’ intrinsic factors [57]. This leads to the fact that socioeconomic disadvantages among girls predict negative knowledge and achievement outcomes [38,58]. For all these reasons, the aim of this study was to analyse the relationship between PA and sleep and the socioeconomic level of girls and boys between 6 and 17 years of age in the Basque Country. ## 2.1. Subjects and Design A cross-sectional observational study was carried out. Participants were selected by non-probabilistic convenience sampling across all schools in the Basque Country. The study sample consisted of 1139 schoolchildren between the ages of six and seventeen (566 boys and 573 girls) from 75 schools (43 public and 32 private) that gave their definitive approval to the study. A proportional and random stratification according to historical territory (Araba, Bizkaia, and Gipuzkoa), sex, age (primary education from 6 to 12 years and secondary education from 12 to 17 years) and ownership of the school (public or private) was taken into account. The public schools were $100\%$ publicly funded while the private schools analysed in this study were co-financed, wherein the education was free but other services such as canteens, transport, materials, and other school activities were not. The qualitative variables of the study were sex, educational stage (primary education and compulsory secondary education), school, and socioeconomic level (SEP). On the other hand, the quantitative variables were body mass index (BMI), PA levels (light, moderate, vigorous, and MVPA), sedentary behaviour (min), total time in bed (min), total sleep time (min), night-time awakenings (min), and sleep efficiency (%). ## 2.2. Instruments The ActiGraph WGT3X-BT accelerometer (manufacturer ActiGraph, 49 East Chase St. Pensacola, FL, USA) was used to collect data related to PA levels as well as sleep parameters. The participants wore the accelerometer for seven consecutive days including a weekend. The device was worn on the non-dominant hand. Recordings were considered valid with a minimum daily exposure of 10 h for at least 3 days, among which at least 2 have to be working days and one on the weekend. In addition, it was requested that the accelerometer be removed during bathing, showering, and/or other water-based activities. They were collected based on the validity and reliability of previous studies [59,60,61,62]. ## 2.3. Procedure In order to carry out the research, approval was requested from the Basque Medicines Research Ethics Committee (Basque Government Department of Health) in accordance with the Law $\frac{14}{2007}$ on biomedical research [63], the ethical principles of the Helsinki Declaration of 2013 [64], and other ethical principles and applicable legislation in the report of the Basque Medicines Research Ethics Committee (CEIm-E) of the Basque Government Department of Health with internal code PI2020011. Likewise, the current regulations on personal data protection were respected: namely (EU) Regulation $\frac{2016}{679}$ of 27 April 2016 (GDPR) [65], Organic Law $\frac{3}{2018}$ of 5 December on Personal Data Protection and guarantee of digital rights (ES) [66], and Royal Decree (ES) $\frac{1720}{2007}$ of 21 December [67]. In all these documents and permits, it was taken into account that the study included school-aged children. Following the approval of the project, the Department of Education of the Basque Government sent an e-mail to all schools in the Basque Country (Figure 1). Subsequently, positive responses were collected from the schools interested in participating and meetings were held with the school’s management teams as well as with the physical education teachers through whom the families were provided with the information and documentation of this study. Among all the families that were willing to participate, participants were selected by means of a draw among those who met the selection criteria established for this study. Then, all the children’s legal guardians signed the informed consent form and the pupils themselves signed their informed consent. Once the participants by school, grade, and sex had been confirmed, a timetable was established for placing and removing the accelerometers. There is a proportional and random stratification according to the historical territory and county, age, sex, educational network (public or private), and SEP index (socioeconomic level based on the deprivation index in the census section, which makes it possible to identify sections with socioeconomic conditions), together with inclusion and exclusion criteria that can be seen in Table 1. In this study, for the determination of socioeconomic status, hereafter SEP, the MEDEA classification was used [68], which is a socioeconomic level that calculates the average per capita income of people living in the district in which the school is located, divided into five groups (see Table 2). ## 2.4. Statistical Analysis For the outcome variables, descriptive statistics were used, reporting the level of significance for the main group (between participants). To avoid a type I error, a post hoc analysis was performed when the interaction effect was found to be significant. Values will be expressed as the mean (SD). Statistical analysis was performed with SPSS software (version 28.0.1.0; IBM Corp; Armonk, New York, NY, USA). Values of $p \leq 0.05$ were considered statistically significant. First, the Kolmogorov–Smirnov test was used to assess the normality of the distribution and Levene’s test to observe the homogeneity of variances, as well as an analysis of the descriptive variables studied (means, standard deviation, etc.). None of the variables studied met the above requirements, so the differences between the groups were compared using the non-parametric Mann–Whitney U test (2 samples) and the Kruskal–Wallis one-factor ANOVA (k samples). After a significant Kruskal–Wallis H test, a Dunn–Bonferroni test was used for pairwise post hoc comparisons. Correlation between the variables was estimated using Spearman’s Rho. ## 3. Results The descriptive results of the sleep variables (Table 3, no statistical differences) and PA variables (Table 4) divided by sex and in each of the SEP categories are shown below. A sex difference was observed in the performance of light and moderate PA between the SEP 1 female and male groups. In the case of vigorous PA, the female group performed less PA than males in all groups, which was a statistically significant difference between SEP groups 2 and 3. Table 5 shows the values for the sleep parameters in the primary and secondary variables. Thus, in the primary stage, both the female and male groups have better sleep efficiency, spend less time in bed, and have fewer night-time awakenings at the two extremes of the SEP (groups 1 and 5) compared to the intermediate levels. At the secondary stage, females spend more time in bed in SEP group 1 than in the other groups, with a significant difference compared to groups 3 and 5. Table 6 shows the results of the PA parameters in both the primary and secondary school stages. In the primary stage, the male SEP group 5 showed greater sedentary behaviour than the rest of the groups. On the other hand, in the female group at the primary stage, although all those in the female SEP group 5 showed greater sedentary behaviour than the rest of the groups, significant differences can only be seen with groups 3 and 4. At the primary stage, intermediate male groups 2, 3, and 4 showed the highest levels of PA with respect to groups 1 and 5. As for the secondary stage, the schoolchildren who showed the greatest sedentary behaviour, amongst both males and females, belong to group 1. In addition, the female SEP group 5 of this stage was that with the least moderate PA and MVPA compared to the rest of the groups. Table 7 and Table 8 show the results with respect to the school ownership variable. Specifically, Table 7 shows the results for BMI and sleep quality. Among the public school students, males in SEP group 4 have the highest BMI among the 5 groups. In addition, males in SEP group 1 have the highest sleep efficiency, with fewer night-time awakenings. On the other hand, regardless of sex, males in SEP group 3 spend the most time in bed, sleep the longest, and have the most night-time awakenings. Among private school children, females in SEP group 2 show lower BMI values compared to the other groups. In addition, males in SEP group 1 show lower values for time spent in bed and time spent asleep compared to groups 2 and 4. Table 8 shows the daily PA values within the school ownership variable (public and private) analysed by sex in terms of the PA variables. Male and female schoolchildren in public schools, specifically those belonging to SEP group 4, showed the highest levels of sedentary behaviour with respect to the rest of the groups, which was statistically significant in females with respect to the rest of the groups. These values of sedentary behaviour are reflected in the higher BMI (see Table 7) in both male and female schoolchildren, although the latter is not statistically significant. In addition, the male SEP group 4 of the public centres were those performing the least vigorous PA. In private centres, it is both male and female SEP groups 1 and 5 who show the highest values of sedentary behaviour. On the other hand, those in SEP group 2 show the lowest values of sedentary behaviour and the highest values of moderate, vigorous, and MVPA compared to the rest of the groups. The female schoolchildren in SEP group 2 in these private centres also have the highest levels of light, moderate, and MVPA PA compared to the rest of the groups. Conversely, in females from public centres, the values of light, moderate, vigorous, and MVPA are lower in SEP group 4 than in groups 1 and 3. Table 9 shows the correlation values of the total sample between all the variables analysed. Thus, there is an inverse relationship between sedentary behaviour and BMI, total bed time, total sleep time, and WASO. There is also an inverse relationship between all levels of PA performed with respect to BMI and total sleep efficiency. ## 4. Discussion The aim of this study was to analyse the relationship between PA and sleep with the socioeconomic level of girls and boys between 6 and 17 years of age in the Basque Country. A meta-analysis on the sleep–obesity relationship in children and adolescents [69] emphasised a research recommendation on the interaction of demographic factors with sleep and obesity. MVPA, sedentary behaviour and demographic information, such as sex, age, and parental education level, were included as covariates because they were reported to influence the weight status and sleep of children and adolescents [70,71]. A short sleep duration is associated with an increased risk of overweight/obesity in children and adolescents in a study performed in China, independently of sleep quality. This relationship is significant for children rather than adolescents. Short sleep duration and sleep quality were significantly associated with overweight/obesity in girls, but not in boys in the same study [72,73]. Considering sleep quality, some studies have found that children and adolescents with poor sleep quality are more likely to gain weight [70,74], while other studies found no significant relationship between the two variables [75,76]. In our case, children have better sleep efficiency, spend less time in bed, sleep less, and have fewer night-time awakenings, which was true for both males and females of primary school age in the SEP extremes (groups 1 and 5) compared to schoolchildren in groups 2, 3, and 4. Females who spend more time in bed are those in a high SEP (group 1), which was a significant difference compared to females in the lowest SEP (group 5). Schoolchildren with lower SEP have a higher PA than their peers with higher SEP (particularly because of their greater participation in more active transportation, more household chores, and more work-related activities), whilst the latter participated more frequently in organised sports and formal activities [77]. However, for the most part, a high SEP is associated with a considerable increase in the frequency of PA [33,77,78,79,80,81,82]. In primary school, children from middle- and high-income families are approximately three times more likely to meet PA recommendations compared to children from low SEP families. For secondary school students, children from middle-income families are twice as likely and those from high-income families are more than three times as likely to ever participate in sports compared to children from low SEP families [39]. This relationship may be due to both the cost of access to sports practices being a barrier for families with lower purchasing power [83]. However, lower wealth is also associated with children’s participation in optimal amounts of health-promoting PA [39]. The research evidence suggests that PA behaviours are socioeconomically shaped, as children with a low socioeconomic status spend less time being physically active during leisure time and engage in less vigorous intensity activities compared to their peers of high economic status [84,85]. In our case, in public centres, males in SEP 4 were those performing the least vigorous PA, and females had the lowest PA values in light, moderate, vigorous, and MVPA, compared to groups 1 and 3. Furthermore, in private centres, both males and females in SEP 2 are those showing the highest PA values in moderate, vigorous, and MVPA. High-income children engage in significantly more vigorous activity than low- and middle-income children do [85]. These domain-specific and intensity-specific differences are important, as vigorous PA is considered to elicit stronger health benefits compared to lower intensity PA [86]. Among most northern, eastern, and southern European countries, children with low parental education played actively/vigorously for longer periods of time [33]. Meanwhile, the opposite situation emerged among Central Asian countries. An inverse socioeconomic gradient also emerges in relation to the practice of sports, with lower SEP children being less involved in these activities. On average, $70.9\%$ of children from families with low parental education level spend less than 2 h/week doing sport compared to $38.2\%$ of children with high parental education level [45]. The higher the level of education, the higher the educational climate and the higher the likelihood of performing PA [87]. In Canada, SEP at the area level was not related to step count or the amount of time children spent performing MVPA [50]. However, another study found that children in areas with higher SEP were more likely to comply with the daily step count recommendations [81]. In our case, in the primary stage, male schoolchildren with a lower SEP showed greater sedentary behaviour compared to the rest of the categories of the index. In contrast, in females of lower SEP in the primary stage, even though they showed greater sedentary behaviour than categories 3 and 4, there were no significant differences with the highest SEP categories (SEP 1 and 2), coinciding with another study that found that the amount of time spent performing MVPA (moderate to vigorous PA) was not statistically different for children from low-, medium-, or high-SEP households [88]. In Australia, children studying in schools of high socioeconomic status are more likely to meet the recommended PA levels as well as healthy cardiorespiratory fitness levels compared to children studying in schools with a low socioeconomic status. In the secondary stage, however, there were no significant differences [39], unlike in the present study, where, in the secondary stage, the schoolchildren who showed greater sedentary behaviour, in both males and females, belonged to the highest SEP (SEP 1). In addition, females with the lowest SEP (SEP 5) at this stage are those with the least moderate PA and MVPA. In turn, most of the barriers are related to curriculum, teaching and school policy, and environment [83]. In the present study, male and female public school children, specifically those belonging to low SEPs, show the highest levels of sedentary behaviour compared to other deprivation levels. In addition, low-SEP males in public schools are the least likely to engage in vigorous PA. Consistently with the results, which indicate that children from low-SEP families may be more likely to engage in sedentary activities, such as screen use, rather than recreational and physical activities, this in turn may lead to poor physical fitness in adolescence [48]. For families of low SEP, the lack of resources to enrol their children in a sporting activity (e.g., football, judo, gymnastics, jazz, ballet, tennis, etc.) could play an important role [47]. Young people of higher SEP are those who dedicate more time to sedentary behaviour, while those of families with lower incomes practice PA in a much less relevant way [89], coinciding with the results obtained in the private centres in our study, that, in an almost opposite way to the results obtained in the public centres, male and female schoolchildren and females with higher SEP (SEP 1) are those that reflect a higher level of sedentary behaviour. However, in another study, there was no clear pattern in terms of socioeconomic status with respect to inactivity, i.e., the subjects analysed all showed practically the same data regardless of whether they were of high or low SEP [82]. Children of low SEP generally have a higher BMI, more behavioural difficulties, report a lower quality of life, and experience more critical life events than children with a higher SEP [90], as shown in the males and females of the public school of the present study, specifically those belonging to low SEP, the highest levels of sedentary behaviour with respect to the rest of the deprivation levels. Values of sedentary behaviour are reflected in the higher BMI in males and in females, although the latter is not statistically significant. Schoolchildren of a lower SEP tend to be more obese ($12.6\%$ compared to $8\%$ of more advantaged children) [91]. Similarly to a representative sample of Swedish adolescents [92], females of high SEP (SEP 2) in private schools have the highest levels of light, moderate, and MVPA compared to the other categories of the index. Conversely, in females in public schools, light, moderate PA, vigorous PA, and MVPA values are lower in low SEP (SEP 4) than in higher and medium SEP (SEP 1 and 3) groups. Children of low socioeconomic backgrounds tend to be more prevalent in groups that combine multiple unhealthy lifestyles. Thus, children whose mother received a low educational level or children from a low-income household have been classified in groups that are physically inactive and engage in significant screen time [47,93]. The limitations of the present study have been associated with the difficulties of involving educational centres and the agents that form part of the schools; teachers, parents, and students. In addition, it was requested that the accelerometers be removed from the wrists when engaging in aquatic activities such as baths or showers, so there is very interesting information that is not collected in the practitioners of aquatic activities. With regard to future lines of research, on the one hand, the relative percentile values of the volume of PA could be analysed as a function of variables such as sex or age. On the other hand, it could be used to study the variables related to sleep quality. It would be highly intriguing to conduct a longitudinal study that researches the temporal changes in PA and sleep quality among child and adolescents in Basque Country, as well as the correlations between all these variables. ## 5. Conclusions The results of our study show a positive trend in both sedentary behaviour and sleep efficiency in Basque schoolchildren of intermediate SEP with respect to the extremes. In addition, it was observed that females performed less high-intensity PA than males in all SEP groups. Likewise, more sedentary behaviour was associated with less time spent in bed and poorer sleep efficiency. Similarly, more PA, regardless of its intensity, was associated with better body composition. These data point to notable inequalities in PA and daily sleep among Basque schoolchildren which, in turn, may be marginalised in our current school system due to the effects of the socioeconomic environment. It is therefore interesting to address future strategies in a cross-cutting manner so that schoolchildren extend their hours of movement both in and out of the classroom. ## References 1. Martín U., González-Rábago Y.. Bilbao’s Health in Figures. A Quantitative Diagnosis (2018.0) 2. Whitehead M., Dahlgren G.. Concepts and Principles for Tackling Social Inequities in Health: Levelling up Part 1 (2007.0) 3. Borrell C., Pons-Vigués M., Morrison J., Díez E.. Factors and processes influencing health inequalities in urban area. J. Epidemiol. Community Health (2013.0) 67 389-391. DOI: 10.1136/jech-2012-202014 4. Diez-Roux A.V., Mair C.. Neighborhoods and health. Ann. N. Y. Acad. Sci. (2010.0) 1186 125-145. DOI: 10.1111/j.1749-6632.2009.05333.x 5. A proposal of policies and interventions to reduce social inequalities in health in Spain. Gac. Sanit. (2012.0) 26 182-189. DOI: 10.1016/j.gaceta.2011.07.024 6. Mutrie N., Carney C., Blamey A., Crawford F., Aitchison T., Whitelaw A.. “Walk in to Work Out”: A randomized controlled trial of a self help intervention to promote active commuting. J. Epidemiol. Community Health (2002.0) 56 407-412. DOI: 10.1136/jech.56.6.407 7. Brown V., Moodie M., Mantilla-Herrera A.M., Veerman J.L., Carter R.. Active transport and obesity prevention—A transportation sector obesity impact. scoping review and assessment for Melbourne, Australia. Prev. Med. (2016.0) 96 49-66. DOI: 10.1016/j.ypmed.2016.12.020 8. Virto N., Etayo-Urtasun P., Sánchez-Isla J.R., Arietanizbeaskoa M.S., Mendizabal-Gallastegui N., Grandes G., Gutierrez B., Coca A., Río X.. Efectos de una intervención de 12 semanas de ejercicio en los niveles de hemoglobina glicada (HbA1c) en pacientes con cáncer (Effects of a 12-week exercise intervention on glycated hemoglobin (HbA1c) levels in cancer patients). Retos (2023.0) 48 153-160. DOI: 10.47197/retos.v48.96221 9. Wang Z., Emmerich A., Pillon N.J., Moore T., Hemerich D., Cornelis M.C., Mazzaferro E., Broos S., Ahluwalia T.S., Bartz T.M.. Genome-wide association analyses of physical activity and sedentary behavior provide insights into underlying mechanisms and roles in disease prevention. Nat. Genet. (2022.0) 54 1332-1344. DOI: 10.1038/s41588-022-01165-1 10. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–2013: A systematic analysis for the Global Burden of Disease Study 2013. Lancet (2015.0) 386 2287-2323. DOI: 10.1016/S0140-6736(15)00128-2 11. Diez-Roux A.V.. Residential environments and cardiovascular risk. J. Urban Health (2003.0) 80 569-589. DOI: 10.1093/jurban/jtg065 12. Humpel N., Owen N., Leslie E.. Environmental factors associated with adults’ participation in physical activity: A review. Am. J. Prev. Med. (2002.0) 22 188-199. DOI: 10.1016/S0749-3797(01)00426-3 13. Ball K., Bauman A., Leslie E., Owen N.. Perceived environmental aesthetics and convenience and company are associated with walking for exercise among Australian adults. Prev. Med. (2001.0) 33 434-440. DOI: 10.1006/pmed.2001.0912 14. Sallis J.F., Hovell M.F., Hofstetter C.R., Elder J.P., Hackley M., Caspersen C.J., Powell K.E.. Distance between homes and exercise facilities related to frequency of exercise among San Francisco residents. Diego residents. Public Health Rep. (1990.0) 105 179-185. PMID: 2108465 15. Rodriguez-Romo G., Garrido-Muñoz M., Lucía A., Mayorga J.I., Ruiz J.R.. Association between environmental characteristics and physical activity. Gac. Sanit. (2013.0) 27 487-493. DOI: 10.1016/j.gaceta.2013.01.006 16. Sallis J.F., Cervero R.B., Ascher W., Henderson K.A., Kraft M.K., Kerr J.. An ecological approach to creating active living communities. Anu. Rev. Public Health (2006.0) 27 297-322. DOI: 10.1146/annurev.publhealth.27.021405.102100 17. Sallis J.F., Glanz K.. The role of built environments in physical activity, eating, and obesity in childhood. Future Child. (2006.0) 16 89-108. DOI: 10.1353/foc.2006.0009 18. Bird E.L., Ige J.O., Pilkington P., Pinto A., Petrokofsky C., Burgess-Allen J.. Built and natural environment planning principles for promoting health: An umbrella review. BMC Public Health (2018.0) 18. DOI: 10.1186/s12889-018-5870-2 19. McGrath L.J., Hopkins W.G., Hinckson E.A.. Associations of objectively measured built environment attributes with moderate-vigorous youth physical activity: A systematic review and meta-analysis. Sport. Med. (2015.0) 45 841-865. DOI: 10.1007/s40279-015-0301-3 20. Nordbø E.C.A., Nordh H., Raanaas R.K., Aamodt G.. Promoting activity participation and well-being among children and adolescents: A systematic review of the determinants of the neighbourhood built environment. JBI Evid. Synth. (2020.0) 18 370-458. DOI: 10.11124/JBISRIR-D-19-00051 21. Ding D., Sallis J.F., Kerr J., Lee S., Rosenberg D.E.. Neighborhood environment and physical activity among young people a review. Am. J. Prev. Med. (2011.0) 41 442-455. DOI: 10.1016/j.amepre.2011.06.036 22. McCrorie P.R.W., Fenton C., Ellaway A.. Combining GPS. GIS and accelerometry to explore the relationship between physical activity and the environment in children and young people: A review. Int. J. Behav. Nutr. Phys. Act. (2014.0) 11 93. DOI: 10.1186/s12966-014-0093-0 23. Boone-Heinonen J., Evenson K.R., Song Y., Gordon-Larsen P.. Built and socioeconomic environments: Patterns and associations with physical activity in US adolescents. Int. J. Behav. Nutr. Phys. Act. (2010.0) 7 45. DOI: 10.1186/1479-5868-7-45 24. Kalak N., Lemola S., Brand S., Holsboer-Trachsler E., Grob A.. Sleep duration and subjective psychological well-being in adolescence: A longitudinal study in Switzerland and Norway. Neuropsychiatr. Dis. Treat. (2014.0) 10 1199-1207. DOI: 10.2147/NDT.S62533 25. Xu F., Adams S.K., Cohen S.A., Earp J.E., Greaney M.L.. Relationship between Physical Activity, Screen Time, and Sleep Quantity and Quality in US Adolescents Aged 16–19. Int. J. Environ. Res. Public Health (2019.0) 16. DOI: 10.3390/ijerph16091524 26. Sanz-Martín D., Ubago-Jiménez J.L., Ruiz-Tendero G., Zurita-Ortega F., Melguizo-Ibáñez E., Puertas-Molero P.. The Relationships between Physical Activity, Screen Time and Sleep Time According to the Adolescents’ Sex and the Day of the Week. Healthcare (2022.0) 10. DOI: 10.3390/healthcare10101955 27. Kline C.E.. The bidirectional relationship between exercise and sleep: Implications for exercise adherence and sleep improvement. Am. J. Lifestyle Med. (2014.0) 8 375-379. DOI: 10.1177/1559827614544437 28. Sanz-Martín D., Zurita-Ortega F., Ruiz-Tendero G., Ubago-Jiménez J.L.. Moderate–Vigorous Physical Activity, Screen Time and Sleep Time Profiles: A Cluster Analysis in Spanish Adolescents. Int. J. Environ. Res. Public Health (2023.0) 20. DOI: 10.3390/ijerph20032004 29. Kronholm E., Puusniekka R., Jokela J., Villberg J., Urrila A.S., Paunio T., Välimaa R., Tynjälä J.. Trends in self-reported sleep problems, tiredness, and related school performance among Finnish adolescents from 1984 to 2011. J. Sleep Res. (2015.0) 24 3-10. DOI: 10.1111/jsr.12258 30. Hansen J., Hanewinkel R., Galimov A.. Physical activity, screen time, and sleep: Do German children and adolescents meet the movement guidelines?. Eur. J. Pediatr. (2022.0) 181 1985-1995. DOI: 10.1007/s00431-022-04401-2 31. Global Matrix 4.0. Physical activity report card scores for children and adolescents: Results and analysis from 57 countries. J. Phys. Act. Health (2022.0) 19 700-728. DOI: 10.1123/jpah.2022-0456 32. Sallis J.F., Pate R.R.. Creating the future of physical activity surveillance in the United States: Better data for better health. J. Phys. Act. Health (2021.0) 18 1-5. DOI: 10.1123/jpah.2021-0182 33. Pérez L.G., Villodres G.C., Sánchez F.J.M., Cara E.M.S., Molina J.J.M.. Adherencia a la Dieta Mediterránea, Actividad Física y su relación con el Nivel Socioeconómico en escolares de Primaria de la capital de Granada. Retos (2021.0) 41 485-491. DOI: 10.47197/retos.v0i41.86166 34. Hallal P.C., Andersen L.B., Bull F.C., Guthold R., Haskell W., Ekelund U.. Global physical activity levels: Surveillance progress, pitfalls, and prospects. Lancet (2012.0) 380 247-257. DOI: 10.1016/S0140-6736(12)60646-1 35. Tapia-Serrano M.A., Sevil-Serrano J., Sánchez-Miguel P.A., López-Gil J.F., Tremblay M.S., García-Hermoso A.. Prevalence of meeting 24-hour movement guidelines from preschool to adolescence: A systematic review and meta-analysis including 387,437 participants and 23 countries. J. Sport Health Sci. (2022.0) 11 427-437. DOI: 10.1016/j.jshs.2022.01.005 36. Papadopoulos D., Etindele-Sosso F.A.. Socioeconomic status and sleep health: A narrative synthesis of three decades of empirical research. J. Clin. Sleep Med. (2022.0). DOI: 10.5664/jcsm.10336 37. Ke Y., Shi L., Peng L., Chen S., Hong J., Liu Y.. Associations between socioeconomic status and physical activity: A cross-sectional analysis of Chinese children and adolescents. Front. Psychol. (2022.0) 13 904506. DOI: 10.3389/fpsyg.2022.904506 38. Dennis E., Manza P., Volkow N.D.. Socioeconomic status, BMI, and brain development in children. Transl. Psychiatry (2022.0) 12 33. DOI: 10.1038/s41398-022-01779-3 39. Tandon P.S., Kroshus E., Olsen K., Garrett K., Qu P., McCleery J.. Socioeconomic Inequities in youth participation in physical activity and sports. Int. J. Environ. Health Res. (2021.0) 18. DOI: 10.3390/ijerph18136946 40. Stalsberg R., Pedersen A.V.. Effects of socioeconomic status on the physical activity in adolescents: A systematic review of the evidence. Scand. J. Med. Sci. Sport. (2010.0) 20 368-383. DOI: 10.1111/j.1600-0838.2009.01047.x 41. Gidlow C., Johnston L.H., Crone D., Ellis N., James D.. A systematic review of the association between socio-economic position and physical activity. Health Educ. J. (2006.0) 65 338-367. DOI: 10.1177/0017896906069378 42. González-Calvo G., García-Monge A., Gerdin G., Pringle R.. Making the familiar strange: A narrative about Spanish children’s experiences of physical (in)activity to reconsider the ability of physical education to produce healthy citizens. Sport. Educ. Soc. (2021.0) 28 227-238. DOI: 10.1080/13573322.2021.2014803 43. Bull F.C., Al-Ansari S.S., Biddle S., Borodulin K., Buman M.P., Cardon G., Carty C., Chaput J.P., Chastin S., Chou R.. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br. J. Sport. Med. (2020.0) 54 1451-1462. DOI: 10.1136/bjsports-2020-102955 44. Reilly J.J., Aubert S., Brazo-Sayavera J., Liu Y., Cagas J.Y., Tremblay M.S.. Surveillance to improve physical activity of children and adolescents. Bull. World Health Organ. (2022.0) 100 815-824. DOI: 10.2471/BLT.22.288569 45. Musić-Milanović S., Buoncristiano M., Križan H., Rathmes G., Williams J., Hyska J., Duleva V., Zamrazilová H., Hejgaard T., Jørgensen M.B.. Socioeconomic disparities in physical activity, sedentary behavior and sleep patterns among 6-to 9-year-old children from 24 countries in the WHO European region. Obes. Rev. (2021.0) 22 e13209. DOI: 10.1111/obr.13209 46. Gerber M., Lang C., Beckmann J., Degen J., du Randt R., Gall S., Walter C.. Associations Between Household Socioeconomic Status, Car Ownership, Physical Activity, and Cardiorespiratory Fitness in South African Primary Schoolchildren Living in Marginalized Communities. J. Phys. Act. Health (2021.0) 18 883-894. DOI: 10.1123/jpah.2020-0839 47. Yang-Huang J., van Grieken A., Wang L., Jansen W., Raat H.. Clustering of Sedentary Behaviours, Physical Activity, and Energy-Dense Food Intake in Six-Year-Old Children: Associations with Family Socioeconomic Status. Nutrients (2020.0) 12. DOI: 10.3390/nu12061722 48. Wong R.S., Tung K.T., Chan B.N., Ho F.K., Rao N., Chan K.L., Ip P.. Early-life activities mediate the association between family socioeconomic status in early childhood and physical fitness in early adolescence. Sci. Rep. (2022.0) 12 81. DOI: 10.1038/s41598-021-03883-8 49. 49. UNICEF How We Protect Children’s Rights with the United Nations Convention on the Rights of the ChildUNICEFNew York, NY, USA2022Available online: https://www.unicef.org.uk/what-we-do/un-convention-child-rights/(accessed on 5 February 2023). How We Protect Children’s Rights with the United Nations Convention on the Rights of the Child (2022.0) 50. Nyström C., Barnes J.D., Blanchette S., Faulkner G., Leduc G., Riazi N.A., Larouche R.. Relationships between area-level socioeconomic status and urbanization with active transportation, independent mobility, outdoor time, and physical activity among Canadian children. BMC Public Health (2019.0) 19. DOI: 10.1186/s12889-019-7420-y 51. Tremblay M.S., Shields M., Laviolette M., Craig C.L., Janssen I., Connor-Gorber S.. Physical fitness of Canadian children and youth: Results from the 2007–2009 Canadian Health Measures Survey. Health Rep. (2010.0) 21 7-20 52. Elmesmari R., Martin A., Reilly J.J., Paton J.Y.. Comparison of accelerometer-measured levels of physical activity and sedentary time among obese and non-obese children and adolescents: A systematic review. BMC Pediatr. (2018.0) 18. DOI: 10.1186/s12887-018-1031-0 53. Bozkurt M.. Metropolitan children’s physical fitness: The relationship between overweight and obesity prevalence, socioeconomic status, urban green space access, and physical activity. Urban Urban Green (2021.0) 64 127272. DOI: 10.1016/j.ufug.2021.127272 54. Galaviz U.Z., González R.V., Arvizuo J.G., Meneses E.F.E., Samaniego C.V., Domínguez I.D.J.T., Gutiérrez A.O.. Socioeconomic status and physical activity during elementary school student recess. Rev. Bras. Med. Esporte (2021.0) 27 80-83. DOI: 10.1590/1517-8692202127012019_0033 55. Peralta L.R., Mihrshahi S., Bellew B., Reece L.J., Hardy L.L.. Influence of School-Level Socioeconomic Status on Children’s Physical Activity, Fitness, and Fundamental Movement Skill Levels. J. Sch. Health (2019.0) 89 460-467. DOI: 10.1111/josh.12761 56. Duffey K., Barbosa A., Whiting S., Mendes R., Aguirre I.Y., Tcymbal A., Breda J.. Barriers and Facilitators of Physical Activity Participation in Adolescent Girls: A Systematic Review of Systematic Reviews. Front. Public Health (2021.0) 9 743935. DOI: 10.3389/fpubh.2021.743935 57. Kennewell E., Curtis R.G., Maher C., Luddy S., Virgara R.. The relationships between school children’s wellbeing, socio-economic disadvantage and after-school activities: A cross-sectional study. BMC Pediatr. (2022.0) 22. DOI: 10.1186/s12887-022-03322-1 58. Rosado J.R., Fernández Á.I., López J.M.. Evaluation of physical activity practice, dietary adherence and behaviour and its relationship with quality of life in Primary Education students. Retos (2020.0) 38 129-136. DOI: 10.47197/retos.v38i38.73921 59. Phillips S.M., Summerbell C., Hobbs M., Hesketh K.R., Saxena S., Muir C., Hillier-Brown F.C.. A systematic review of the validity, reliability and feasibility of measurement tools used to assess physical activity and sedentary behaviour in preschool children. Int. J. Behaviour. Phys. Nutr. Act. (2021.0) 18 141. DOI: 10.1186/s12966-021-01132-9 60. Cooper A.R., Goodman A., Page A.S., Sherar L.B., Esliger D.W., van Sluijs E.M., Anderson L.B., Anderssen S., Cardón G., Davey R.. Objectively measured physical activity and sedentary time in youth: The international children’s accelerometry database (ICAD). Int. J. Behaviour. Phys. Nutr. Act. (2015.0) 12 113. DOI: 10.1186/s12966-015-0274-5 61. Pate R.R., Almeida M.J., McIver K.L., Pfeiffer K.A., Dowda M.. Validation and calibration of an accelerometer in preschool children. Obesity (2006.0) 14 2000-2006. DOI: 10.1038/oby.2006.234 62. Chandler J.L., Brazendale K., Beets M.W., Mealing B.A.. Classification of physical activity intensities using a wrist accelerometer in children aged 8–12 years. Pediatr. Obes. (2016.0) 11 120-127. DOI: 10.1111/ijpo.12033 63. Ley 14/2007, de 3 de julio de 2007, de Investigación Biomédica 64. Editorial E.. Declaración de Helsinki de la Asociación Médica Mundial. Arbor (2008.0) 184 349-352 65. Reglamento (UE) 2016/679, de 27 de abril de 2016, Relativo a la Protección de las Personas Físicas en lo que Respecta al Tratamiento de Datos Personales y a la Libre Circulación de Estos Datos 66. Ley Orgánica 3/2018, de 5 de diciembre, de Protección de Datos Personales y Garantía de los Derechos Digitales 67. Real Decreto 1720/2007, de 21 de diciembre, por el que se Aprueba el Reglamento de Desarrollo de la Ley Orgánica 15/1999, de 13 de diciembre, de Protección de Datos de Carácter Personal 68. Domínguez-Berjón M.F., Borrell C., Cano-Serral G., Esnaola S., Nolasco A., Pasarín M.I., Ramis R., Saurina C., Escolar-Pujolar A.. Construction of a deprivation index from census data in large Spanish cities: (MEDEA Project). Gac. Sanit. (2008.0) 22 179-187. DOI: 10.1157/13123961 69. Guidolin M., Gradisar M.. Is shortened sleep duration a risk factor for overweight and obesity during adolescence? A review of the empirical literature. Sleep Med. (2012.0) 13 779-786. DOI: 10.1016/j.sleep.2012.03.016 70. Khan M.K.A., Chu Y.L., Kirk S.F.L., Veugelers P.J.. Are sleep duration and quality associated with diet quality, physical activity and body weight status? A population-based study of Canadian children. Can. J. Public Health (2015.0) 106 277-282. DOI: 10.17269/cjph.106.4892 71. Garfield V.. The association between body mass index (BMI) and sleep duration: Where are we after nearly two decades of epidemiological research?. Int. J. Environ. Res. Public Health (2019.0) 16. DOI: 10.3390/ijerph16224327 72. Chen H., Wang L.J., Xin F., Liang G., Chen Y.. Associations between sleep duration, sleep quality and weight status in Chinese children and adolescents. BMC Public Health (2022.0) 22. DOI: 10.1186/s12889-022-13534-w 73. López-Gil J.F., García-Hermoso A., Smith L., Firth J., Trott M., Eumann-Mesas A., Jiménez-López E., Gutiérrez-Espinoza H., Tárraga-López P.J., Victoria-Montesinos E.. Global Proportion of Disordered Eating in Children and Adolescents: A Systematic Review and Meta-analysis. JAMA Pediatr. (2023.0). DOI: 10.1001/jamapediatrics.2022.5848 74. Fatima Y., Doi S.A.R., Mamun A.A.. Sleep quality and obesity in young subjects: A meta-analysis. Obes. Rev. (2016.0) 17 1154-1166. DOI: 10.1111/obr.12444 75. Mota J., Vale S.. Associations between sleep quality with cardiorespiratory fitness and BMI among adolescent girls. Am. J. Hum. Biol. (2010.0) 22 473-485. DOI: 10.1002/ajhb.21019 76. Wang J., Adab P., Liu W.J., Chen Y.J., Li B., Lin R., Liu W., Cheng K.K., Pallan M.. Prevalence of adiposity and its association with sleep duration, quality, and timing among 9-12-year-old children in Guangzhou, China. J. Epidemiol. (2017.0) 27 531-537. DOI: 10.1016/j.je.2016.11.003 77. Muthuri S.K., Wachira L.J.M., Leblanc A.G., Francis C.E., Sampson M., Onywera V.O., Tremblay M.S.. Temporal trends and correlates of physical activity, sedentary behaviour, and physical fitness among school-aged children in Sub-Saharan Africa: A systematic review. Int. J. Environ. Res. Public Health (2014.0) 11 3327-3359. DOI: 10.3390/ijerph110303327 78. Lavin J., Bruzzone F., Mamondi V., González E., Berra S.. Physical activity and sedentary behaviors of schoolchildren from Córdoba in their free time and in Physical Education class. Rev. Argent. Salud. (2015.0) 6 15-21 79. Owen K.B., Nau T., Reece L.J., Bellew W., Rose C., Bauman A., Halim N.K., Smith B.J.. Fair play? Equity of participation in organised sport and physical activity among children and adolescents in high-income countries: A systematic review and meta-analysis. Int. J. Behav. Nutr. Phys. Act. (2022.0) 19 27. DOI: 10.1186/s12966-022-01263-7 80. Heradstveit O., Haugland S., Hysing M., Stormmark K., Sivertsen B., Boe T.. Physical inactivity, non-participation in sports and socioeconomic status: A large population-based study among Norwegian adolescents. BMC Public Health (2020.0) 20. DOI: 10.1186/s12889-020-09141-2 81. Stalsberg R., Pedersen A.. Are Differences in Physical Activity across Socioeconomic Groups Associated with Choice of Physical Activity Variables to Report?. Int. J. Environ. Res. Public Health (2018.0) 15. DOI: 10.3390/ijerph15050922 82. Guthold R., Stevens G., Riley L., Bull F.. Global trends in insufficient physical activity among adolescents: A pooled analysis of 298 population-based surveys with 1–6 million participants. Lancet (2019.0) 4 23-35. DOI: 10.1016/S2352-4642(19)30323-2 83. Fernández-Prieto I., Giné-Garriga M., Canet O.. Barriers and motivations perceived by adolescents in relation to physical activity. Qualitative study through focus groups. Rev. Esp. Salud. Publica (2019.0) 93 1-12 84. Falese L., Federico B., Kunst A.E., Perelman J., Richter M., Rimpelä A., Lorant V.. The association between socioeconomic status and vigorous physical activity among adolescents: A cross-sectional study in six European cities. BMC Public Health (2021.0) 21. DOI: 10.1186/s12889-021-10791-z 85. Love R., Adams J., Atkin A., van Sluijs E.. Socioeconomic and ethnic differences in vigorous intensity physical activity in children: A cross-sectional analysis of the UK Millennium Cohort Study. BMJ Open (2019.0) 9 e027627. DOI: 10.1136/bmjopen-2018-027627 86. Tarp J., Child A., White T., Westgate K., Bugge A., Grøntved A., Wedderkopp N., Andersen L.B., Cardon G., Davey R.. Physical activity intensity, shift duration and markers of cardiometabolic risk in children and adolescents. Int. J. Obes. (2018.0) 42 1639-1650. DOI: 10.1038/s41366-018-0152-8 87. Farinola M.G., Tuñón I., Laíño F., Marchesich M., Rodríguez M.P.. Socio-educational and economic profile of Argentine elite adolescent athletes. Retos (2018.0) 34 172-176. DOI: 10.47197/retos.v0i34.60072 88. McCormack G.R., Giles-Corti B., Timperio A., Wood G., Villanueva K.. A cross-sectional study of individual, social, and constructed environmental correlates of pedometer-based physical activity among primary schools children. Int. J. Behav. Nutr. Phys. Act. (2011.0) 8 30. DOI: 10.1186/1479-5868-8-30 89. Sluijs E., Ekelund U., Crochemore-Silva I., Guthold R., Ha A., Lubans D., Oyeyemi A., Ding D., Katzmarzyk P.. Physical activity behaviours in adolescence: Current evidence and opportunities for intervention. Lancet (2020.0) 398 429-442. DOI: 10.1016/S0140-6736(21)01259-9 90. Poulain T., Vogel M., Sobek C., Hilbert A., Korner A., Kiess W.. Associations Between Socio-Economic Status and Child Health: Findings of a Large German Cohort Study. Int. J. Environ. Res. Public Health (2019.0) 16. DOI: 10.3390/ijerph16050677 91. 91. Gasol Foundation The Gasol Foundation and Fundación Probitas Resume Their Partnership to Promote the Second Edition of the PASOS StudyGasol FundationBarcelona, Spain2022Available online: https://cutt.ly/DI0o4qx(accessed on 10 January 2023). The Gasol Foundation and Fundación Probitas Resume Their Partnership to Promote the Second Edition of the PASOS Study (2022.0) 92. Nyberg G., Kjellenberg K., Fröberg A., Lindroos A.K.. A national survey showed low levels of physical activity in a representative sample of Swedish adolescents. Acta Paediatr. (2020.0) 109 2342-2353. DOI: 10.1111/apa.15251 93. Leech R.M., McNaughton S.A., Timperio A.. Clustering of children’s obesity-related behaviours: Associations with sociodemographic indicators. Eur. J. Clin. Nutr. (2014.0) 68 623-628. DOI: 10.1038/ejcn.2013.295

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.